Part I - Systemic Shifts in the Global Economic Order

2 Trade Law in a Data-Driven Economy The Need for Modesty and Resilience

Gregory Shaffer *

Data, it is said, is the new oil. Treated as a raw material, once processed, once refined, it fuels the new economy. But unlike oil, data is not only nondepletable, it is also constantly generated and exponentially growing. Unlike oil, data exhaust from transactions is not waste but recycled for further use. And unlike oil, data is nonrivalrous as it can be exploited by multiple users. Despite the differences, the metaphor is powerful not only because data fuels the new economy, but also because data is extracted (in this case not only from land but from humans), and because data is protected and traded through law as a new form of property. The value of data vastly surpasses that of oil as measured by the capitalization of the world’s largest firms.¹ The only question for companies is how to gather, store, analyze, and deploy data ever more efficiently since data can significantly reduce transaction and production costs. This chapter examines the social challenges posed by such an economy, their implications for trade law, the current trade negotiating context, and a way forward that can both enhance trade and regulatory efficacy. Section I sets the stage regarding law as a “channeling” tool in the digital economy. Section II examines eight critical challenges. Section III presents the negotiating context in which major powers advance different governance models. Section IV provides a governance framework for moving forward in light of the challenges, a framework that is modest and that foregrounds the importance of building resilience and engaging in problem solving, learning, and adaptation. Section V concludes.

I The Data-Driven Economy and Law

The data-driven economy refers to the collection, aggregation, organization, analysis, exchange, and exploitation of digital information, whether for use in production (such as in “smart manufacturing” and “smart agriculture”), the sale of goods and services (such as through electronic commerce), the provision of services (such as through online platforms like Uber), or trade in data itself (whether for advertising, solicitation, or assessment, such as for credit ratings).² The data-driven economy is fueled by the data generated from connected devices, which is then used to innovate, produce, operate, and sell responsive machines, goods, and services.³ McKinsey estimates that the value of global data flows surpassed that of trade in goods as early as 2014.⁴

Technology has been and is being developed through the exponential rise in computing power, storage, and bandwidth to exploit data. 5G wireless technology expands capacity, enhances the speed of information flows, reduces latency for near-real-time communication, and transforms scalability for new services. Data-trained artificial intelligence (AI) industrializes learning, which increases productivity, reduces costs, and improves logistical services, facilitating trade.⁵ Microchips enable powerful computers at our fingertips, generating new data to be processed. “Smart” manufacturing self-automates, trumpeted in Germany as “Industry 4.0” and in the United States as the “Industrial Internet.” Linking big data, cloud computing, wireless sensor networks, and automated analytic tools with industrial equipment, it makes manufacturing more efficient, more precise, and more responsive.⁶ Daily life – from driverless cars to heart monitors and security locks – revolutionize through the so-called Internet of Things.⁷

Many if not most commentators on trade and technology are technological optimists since, in basic economic theory, “technological progress by definition shifts out the production possibilities frontier” and thus enhances aggregate social welfare.⁸ Basic trade law and economics casebooks deploy parables that compare trade with technology with a moral that countries should embrace the social welfare benefits of trade.⁹ Even Dani Rodrik, a leading critic of the trade regime for having liberalized too far, has argued that technology is more benign than trade in its distributional effects.¹⁰

If one is a technology optimist, then our task is less daunting: law should incentivize technology’s development and use. Ronald Gilson, writing from Stanford in the nerve center of Silicon Valley, famously called lawyers “transaction cost engineers.”¹¹ It is lawyers who grease the wheels and driverless cars of innovation – the creative disruptor of not just commerce but our life worlds. It is law and lawyers that construct the intangibles of the data-driven economy, such that its potential as energy is released.¹² Because AI systems require huge quantities of updated data to “train” themselves and continuously learn, improve, and refine their output, the data-driven economy relies on the free flow of data across borders generated from digitized societies.¹³ Predictions made through AI improve with more data, driving its demand. For economic globalization, the “free flow of data” becomes the “fifth freedom” alongside the free movement of goods, services, capital, and labor – the “four freedoms” of the European Union’s internal market.¹⁴

The great contract scholars and rule-of-law theorists stressed law’s channeling function.¹⁵ For the technological optimist, lawyers’ role is to free up data flows so as to release pent-up energy for a leap in efficiency, facilitating the making of responsive, just-in-time products adapted to individual and group desires and needs. The challenge of trade law scholars then is to combat constraints on data flows such as data localization requirements that are proliferating,¹⁶ and new digital taxes,¹⁷ constituting a new protectionism impeding progress in the digitalized world. The challenge is to press for interoperative standards to ensure frictionless flow across borders and combat fragmentation. The dream is a world where small- and medium-sized enterprises (SMEs) – the Jeffersonian democrats of the marketplace – can compete on a fair footing with the multinational behemoths.¹⁸ The goal is a terrain where developing country entrepreneurs can better participate and compete because deficiencies of physical infrastructure matter less in a world of handheld computers and digitalized communications and services.¹⁹ The vision is a world of affordable products tailored for individual wants produced in an environmentally sustainable way.²⁰ With advances such as 3D printing, we conceivably could live in a more localized society that would, in the words of Richard Baldwin in his book The Globotics Upheaval, “make for a better society.”²¹ Trade lawyers’ role, from this vantage, is to release the potential of microchips, circuits, and smart machines through the free flow of data. If only government representatives could see policy from the individual consumer’s perspective and understand the utilitarian benefits of global markets, the neoclassical trade theorist posits, the world would be more prosperous, more free, and more peaceful.

But if one is not a technology optimist, if one is a pragmatist who believes that there are tradeoffs, if one finds that technology may be unstoppable but there remain choices for governing it, if one is concerned about not just pathologies but also pathogens that law can channel, then what channel should governments choose? In a world of uncertainty, of speculation, amidst the fog of transnational distrust, insecurity, and rivalry, governments face a daunting task.

II The Challenges Posed

Let us consider eight risks that the technological tsunami of AI could unleash, which are both distinct and interrelated. They are the rise of social inequality and “winner-takes-all” industries, social control through surveillance, risks to democracy, national security threats, economic vulnerability and systemic risk, premature deindustrialization implicating development, geopolitical conflict, and threats to personal privacy and dignity. Although technological change offers great societal benefits, it also raises new regulatory challenges for which responses vary depending on societal contexts and preferences. There is thus reason for pause before concluding ambitious trade agreements that free data flows, at least without significant safeguards. This section examines each of these challenges before the next sections address future trade governance options.

First, the data-driven economy could spur growing inequality in multiple ways, raising social conflict. On the one hand, because of network effects, increasing returns of scale and scope, and the dynamic of first-mover advantage, the data-driven economy increasingly gives rise to winner-takes-all – or winner-takes-most – companies, such as Amazon for e-commerce, Google for search engines, and Facebook for social networking.²² Companies proficiently using AI can serve additional customers globally at little marginal cost at the same time as they enhance quality, enabling the owner of this form of capital to capture unprecedented rents.²³ Unlike traditional industries, scale can be increased without the costs of “mass” because data is weightless; its storage is in the cloud. It thus entails near-zero marginal production costs. From this vantage, the trumpeting of e-commerce in terms of how it will benefit SMEs could be a utopian fantasy.

A data-driven economy not only enables economic behemoths to monopolize but also enables them to engage in price discrimination so that they price at what each individual consumer is willing to pay. One of the staple arguments for the benefit of markets in neoclassical economics – that of “consumer surplus” – is thus extracted from individuals since companies have the ability to predict what exactly each consumer is willing to pay and charge that amount.²⁴ In addition, AI permits companies to engage in cartel-like behavior through reactive, tit-for-tat responses to coordinate prices, once more to extract rents.²⁵ As winner-takes-most companies reap monopoly, oligopoly, and collusive rents, inequality proliferates (including between high- and low-skilled workers). The trade regime is already under considerable stress; simply removing trade barriers to data flows could contribute to social conflict within countries unless growing inequality is addressed.

Second, companies and governments gather data through surveillance that can exploit and shape us. The algorithms they use to process our data enable them to know our future and predict what we want and when we want it better than ourselves. Our wired world creates opportunities for enhanced state control through harnessing social pressure, epitomized by the development of “social credit” systems in China.²⁶ In parallel, it enables “surveillance capitalists” to steer us toward products to maximize their profits.²⁷ Not only can our data be automated, but we can too. When social media become constitutive of social participation, we become increasingly numb to companies and governments knowing everything about us, while we know nothing about how and what they know.²⁸ We as consumers are consumed. We participate in our commoditization to fuel trading in data to make the commodities that we buy. It is law that helps constitute that relationship, including through protecting company algorithms through property law.²⁹ Law could, for example, ban particular algorithmic practices and otherwise require disclosure and monitoring of algorithms so that they can be contested, whether for different forms of bias or for their social consequences.

Third, the data revolution poses massive problems for democracies. To start, the dynamic of increased economic inequality facilitates conditions for decreased social solidarity and increased social conflict, which can erode democracies. More specifically, the data revolution enables others to manipulate our views, including through the proliferation of “fake news” that harnesses predictive power regarding our psychology and behavior.³⁰ Tech developed and harnessed by groups such as Cambridge Analytica relentlessly targets individual vulnerabilities and spurs “thinking fast” tribal responses, thus manipulating behavior to win elections and embed leaders in power.³¹ Foreign authoritarian powers can harness these mechanisms, creating “vast numbers of fake persons orchestrated by shadowy intelligence warfare units building momentum for online paranoia and conspiracy theories.”³² In addition, entrepreneurs can profit by targeting search results from prior preferences, which divides societies into information bubbles, leading to increased social fragmentation and political polarization. While foreign governments target conspiracy theories at vulnerable groups to create social chaos, entrepreneurs do so to profit from “clickbait.”³³ As Larry Diamond documents, antidemocratic politics are spreading globally.³⁴ Democracies risk becoming a shell, unless governments, companies, and societies rise to the regulatory challenge.

Fourth, data and AI pose national security challenges, in part because the refined data and technology have dual uses, giving rise to a shift in trade analysis toward geopolitics and “geoeconomics.”³⁵ This shift places traditional trade liberals, with their analysis of trade’s mutual benefits, on the defensive. At the core of the USA–China trade war is technology, which will determine the global leaders of tomorrow and whether those leaders are Chinese or American. As part of this competition, China competes with the United States, Europe, and Japan in creating standards for the data-driven economy, such as for 5G infrastructure and the future of manufacturing. The US contestation of China’s 2025 innovation initiative, in part, is because China threatens to take the lead in “smart manufacturing” at the cutting edge of technology. Yet that technology also can be used for military purposes. Sales of Huawei 5G infrastructure, for example, become security concerns not only because they facilitate espionage but also because a country’s economy can be held hostage under the threat of a shutdown of wireless services. Technology can be “weaponized” by withholding key components in a trade war or in an actual conflict.³⁶

Fifth, the technology poses significant systemic concerns regarding the risks of system vulnerability, integrity, and availability.³⁷ If the 5G network were to shut down without a backup, social chaos could spread, giving rise to a Margaret Atwood MaddAddam dystopia.³⁸ Economics, ecology, engineering, and psychology – from their different vantages – all stress the importance of resilience to guard against system collapse,³⁹ which the risks of the COVID-19 virus exemplify. The so-called global financial crisis was not in fact global because China and Chinese banks were less ensnared in the market disintegration triggered by the US housing and mortgage-backed securities bubble. Countries could sell their products to China, enabling the global economy to staunch contagion and recover more quickly. Imagine the counterfactual if China’s economy had been “just like us” (i.e. the United States) in 2008, with free capital flows and globally integrated banks, and had crashed as well. Because it differed, there was greater resilience for the global economy, benefiting everyone. This experience holds lessons for the risks posed by a global economy dependent on single technological systems, regardless of whether geoeconomic conflict can be managed.

Sixth, the technological revolution can lead to “premature deindustrialization” of developing countries, possibly trapping them at low-income levels in services sectors, widening the global economic divide by a “digital divide.”⁴⁰ Development economists worry about the consequences for development since manufacturing helped make many developing countries, particularly in Asia, richer.⁴¹ With deindustrialization, smart manufacturing enterprises operate more like software companies, requiring employees to design, program, operate, and debug “smart” machines. That know-how will more likely reside in a few leading countries, with the United States, Europe, China, and a few others vying for leadership. In the winner-takes-most economy, large countries that require data localization, such as China, can grant privileged access to their nationals’ data to national companies. That is why populous countries such as India, Indonesia, and Brazil envy Chinese Internet companies’ fortunes. The calculus for smaller developing countries is less favorable. They most likely benefit from free data flows for foreign companies serving their constituents, but they will also face foreign monopolists’ economic clout.

Seventh, because of rising inequality within countries (which the data-driven economy facilitates), combined with declining inequality between the West and a few emerging powers (notably China with its massive investments in AI and data-linked technology), social conflict both within and between countries could rise. Violence looms, threatening national civic and global peace. For some, the link between the threat of violence and this combination of increasing inequality within countries and decreasing inequality between them may seem paradoxical. Within national contexts, rising domestic inequality increases domestic social conflict. US President Trump’s references to the prospect of “civil war” were he to be impeached are symptomatic,⁴² as are the mass protests of the “yellow vest” movement in France.⁴³ At the international level, declining inequality between the United States and China threatens US hegemony and, possibly in turn, the stability of the international system to the extent that it depends on one country being hegemonic (per “hegemonic stability theory”).⁴⁴ In parallel, populist leaders harness US and European workers’ lost sense of status from the shift of jobs to China and the East, harnessing nationalist fervor. Scholars now warn of the “Thucydides Trap” in which a rising power and an incumbent heedlessly and inescapably march toward war.⁴⁵

Eighth, and finally, there are risks to personal privacy and dignity. We have so far stressed societal risks as opposed to individual ones, as the latter have been most frequently addressed in legal scholarship.⁴⁶ Yet many of these societal risks build on individual ones. Even if societal risks are addressed, the risks to individuals regarding their privacy, dignity, and safety can be ruinous, whether the individual are coerced by authoritarian governments or privately, such as through social media.⁴⁷

III The Current Negotiating Context

Trade negotiations often take a mercantilist orientation where trade negotiators aim to protect domestic industries while opening foreign markets. Through the mechanism of reciprocity, these negotiations, complemented by litigation, have led to greater trade liberalization over time. For those focused on reciprocally opening markets, their starting point is no different for data than it is for goods and services – how to free up flows, in this case data flows that are intrinsic to the new data-driven economy. In this way, trade law can reduce transaction costs for business and the costs of segmented markets.⁴⁸

Trade scholars have focused on the fit of current trade rules with developments in the new economy, finding the fit wanting.⁴⁹ The same conclusion applies to international law more generally,⁵⁰ rendering the challenges for trade law even greater. Given that World Trade Organization (WTO) rules were negotiated over a quarter of a century ago, before the Internet existed, scholars naturally conclude that trade rules must be updated. WTO rules still address primarily goods, a legacy of the 1948 General Agreement on Tariffs and Trade (GATT), an era where industrial manufacturing represented the commanding heights of the economy. In 1995, with the creation of the WTO and its inclusion of a General Agreement on Trade in Services (GATS), the trade regime partially and indirectly addressed services that are linked to the digital economy, as well as technical regulations affecting trade in goods.⁵¹ But that too was a quarter of a century ago and technology has changed radically. Today, services constitute the largest and fastest-growing part of the global economy in terms of output, value added, and employment.⁵² The GATS only rudimentarily addresses digital issues where the line between a “good” and a “service” blurs and could eventually disappear.⁵³ Not only are an increasing number of goods now inextricably linked with “services” (the Internet of Things), but know-how and data have become the most valuable components of trade across borders.⁵⁴

While WTO negotiations have failed to fill key regulatory gaps for digital trade, countries have negotiated bilateral and plurilateral trade agreements to instill their priorities and values into standards for the digital economy. Trade negotiations reflect competition between systems, since countries’ positions reflect their internal policies. One can broadly speak of three distinct approaches for digital governance advanced by the WTO’s three most powerful members – the United States, the European Union (EU), and China.⁵⁵

The United States has trumpeted a world of free “data flows” that would benefit its companies.⁵⁶ The United States-Mexico-Canada Agreement, which entered into force on 1 July 2020, illustrates the US approach.⁵⁷ Chapter 19 of the agreement is on “Digital Trade,” and it represents the first time that a trade agreement has a chapter with such a title. Although this change is, in part, semantic since the chapter borrows significantly from chapter 14 of the earlier Trans-Pacific Partnership (TPP) agreement on “Electronic Commerce,” the title signifies a broader concern than trade in goods, and the chapter indeed further tightens rules in favor of US technology companies. It includes provisions mandating free movement of data, a permanent moratorium on customs duties, and bans on data localization requirements, forced disclosure of source codes, and other forced technology transfers. It also includes a new provision providing that Internet platforms should not be held civilly liable for their users’ actions, which is modeled on section 230 of the US Communications Decency Act.⁵⁸

Although the EU has advanced liberalization objectives, it imposes significant restraints on the free flow of data on privacy grounds. The EU’s position is reflected in its General Data Protection Regulation (GDPR), together with EU judicial oversight of its negotiation of “adequacy decisions” with third countries, such as through “safe harbors” and “privacy shields,” in order for data on European citizens to leave the continent. On 16 July 2020, the European Court of Justice invalidated the US-EU Privacy Shield because it provides inadequate protection to EU citizens’ privacy from surveillance, just as the court had in October 2015 as regards the previous US-EU “Safe Harbor Privacy Principles.”⁵⁹ Because the EU lacks leading digital firms, it is politically easier for it to champion such regulation.⁶⁰

China, in contrast, applies “data localization” requirements on sovereignty grounds, rather than the protection of citizen rights. In this way, the Chinese state and Chinese companies control data over China’s 1.4 billion citizens, facilitating social control while creating a competitive advantage for Chinese enterprises.⁶¹ A result is the rise of Chinese information technology titans such as Alibaba and Tencent.

Other countries choose among these three models, although those “choices” occur within negotiating contexts that can involve highly asymmetric power. In this way, these different models are adopted around the globe. Japan has adopted the US approach, as reflected in the 2019 US-Japan Trade Agreement.⁶² Australia and Canada have hybrid approaches that include stronger data privacy protection as under the EU model, while India, Indonesia, and Brazil are enticed by China’s requirements of data localization to create national champions.⁶³

A question arises regarding how these models will interface. It is conceivable that the United States and EU could negotiate a further compromise. Although the EU will periodically challenge US tech giants, it unlikely will develop its own. Nonetheless, the EU generally favors a single market for data flows, subject to adequate privacy and consumer protection. The United States and China, however, are less likely to negotiate a compromise (unless it includes significant carve-outs on national security and other public policy grounds) given the advantages for China of requiring data localization so that foreign companies do not gain access to the Chinese data trove. India will likely follow this route.⁶⁴ Economic behemoths from just a few countries could dominate the globe. From an economics perspective, the global market will not be based on perfect competition reflected in neoclassical models, but rather “strategic trade” in which a few countries compete to support national champions that reap oligopolistic and monopolistic profits, potentially having positive spillover effects for their national economies.

IV Governance Framework for a Way Forward: A Call for Modesty

Although the challenges are severe, if they are to be met, law must play a critical role nationally, internationally, and transnationally. This section provides a framework for addressing the challenges posed. It places trade law in a broader regulatory context that involves competing ways to frame the “problem” to be addressed in a world characterized by uncertainty and rapid technological change. It then applies it to particular issues.

Addressing the challenges requires regulation. Two key issues are: (i) at what level regulation should occur; and (ii) what form and content such regulation should take. In practice, regulation can occur at multiple levels and take different forms, public and private, hard and soft. Moreover, regulation in any one country will have impacts on constituencies outside that country, so that countries have incentives to address these externalities. In parallel, common problems may require regulatory coordination among countries to address it. The key questions thus can be reframed as: (i) What regulation, if any, should occur at the international level and how should it interface with national regulation? (ii) What forms should regulation take and how should these forms interface?

1. Three Governing Principles.

To determine a framework for governance of the respective challenges, we start with three principles. First, traditional trade agreements are not optimal for regulatory agreements and thus trade agreements need to be viewed as part of a broader ecology of governance of the new data-driven economy, which creates links between different rule-making and monitoring bodies at different levels of social organization. Second, for most of the issues raised in Section II, there should be no single system of hierarchical rules. Rather, in a world of radical uncertainty and different preferences regarding the regulation of these issues, countries will benefit from experimentation with different regulatory approaches. A diversity of regulatory approaches provides greater resilience against the systemic risks posed when single systems fail. Third, given the transnational impacts of the risks, as well as of national regulation addressing (or failing to address) them, there is a need for systems of regulatory coordination over options and experiences that will facilitate trade while enhancing regulatory efficacy, learning, and adaptation. Once one turns to issues of coordination and the interface of different national regulatory systems, one is in a world of transnational legal ordering that is not just top-down but also bottom-up, horizontal, and transversal.

Elsewhere I have developed a theory of transnational legal orders with the sociologist Terence Halliday.⁶⁵ That approach focuses on how problems are framed, norms develop transnationally in response to such framings, and norms settle and unsettle as part of recursive processes of interaction between different levels of social organization, from the international to the national and local. That framework has been predominantly positivist in its approach, aimed at generating empirical research for how legal norms develop, diffuse, and change transnationally.

Such an approach, however, also has normative payoffs when combined with what has been called new governance theory regarding systems of adaptive regulation in light of uncertainty involving changing problems and regulatory contexts. New governance theory, when applied transnationally, emphasizes the need for the development of new transnational institutional structures for regulation comprising a common forum for deliberation, principles to guide discussions, an open menu of options for addressing regulatory choices, and peer review and information sharing to enhance trust and learning.⁶⁶ Through such structured processes of regulatory dialogue, both hard and soft international law norms can develop.⁶⁷ The question becomes: How can trade law help to facilitate and channel these processes?

Under a new governance approach, countries jointly create regulatory goals and measures to gauge achievement and permit variation in how regulatory agencies pursue the attainment of these goals. These agencies then report to each other and participate in peer-review processes regarding regulatory outcomes, aimed at continual improvement and potential reassessment of goals in light of experience.⁶⁸ This approach, in the pragmatist tradition, entails ongoing mutual scrutiny of outcomes and their effectiveness based on continuous information exchange by regulators committed to regulatory improvement and attentive to risk, including potentially catastrophic risks. Under this approach, regulators exchange information, conduct joint trials and risk assessments, monitor results, and adapt regulatory practices.⁶⁹ Transparency is central to this model through processes of information sharing, peer review, questioning, and response. Through regulatory learning, norms and practices can recursively change.

At the global level, Charles Sabel and Bernard Hoekman note the possibility of open plurilateral trade agreements that could create frameworks for developing such a regulatory approach. A core group of countries initially would join the agreement, but others could join it subsequently.⁷⁰ In June 2020, Chile, New Zealand, and Singapore signed a Digital Economy Partnership Agreement that aims to develop mechanisms that build trust in data flows, which is open for other parties to join.⁷¹ More broadly, the United States and EU discussed the development of new transatlantic regulatory mechanisms in their negotiations to create a Transatlantic Trade and Investment Partnership.⁷² The EU proposed the creation of a new transatlantic body, called a Regulatory Cooperation Body, to support specific regulatory cooperation initiatives and oversee them. Hoekman noted how other institutions, including private ones, could complement it in particular regulatory domains.⁷³ Through ongoing interactions, national regulators eventually could recognize each other’s regulations as functionally equivalent, facilitating trade. These programs could lead to the institutionalization of broader sectoral frameworks, giving rise to cooperative regulatory systems that reduce barriers to trade while enhancing regulatory responsiveness in an increased number of domains. Food safety is one area where such a system has been applied transnationally.⁷⁴ The governance challenges posed by the digital economy beckon for new institutional initiatives in this vein.

New governance theory is particularly useful in a world of radical regulatory uncertainty. Given the risks, uncertainties, and differences in values, interests, and priorities, international trade law must not foreclose experimentation and variance. Yet, given these very same risks, uncertainties, and differences, international trade law and institutions are needed to foster cooperation, deliberation, and exchange of ideas. International trade law must foster transnational engagement, while not foreclosing regulatory policy space to engage with the challenges posed. Seeking and adjusting the “right” balance between coordination, harmonization, and experimentation will be an ongoing challenge.

2. Electronic Commerce.

Regarding electronic commerce, a WTO trade agreement is most achievable if it adopts a decentralized model that accommodates regulatory flexibility in which countries of varying levels of development have different implementation periods conditioned on regulatory capacity building and technical assistance. The WTO’s 2017 WTO Trade Facilitation Agreement offers a model of how this can be done.⁷⁵ The Trade Facilitation Agreement provides for flexibility in relation to a country’s level of development, and it facilitates provision of technical assistance and resources for developing countries to adapt their regulatory systems. A new digital trade agreement could have a similar structure, in this case organized to accommodate not only countries at different levels of development but also to support the interface and interoperability of different regulatory systems that reflect varying national practices and preferences.⁷⁶ It could establish digital norms to ensure the validity of contracts, recognition of electronic authorizations and signatures, protection against fraudulent practices, and the banning of unsolicited commercial messages. In this way, parties would commit both to foster consumer trust by protecting information and preventing fraud, and to cooperate to tackle transnational problems, such as spam generated from abroad.⁷⁷ Developing country adherence to them, however, would be subject to the receipt of technical assistance, as under the Trade Facilitation Agreement.

These norms could be negotiated and developed in conjunction with other venues, such as before the United Nations Commission on International Trade Law (UNCITRAL), the United Nations Conference on Trade and Development (UNCTAD), the Organisation for Economic Co-operation and Development (OECD), and the G20, each of which has ongoing programs to develop, share, assess, and provide capacity building for the adoption of e-commerce regulations.⁷⁸ Though developed elsewhere, the norms could be incorporated by reference into the trade agreement and be updated over time. They could constitute minimum standards, while permitting countries to deviate from them on legitimate regulatory grounds. There is precedent for this approach in WTO and other trade agreements. Within the WTO, the Agreement on Sanitary and Phytosanitary Standards references standards developed by Codex Alimentarius and other standard-setting bodies, and the Agreement on Technical Barriers to Trade references international standards more generally, including those developed in the International Organization for Standardization (ISO). More directly on point, in the EU-Canada agreement known as CETA (Comprehensive Economic and Trade Agreement), the parties agree that they “shall take into due consideration international standards of data protection of relevant international organizations of which both Parties are members.”⁷⁹ In each case, they permit parties to apply more stringent standards for legitimate regulatory reasons.

Such an agreement could also include provisions that are standard in trade agreements. It could require nondiscrimination between domestic and foreign digital products, potentially subject to negotiated product and sectoral carve-outs and general exceptions on regulatory policy grounds, including national security. It could incorporate basic due process commitments, including the right to be heard and to receive reasoned justifications before administrative and judicial processes. It could address (and either ban or otherwise limit) customs duties on electronic transmissions. It could likewise cover the use of digital taxes, possibly, once more, by reference to standards developed elsewhere, whether in the OECD or otherwise.⁸⁰ It also could clarify and enhance parties’ market access commitments to services that affect digital trade, which is currently being negotiated in the form of a Trade in Services Agreement on a plurilateral basis among a subset of twenty-three WTO members, including the United States and EU.⁸¹

Such an agreement could include an ongoing new governance component as well. It could require regulatory transparency and create a framework for regulators, standard setters, and commercial enterprises to engage with and learn from each other to address the uncertainties that new technologies pose and share information through peer-to-peer processes.⁸² Although WTO committee and working group processes offer one means,⁸³ these groups also can work in coordination with other international organizations and standard-setting bodies where the primary regulatory peer review could be done. These latter bodies could then report to the WTO committee. In parallel, bilateral and plurilateral trade agreements can serve as learning laboratories for the development of norms.

3. Cybersecurity and Resilience.

To turn to the other challenges raised in Section II, they will be more difficult. For example, concerns over resilience represent a critical reason why a very “ambitious” trade agreement would be problematic at this stage. Governments must be free to regulate and require different standards, product controls, and even product bans, on security grounds to ensure resilience. States and companies will need to develop backup, modular, and exit systems involving redundant and diverse infrastructure that is adaptive to 5G communications and other breakdowns.⁸⁴ It is a critical question for engineering and for regulatory policy. There are always tradeoffs in product performance and costs, on the one hand, and security, on the other. But here it is not a question of simple market failure and “second-best” government intervention to “correct” it. Rather, the risks can be catastrophic. Regulation of these concerns should thus be left predominantly at the national level, addressed primarily by security and not trade law professionals.

The current GATT Article XXI exception on national security grounds was not drafted with cybersecurity concerns at stake and it will need to be updated to address cyber threats.⁸⁵ The article currently refers to “action … taken at time of war or other emergency in international relations.” National cybersecurity precautions do not neatly fall within this text. It accordingly should be expanded to grant governments greater flexibility to define their security policies in relation to new threats (beyond immediate “emergencies”), while remaining subject to oversight through peer-review mechanisms and (possibly) judicial application of proportionality analysis on a deferential basis. For example, article 17.13 of the Regional Comprehensive Economic Partnership now includes measures “taken so as to protect critical public infrastructure, including communications, power, and water infrastructures” under the list of legitimate national security concerns.⁸⁶

The bulk of such regulatory efforts must be national where regulators and politicians are most easily held to account. Nonetheless, given the externalities of one country’s regulations on others and given reciprocal regulatory concerns, there is a role for regulatory architectures where countries adopting different economic models, holding different preferences, and advancing different interests can cooperate. That calls, on the one hand, for the retention of policy space, including the development of “regulatory sandboxes” to keep up with a rapidly changing digital world in which diverse countries may gain regulatory experience and develop alternative regulatory models.⁸⁷ On the other hand, it calls for the development of new oversight and peer-review mechanisms, together with standard setting, possibly on a voluntary, soft-law basis. Such standard setting and oversight can be allocated between the ISO, the International Electrotechnical Commission, the WTO, and other organizations, catalyzing interlinked networks of institutional oversight and peer review to foster policy learning, cooperation, and coordination.

In a world of increasing geoeconomic competition and accompanying national security concerns, there are limits to what trade agreements can accomplish. Because technological shifts give rise to automated and wirelessly connected products that are vulnerable to hacking, trade in such products acquires a greater security dimension. The US blacklisting of Huawei and other Chinese companies, the banning of the use of Huawei’s 5G technology for their wireless networks by Australia, Japan, New Zealand, and the United Kingdom, and Europe’s internal debates exemplify the concerns. In this area, the 2017 US national security plan declares, “economic security is national security.”⁸⁸ Trade wars and the decline of the rule of law for trade could follow.⁸⁹

And yet, law can be structured to alleviate some of these concerns by facilitating international coordination. To start, rising US concerns over national security suggest that the US position also is shifting toward more expansive exceptions to free data flow commitments.⁹⁰

This shift potentially could facilitate agreement, provided the exceptions are broad enough to encompass privacy and public order interests advanced by the EU, China, and others. In addition, a transnational governance architecture that convenes regulators to address common problems can enhance deliberation, reduce tensions, and thus (indirectly) be more conducive to peace. The approach set forth in this chapter is vastly preferable to the current situation in which trust that underpins a cooperative international trade legal order is eroding.

4. Competition Law.

Similarly, policymakers are reevaluating competition policy in response to digitalization and the data-driven economy.⁹¹ Policy options include regulating property rights in data,⁹² blocking oligopolists’ expansion through acquisitions that preempt competition, breaking up companies, and regulating oligopolists like utilities or fiduciaries.⁹³ There is considerable debate regarding them.⁹⁴ Winner-takes-most companies profit globally through trade, raising tensions between the companies’ home countries and third countries regulating them, as in the case of the EU investigating the practices of US data-exploiting multinational companies such as Google.⁹⁵ Countries deploying competition law to discipline foreign companies will continue to trigger trade conflicts.⁹⁶ Given divisions on competition policy, especially between the United States and EU, and given divisions in economic theory, including in relation to the diversity of social contexts, it may be difficult to address this issue in a trade agreement. Nonetheless, the issue calls for dialogue and regulatory response, including within the International Competition Network (ICN), the OECD, and UNCTAD, as well as the WTO’s system of committees and working groups. Although the OECD has organized a series of sessions on these issues and the ICN spent its annual meeting in 2019 at Cartagena addressing them, much more work needs to be done regarding the competition law challenges that digitalization poses.⁹⁷

5. Data Privacy.

As regards data privacy regulation, countries’ approaches again will diverge based on different preferences. Nonetheless, structures can be developed where countries discuss their common concerns and work to free data flows so long as core concerns are met. The European Court of Justice’s invalidation of the US-EU Privacy Shield, while upholding the validity of the EU’s “standard contractual” clauses for data transfers, illustrates the challenges posed.⁹⁸ Nonetheless, bilateral negotiations, complemented by the development of common international principles and standards, together with ongoing judicial oversight, present a path forward. In practice, jurisdictional conflicts, including the assertion of de facto or de jure extraterritorial jurisdictional power, must be managed continuously.⁹⁹ In these cases, a key challenge for all countries will be how to protect individual information in a world where AI increasingly can identify individuals even when data is processed to be anonymized and deemed “nonpersonal.”¹⁰⁰ Given the transnational implications of any policy, and given the role of companies in governing data usage, there is a need not only for governments to develop rules, but also for domestic and transnational civil society organizations to be incorporated within governance mechanisms to engage with governments and corporations.¹⁰¹ Once more, structures can be developed outside the WTO for information exchange, peer review, and norm development to address privacy regulation concerns. But the WTO committee system also can be engaged in coordination with such other international bodies.

6. Inequality.

Not to be forgotten, societies face rising inequality that the data-driven economy exacerbates. Like trade in goods, free flow of data enhances efficiency and thus welfare gains, but also facilitates economic processes that exacerbate inequality in ways that can threaten social stability and international cooperation. Liberalization of data flows should not be addressed without complementary social policies. For conventional trade theorists, social equality and trade adjustment assistance should be left entirely to the national level. Many reference Scandinavian social welfare and job flexicurity policies to show how this can be done.¹⁰² Again, such regulatory power should reside predominantly at the national level, which is most democratically legitimate. However, trade agreements can facilitate governments’ ability to address social inclusion policies.¹⁰³ At a minimum, trade agreements should not directly or indirectly constrain governments from adopting necessary policies domestically. They must accommodate (and not foreclose) mechanisms that enable states to address labor and other social concerns.¹⁰⁴ Such agreements also should address (or support addressing) tax evasion and avoidance so that governments can fund social welfare and job flexicurity policies.¹⁰⁵ These agreements should be developed primarily outside of the WTO. However, since these policies implicate trade, trade liberalization initiatives for the digital economy could be made contingent on their conclusion. Similarly, trade agreements could explicitly recognize the ability of countries to address social dumping concerns, as most recently developed in the United States–Mexico–Canada Agreement.¹⁰⁶ Otherwise, the economic dynamics of trade liberalization in the digital economy could further increase inequality, undercut domestic solidarity, and, in turn, empower nativist and politically populist domestic movements that undermine international cooperation and good will, as well as national democratic systems.

V Conclusion

The world needs international institutions to enhance international deliberation, cooperation, and exchange, but international institutions must be careful not to overreach. Normatively, there are efficiency and fairness reasons for agreements to accommodate policy space so that governments may respond effectively to different preferences and priorities. Politically, when international trade law overreaches, it can spur populist backlash so that the system unravels.

The challenges raised in Section II will not be resolved through a traditional trade agreement alone. The data-driven economy is developing at a rapid pace for which governments lack regulatory experience. Given the risks – ranging from systemic risks to risks to democratic institutions, national security, and personal privacy – trade negotiators should proceed with caution and humility. The issues raised are not clearly protectionist, as with tariffs, but rather entail regulation addressing diverse public policy concerns. While one of the purposes of international trade agreements is for national political bodies to “tie themselves to the mast” to avoid the siren call of protectionism,¹⁰⁷ this rationale is inapt when applied to regulation. Democracies should be able to elect leaders that change orientation regarding the appropriate mix of free data flow and regulation to protect security, privacy, and other concerns. Trade agreements constraining their ability to do so curtail democracy. Because governments weigh tradeoffs regarding the balance between free data flow and other policies in different ways, each country should be free to change its mind. International trade law should not foreclose these domestic debates and choices.

At the same time, governance mechanisms are needed in an interdependent world to address common challenges and the externalities that one country’s regulations pose for others. Governments should be required to provide equal treatment and due process to affected foreigners domestically, and to provide public policy justifications for the regulations they adopt before transnational peer review and other mechanisms. Section IV advanced a pragmatist, transnational governance architecture focused on regulatory cooperation and learning as an essential complement to WTO “hard” rules backed by dispute settlement.¹⁰⁸ This form of international governance can interact with national regulation in ways that enhance trade, with its accompanying welfare benefits, as well as regulatory efficacy, learning, and adaptation. It is the best way for trade law to address the challenges raised by the data-driven economy.

The approach set forth in this chapter differs significantly from the “grand bargains” that characterized the creation of the WTO. It is much humbler, grounded in uncertainty regarding the digital world and what it means for societies and individuals. Trade agreements, in turn, should approach the issues with caution, leaving regulation predominantly at the national level, while recognizing common standards in some areas to facilitate trade in goods and services (such as regarding electronic signatures and authorizations), coupled with structures that catalyze experimentation and exchange of knowledge and practices regarding the challenges that all societies will continue to face and to which they must respond. These processes can facilitate learning and, possibly but not necessarily, convergence over time.

There are clear limits to this governance alternative. Commercial interests and countries will contend that there is certainty regarding “best” policies and they will attempt to use leverage and persuasion to extend these policies globally. “Learning” is difficult to facilitate where interests have strong incentives to think otherwise. Yet, even then, such processes will make differences more transparent, while still leaving open the possibility of learning from experience that, potentially, can lead to policy adaptation.

The future can be governed in worse ways or better. Law at the international and national levels and their transnational interaction will help constitute that world. In the face of uncertainty, there is a critical need for agonistic deliberation, debate, and policy experimentation. Karl Polanyi in his book The Great Transformation described what occurred when governments lost control of unleashed markets in the first half of the last century.¹⁰⁹ We know how that ended. With the data revolution and the rise of AI, the risks are high. The choices societies make today will shape which science fiction remains fiction. It is a brave new world. A future will arrive that we have yet to imagine, but that (hopefully) we can muddle through.

3 Global Law in the Face of Datafication and Artificial Intelligence

Rolf H. Weber

I Introduction

Law is regularly challenged by new societal developments. Therefore, its stabilizing function is at risk in the globalized world if technology moves fast and changes the bases of human interactions. Eventually, law is no longer able to provide support for the reorientation of civil society in the context of a potentially highly dynamic environment. However, in a transnational legal system, such as the international trade regime, the evolution of expectations in world society must be channeled in order to avoid social differences that lead to disruptions.¹ Therefore, law cannot disregard technological developments.² This contribution examines challenges to the global legal framework caused by recent (primarily technological) developments. At the outset, the characteristics of the law as a structural system are outlined. Thereafter, potentially changing factors, such as technology-driven datafication (big data, cloud computing) and artificial intelligence (AI), will be briefly addressed. Based on this foundation, the main component of this contribution analyzes a desirable digital governance and the regulatory principles of a data-driven world with regard to the establishment of global legal standards in the international trade context.

II Legal Framework and Technological Advances

A Law as a Structural System

Law as a structural system gives guidance about desired behavior, thereby stabilizing normative expectations.³ In principle, legal concepts can help to support adequate normative reasoning, since the addressees of legal provisions are supposed to acknowledge the authority of the rule-making body and comply with the law.⁴ The functions of law are crystallized in rules and institutions that underpin civil society, facilitate orderly interactions, and resolve disputes and conflicts that arise in spite of such rules.⁵ The normative framework allows people and businesses in a community to determine the limits of what can and cannot be done in the collective interest.⁶ Therefore, the rule of law helps to achieve a high, discretion-limiting degree of certainty and predictability in social relations and transactions.⁷

Irrespective of the manner in which norms actually influence behavior, law-making bodies must understand the different processes that facilitate legal developments, insofar as law often proves to be path-dependent.⁸ This assessment corresponds to the reality that the legal system is linked to other social systems, such as technological advances or ethical relations; that is, law only enjoys relative autonomy and is confronted with technological uncertainty.⁹ The structural coupling that occurs between and among the legal system and other systems requires the implementation of mechanisms that allow a change of law called for because of societal needs and circumstances.¹⁰ Only when such mechanisms are institutionalized can the continuous existence of the legal system be secured over time.¹¹ In other words, even if the law requires predictability and a stable structure, the adaptability of legal rules keeps the law intact in cases of social variation.¹²

Nevertheless, some substantive legal values are so fundamental that their abolition would totally undermine the function of law in society. On the one hand, human rights, such as the freedom of expression or the nondiscrimination principle, represent major (even untouchable) constitutional values. On the other hand, legal order can hardly function without property rights and privacy rights.¹³ As a consequence, certain legal pillars may not be “overruled” by technological developments.

These pillars of law as a structural system must be mirrored against the challenges caused by recent technological advances in order to avoid “clashes” that may harm civil society.

B Datafication as a Technological Trend

The term “datafication” was coined by Mayer-Schönberger and Cukier in 2013, primarily with respect to the then-new phenomenon of big data.¹⁴ Datafication, which has become a buzzword in the new IT world, refers to a technological trend that is able to “convert” many aspects of modern life into computerized data.¹⁵ Hereinafter, not only are big data and cloud computing addressed but also new developments in the AI context.

1 Big Data and Cloud Computing

Over the past few decades, an immense amount of data has been generated through the (cross-border) flow of information, humans, products, services, and capital. These developments have resulted in data protection concerns, as well as the specific problem of users of online goods and services “paying” for products offered by disclosing data without assessing the nature and value of the data. The recent European Union (EU) Directive on digital content even regulates the payment of online services through the provision of data.¹⁶ Furthermore, data holders are often unaware of how much data is collected and stored about them.

Big data is a term coined for datasets whose size is beyond the ability of commonly used software tools to capture, curate, manage, and process within a tolerable degree of time. The phenomenon of big data analytics is often characterized by four elements, namely volume, variety, velocity, and veracity (the “4 V”). “Volume” refers to the especially large amount of data; “variety” makes it clear that the data is derived from manifold sources and formats; “velocity” mirrors the high speed of the data processing; and “veracity” reflects the reliance on the correctness of the data. A special feature of big data from a legal perspective is the fact that the traditional concept of causation is replaced by the concept of correlation.¹⁷

For some years, international organizations such as the Organisation for Economic Co-operation and Development (OECD), as well as academic voices, have assumed that big data is potentially a key driver of innovation, productivity growth, and economic competitiveness.¹⁸ The global exchange of data requires unrestricted cross-border data flows in order to realize its merits. In this context, big data analytics is able to improve the outcome of data processing in manifold areas.¹⁹ Examples include the automotive industry, which evaluates ideas submitted through its “virtual innovation agency,” and the health sector, which collects data on larger populations in order to reduce disease, bringing rapidly and accurately identified drugs to the market and providing better healthcare by enabling the application of evidence-based interventions.²⁰

A further technical innovation is cloud computing, which facilitates cross-border data flows in order to take advantage of cheaper on-demand computer capacity that can be scaled and paid for as needed. Cloud services encompass software and access to processing, email, storage, and other computer resources. From a business perspective, cloud computing turns a fixed IT cost into a variable operating cost, and from the angle of human resources, cloud computing underpins the capacity of people to work remotely.²¹ Major challenges include connectivity and the compliance of cross-border data flows with data protection and data security requirements.²²

2 Artificial Intelligence

Algorithms are able to automate the “production” of goods/services and to facilitate the selection and filtering of information in various ways, thereby attributing relevance to certain data and moderating content. Algorithms are the foundation of AI as a machine-based system that is capable of influencing the environment by making recommendations and predictions without human input.²³ An AI system lifecycle encompasses various phases, such as the design and modeling of data, the verification and validation of data, as well as the deployment, operation, monitoring and storage of data.²⁴

AI allows for the implementation of a “regime” of automated decision-making to be conducted in a highly timely and effective manner. Such automation is primarily feasible in situations that do not require a specific human intervention. The new technologies can also transform traditional manufacturing into smart manufacturing, focusing on digitization from early product design through maintenance at the product’s end of life by using advanced sensors and big data analytics.²⁵ As examples, and only at the very beginning of data-driven innovation potential, the following phenomena can be identified: (i) driverless cars will become a reality, and (ii) robotics will move forward to become a widespread industry tool based on rapidly advancing AI.

Notwithstanding the fact that AI has many benefits, some risks also cannot be overlooked. Therefore, from a normative concept of civil society, a few questions must be considered in an interdisciplinary manner:²⁶ (i) Do AI processes comply with fundamental constitutional principles? (ii) Is the AI application based on a sufficient legal foundation? (iii) Do AI processes comply with the applicable requirements of data protection laws? (iv) Who is in charge of monitoring the socially responsible use of AI, and who is liable in case of a failure caused by an algorithm?

These questions merit appropriate answers, even if solid responses that serve the needs of global society are not easy to find. Nevertheless, a reconciliation of AI’s chances and risks must be kept in mind if an adequate legal framework for digital governance in global relations, particularly in the international trade regime, is to become a reality.

III Reconciling Global Trade with Global Law

A Rule-Making in the Digital World

1 Globalization and Governance

Globalization is not a clearly defined term; its concrete meaning depends on the given substantive component of societal life.²⁷ (i) Legal globalization concerns the harmonization of states’ normative orders, or the implementation of cross-border legal rules. (ii) Cultural globalization addresses those issues related to manifold social policies. (iii) Commercial globalization reflects the existence of increased transnational businesses and economic activities. Global law as an aspect of legal globalization is confronted with new concepts, examining institutional differentiations and elaborated procedural techniques.²⁸

The term “governance,” which stems from the Greek word “kybernetes” and the Latin word “gubernator,” means a steersman, and it must be recognized for its importance. Governance can be addressed from the perspective of different disciplines; nevertheless, at whatever level of social organization it may take place, governance refers to the appropriate business conduct of a private or a public body. In this context, some key questions must be asked and answered:²⁹ (i) Who is entitled to set the rules? (ii) In whose interest are the rules? (iii) By which mechanisms are the rules created? (iv) For which purposes are the rules designed? There is a need to develop overarching networks and negotiation systems between different stakeholders, thus forming a “cooperative approach to governance” that includes the entire society, hence dividing responsibilities between public and private actors.

Based on the described notions of “globalization” and “governance,” the development of a broader rule-making approach encompassing the needs of the international trade regime and of the digital environment of today’s societies appears to be unavoidable and equally justified.

2 A Broader and Better-Coordinated Rule-Making Approach

Assessing the dichotomy of regulatory sources and the emergence of new regimes introduced by civil society, adapted transnational concepts must be developed in the global law environment.³⁰ Institutions can lead states to more cooperative behavior than they otherwise might have adopted, building mutual connections from peripheral points in federative or associate forms.³¹ In realizing an appropriate governance framework, civil society involvement should be strengthened. The facilitation of networking opportunities and public support from concerned persons and organizations in the relevant policy field should also be considered; that is, a policy vision must be developed.³²

In the Internet governance context, a new approach has been developed and applied in part, namely the multistakeholder participation model, which attempts to involve all concerned persons and organizations in the public and private sphere in the discussions and negotiations of the regulatory framework.³³ Practical experience has shown that some basic challenges need to be addressed in order to ensure that the multistakeholder concepts are successful. Four fundamental questions must be tackled:³⁴ (i) How do governance groups best match challenges with organizations and networks? (ii) How can governing bodies/entities be most able to help develop legitimate, effective, and efficient solutions? (iii) How should the flow of information and knowledge necessary for successful governance be structured? (iv) How can different governance groups approach coordination between available governance networks in order to avoid conflicting interests?

Over the last few years, the globalization of the international legal framework, among other fields in international trade law (particularly due to the “outdated” classification regime in the World Trade Organization (WTO) Agreements), and equally in the area of Internet law (due to a stronger emphasis on state sovereignty), has suffered some setbacks. In light of the fact that the legal fragmentation caused by national laws jeopardizes an appropriate design of global law in a network society, coherence between different initiatives should be strengthened in order to overcome trends leading to various forms of fragmentation or distorted regulatory regimes.

The recently developed term “legal interoperability” addresses the process of creating legal rules that cooperate across jurisdictions.³⁵ This objective can be realized in a matter of degrees, as many options exist between a full harmonization of normative rules and a complete fragmentation of legal systems.³⁶ As is so often the case in the real world, striking the correct balance is of utmost importance. While an excessively high level of interoperability could cause difficulties in the management of the harmonized rules and fail to acknowledge social and cultural differences, a level too low could present challenges to smooth social interaction.³⁷

B Regulatory Principles for the Data-Driven World

In view of these technological innovations, the legal order is confronted with the need to establish an international regulatory framework for the data-driven world that implements the following basic regulatory principles.

1 Transparency

All involved stakeholders should promote a culture of transparency, enshrining the disclosure of data logics and access to the applied algorithms and datasets.³⁸ Transparency is usually defined as “easily seen through … evident, obvious, clear.”³⁹ Transparency means understandable and forward-looking information, appropriate to the context and the state of the art, in order to make stakeholders aware of their interactions (in an ex ante or ex post data-centered decision-making process).⁴⁰

Transparency requires robust and general rules, not necessarily more regulation; that is, the improvement of transparency does not mean a quantitative increase in information, but rather “more” in terms of higher information quality.⁴¹ A future-oriented understanding of transparency should observe the following elements:⁴² (a) the existence of publicly reliable information; that is, substantive quality standards related to information, supported by an adequate legal framework; (b) the designation of the information recipient as a holder of rights and an essential component for perception and transparency; and (c) the availability of disclosure procedures and observance of the time element; that is, transparency implies constant visibility of information.

Providing information about the type of input data and the expected output, explaining the variables and their weight, and shedding light on analytics architecture usually contribute to transparency with respect to AI algorithms.⁴³ Nevertheless, a generic statement on the use of AI does not allow for the easy assessment of all challenges and risks; concrete circumstances do play a role, which means that solutions focused on disclosing specific information about the applied algorithms may be the best option.⁴⁴

2 Accountability

All stakeholders involved in datafication and AI mechanisms should be accountable for the proper functioning of the systems employed, as well as the integrity of the regulatory environment.⁴⁵ Accountability helps to ensure an environment in which individuals and enterprises assume their respective responsibilities. The first legislative attempts to meet this requirement can already be seen: for example, in the field of data protection, the EU General Data Protection Regulation calls upon organizations to apply a “Privacy by Design/Default” approach and – under certain circumstances – to conduct a “Data Protection Impact Assessment.”⁴⁶

Accountability encompasses the obligation of one person to another, according to which the former must give an account of, explain, and justify his/her actions and decisions against criteria of the same kind.⁴⁷ Therefore, the proportionality principle, which inspires an adequate and appropriate deployment of big data analytics and AI, should apply.⁴⁸ Accountability also relates to good governance, which was previously addressed. The development of respective concepts in public institutions and private enterprises requires publicly assessable accounts as a precondition for a sustainable society.⁴⁹

The obligation to be accountable encompasses the task of disclosing information about the actual AI processes. To improve respective foreseeability, standards should be developed and introduced that moderate the behavioral requirements in a concise manner. Furthermore, the responsibility of the accountable person to ensure that concerned individuals are protected from damages when having suffered a detriment is to be legally developed in a precise way.⁵⁰

3 Safety and Robustness

Technological innovations must be safe and robust throughout their entire lifecycle so that data-driven communications and transactions can overcome adverse conditions or foreseeable potential misuse. Safety and robustness also extend to the terms of security and resilience. Therefore, the traceability of the datasets, processes, and decisions must be secured. Furthermore, the risks with respect to safety and robustness should be managed throughout the entire process of hardware or software applications. Consequently, the execution of impact assessments with respect to technological risks is necessary.⁵¹

International instruments already state that the likely impact of AI on civil society must be adequately taken into account in order to safeguard fundamental rights. For example, the Recommendation of the OECD Council on Artificial Intelligence refers in part 1.4 to robustness, security, and safety as follows:⁵²

AI systems should be robust, secure and safe throughout their entire lifecycle so that, in conditions of normal use, foreseeable use or misuse, or other adverse conditions, they function appropriately and do not pose unreasonable safety risk. To this end, AI actors should ensure traceability, including in relation to datasets, processes and decisions made during the AI system lifecycle, to enable analysis of the AI system’s outcomes and responses to inquiry, appropriate to the context and consistent with the state of art.

Furthermore, the use of AI through modern data-processing techniques and the trend toward implementation of data-intensive processes require a more advanced understanding of risk assessment by individuals, businesses, and public organizations, since possible adverse outcomes stemming from data processes cannot be excluded.⁵³ Risks and compliance assessments are not only justified by collective social and ethical values, as well as the nature of fundamental rights and treatments affected by AI application. They also represent an opportunity to better foster public trust as a key objective of the information society.⁵⁴

4 Interim Assessment

As stated earlier, the use of modern data-processing techniques and the trend toward implementation of information-intensive data analytics require a more advanced understanding of risk assessment. In particular, the fact that automated decision-making may have an impact on fundamental rights (including the nondiscrimination principle), as well as collective social and ethical values, must be addressed.⁵⁵ AI programs can affect various human rights, i.e. the right to human dignity, the principle of nondiscrimination, privacy rights, and the guarantee of self-determination.⁵⁶

Furthermore, risks and compliance assessments are not only justified by the nature of the rights and freedoms potentially affected by datafication (or big data analytics), as well as AI applications. In a participatory environment such assessment can contribute to an increased level of trust.⁵⁷ As the most recent political developments in the political arena have shown, trust plays an important role in the context of the international trade regime. Trust can even be seen as a central pillar of the globalized governance, as previously discussed.

C Combatting Distortive Interferences

In the digital society and economy, the factual possession and control of data is key. New technologies, such as datafication and AI, lead to situations in which the data control function is primarily assumed by large private enterprises and by governments. The use of big data results and the application of algorithms give these entities information power,⁵⁸ which can be exercised in either a good or bad way. Improvement in healthcare or the strengthening of measures to protect cybersecurity undoubtedly have a positive effect. However, the misuse of data is also possible, for example with the objective of spreading inaccurate, embarrassing, or misleading information or controlling the data exchange for one’s exclusive benefit, insofar as the holder of data can become a “data demagogue,”⁵⁹ contradicting the basic principles of an appropriate international trade regime.⁶⁰

1 Anticompetitive Behavior

From the perspective of competition law, different issues are at stake. An initial aspect concerns the changed market parameters. Digital markets, as well as the exchange of communications in a digital society, should increase the possibilities for participation among all interested individuals and organizations/businesses. However, the markets tend to be dominated by a few firms (Alphabet (Google), Amazon, Facebook, and Apple (GAFA)). Similar experiences are found in East Asia (Alibaba, Tencent).

The overwhelming dominance of a few players in digital markets causes anticompetitive concerns, which are primarily due to the risk that a market-dominant position has been misused.⁶¹ The oligopolistic market structure is now challenged by competition authorities (for example, the Directory Competition of the EU), as well as political bodies in the United States and Europe. The outcome of these interventions is unresolved for the time being.

Other behavioral problems are also on the horizon – for example, the tacit collusion by big firms through the use of parallel algorithms.⁶² For good reasons, primarily the OECD, as one of the most important international organizations in the economic field, is thoroughly analyzing the respective challenges.⁶³ So far, competition law does not appear to be fully fit to tackle these problems. The lack of general competition law principles in the WTO framework must be seen as another disadvantage for the international trade regime.⁶⁴ Furthermore, the lack of coherent competition policies among jurisdictions leads to the disadvantage that national competition authorities are ill equipped to effectively address the anticompetitive data practices; the need for more streamlined standards between antitrust regimes is obvious.

2 Denial of Network Neutrality

A second issue is network neutrality: all market participants (providers of goods and services, as well as consumers) should have unfettered access to the digital infrastructure.⁶⁵ The preferential treatment of some businesses or individuals results in the risk of competition distortion. Therefore, discrimination toward certain market participants must be considered unjustified.

In some countries (for example, in the EU), network neutrality provisions do exist.⁶⁶ The national implementation of the neutrality principle, however, does not always cover existing needs. In the United States, the trend has moved away from regulatory intervention because of the existing political climate. Obviously, rules cannot replace the market. However, illegitimate discrimination does have a substantial negative effect on the international trade framework.⁶⁷

3 Data Localization

A third challenge, more due to efforts of governments than private enterprises, concerns data localization requirements. Around the globe, such requirements are implemented in a variety of forms.⁶⁸ Information and communications technology companies may be obligated to host all subscriber and consumer data locally within the country; in some instances, only information covering certain substantive areas (for example, health) must be stored in the country.

Data localization reduces access to data and digital technologies and may also be counterproductive. Requiring data localization in relation to cybersecurity increases data vulnerability in a single jurisdiction, making it easier to target and possibly preventing data backups in globally distributed data centers.⁶⁹ Obviously, such provisions raise the costs of access to, and use of, data, thereby reducing gains from digital trade.

4 Interim Assessment

As the aforementioned deliberations demonstrate, global law is exposed to “data demagogues” who strive to interfere with cross-border data flows. Various forms of unjustified market interventions, such as anticompetitive behavior, denial of network neutrality, and data localization provisions, can have a competition distortion effect. Such an outcome must be avoided or, at the very least, mitigated by the regulator.

In addition, legal uncertainty, caused by technological advances, leads to discretionary power that is not always consistent with the rule of law.⁷⁰ This fact primarily concerns governments and other regulatory bodies. However, private enterprises can also have power in factual forms. A balance between differing interests is often difficult to find, but efforts to reconcile such interests appear to be unavoidable.⁷¹

D New Digital Trade Regime

1 Outdated Goods and Service Classifications

So far, a commonly agreed-upon definition of what represents digital trade does not exist. The WTO has addressed the “production, distribution, marketing, sale, or delivery of goods and services by electronic means” as part of digital trade in its Work Programme on Electronic Commerce of 1998.⁷² Cross-border data flows enabling digital trade have not played a major role in the previous discussions, notwithstanding the fact that their appearance is important.

Legal scholars of WTO law have been tackling the distinction between goods (General Agreement on Tariffs and Trade, GATT) and services (General Agreement on Trade in Services, GATS) for many years, since it is not clear which “product category” and consequently which Agreement would be suitable in addressing digital assets.⁷³ The differentiation is practically important, since the GATT offers less room for maneuvering to member states that are unwilling to liberalize digital markets than the GATS.⁷⁴

The WTO itself is well aware of the existing problems, and efforts have been initiated to remedy the situation. In the context of the 11th Ministerial Conference (December 2017, Buenos Aires), a short “Joint Statement on Electronic Commerce” was adopted, inviting participating member states to undertake further work on digital trade.⁷⁵ Most developed countries have agreed to this Joint Statement; China did it almost at the last minute. However, no concrete outcome can be seen for the time being, even though several proposals have been submitted and discussed in the WTO forum. The World Trade Report 2018 of the WTO also intensively addressed digital trade, pointing to the further transformation of the global trade regime. As a result of these investigations, the WTO has pleaded for a technology-induced reshaping of the regulatory environment.⁷⁶ Nevertheless, a solution to the existing problems may not surface in the near future.

2 New Data-Oriented Regulatory Approaches in World Trade Organization Law

An important component of the regulatory challenges concerns cross-border data flows. Foundational principles of data regulation in international trade law must be developed.⁷⁷ The elements encompass fostering digital trust and ensuring interoperability and transparency, in support of the free flow of information. Therefore, a hybrid regulatory approach based on a polycentric governance model⁷⁸ appears to be suitable.

New business models must become part of the trade rule framework. For example, digital platforms should be able to overcome barriers that have prevented small companies from participating in international trade, and from facilitating the building of trust in global transactions.⁷⁹ In summary, regulatory support for the implementation of digital platforms that are stable and trustworthy is an objective that must be pursued.

The WTO rules typically do not refer to private standards, industry best practices, or multistakeholder institutions.⁸⁰ However, these models are commonplace in the digital world and therefore must also be incorporated in international trade law. Self-regulation and – possibly – coregulation do have important merits; therefore, it appears to be worthwhile for WTO bodies to cooperate with international organizations and associations that are involved in the development of international trade standards.

Another issue concerns ongoing discussions about digital trade barriers.⁸¹ The respective objectives are often at least partly founded in sound policy or social reasoning. Examples include cross-border cooperation on cybersecurity issues, the implementation of an appropriate privacy framework, and the integration of consumer trust-enhancing measures.⁸²

Finally, legal scholars are generally of the opinion that the WTO framework should be expanded in the light of smart goods and smart services, constituting the Internet of Things.⁸³ Apart from horizontal obligations on cross-border data flows and data localization,⁸⁴ efforts must be undertaken to amend the international trade regime by way of new provisions that enable digital innovation and promote business trust.⁸⁵ For example, technical standards for digital services that are consistent with internationally recognized standards should be adopted.

E International Regulatory Cooperation

Cross-border data flows and digital trade are in the process of transforming the global legal framework. In contrast, experience over the last five years has shown that many governments have become less inclined to agree on new, internationally accepted rules and are increasingly restricting cross-border data flows.⁸⁶ The discussed data localization requirements can be seen as a new form of digital protectionism that extends beyond justified objectives, such as protection of privacy, law enforcement needs, and cybersecurity concerns. Fragmentation of the Internet is another widely discussed topic.⁸⁷

The developments identified herein are leading us in the wrong direction. Rather, efforts should be undertaken to strengthen international regulatory cooperation. Therefore, the interplay between global needs and national interests should be better analyzed, and should also lead to an appropriate design of such cooperation.⁸⁸ Legal interoperability (for example, through mutual recognition understandings) must become a regulatory objective,⁸⁹ and a convergence toward principles and standards in areas such as privacy and cybersecurity is desirable.⁹⁰ Such developments can only be successful if international regulatory cooperation between actively public and private cross-border organizations is improved (in the form of interagency coordination and compliance management measures).

In the light of growing nationalism and deepened sovereignty interests in many large and small states around the globe, international regulatory cooperation represents a difficult stand. Nevertheless, from an academic and political perspective, coordination is always better than confrontation. This assessment should motivate policymakers to place more emphasis on the harmonization of international bodies’ activities.

IV Outlook

Overarching key elements of global law in the context of the international trade regime are transparency, trust, and traceability. An optimal design for a balanced policy environment in the global trade ecosystem must consider aspects including risk assessment and ethical considerations, thereby strengthening the trust of all involved stakeholders in the global legal framework.

Regulation should become an enabler of digital innovation and should not limit business activities in an undue way. Interoperability, with respect to technical standards, data models, and AI processes, may help in the design of an appropriate normative framework. Governance measures that avoid further fragmentation may also help to realize a globally accepted legal environment. The difficulties ahead are remarkable, but not insurmountable. However, a new way of thinking is needed, setting traditional sovereignty considerations aside and moving toward new intellectual concepts.

4 Trading Artificial Intelligence Economic Interests, Societal Choices, and Multilateral Rules

Dan Ciuriak and Vlada Rodionova

I Introduction

After technology’s decade of disillusion, societies confront the decade of decision: how to address the myriad issues already encountered with digital technology in reality or conceptualized in virtual realities, as use cases proliferate and as applications gain power. As a general-purpose technology with applications that can touch on virtually any human endeavour, the integration of artificial intelligence (AI) into social and economic frameworks poses particularly thorny issues. The full extent to which it will be embraced and the terms and conditions under which it will be allowed into our lives will likely vary across jurisdictions, reflecting differences in governance structures, societal preferences, and economic interests, with regulatory decisions being made in a context of limited experience, highly imperfect information, and at best a rudimentary understanding of the complex feedbacks that will be unleashed as the integration of AI proceeds.

From a trade perspective, regulatory decisions concerning the operation of AI within societies will constitute non-tariff measures (NTMs) that condition market access for the hyper-specialized AI applications that are already in use and the many more that are under development and slated to be brought to markets over the coming years.

The multilateral trade system has some experience addressing issues encountered with the introduction of new technologies, including the range of considerations bearing on risk tolerance, such as, inter alia, the use of available scientific evidence, the factors to be considered in assessing risk, the role of international standards in establishing acceptable levels of risk, and even in providing flexibility for differences in consumer tastes and preferences (i.e. political choice, including involvement of civil society) with regard to risk, including through the invocation of the precautionary principle.

At the same time, the “dual-use” character of AI¹ and the data that train it² make national security entanglements seemingly unavoidable and perhaps even ultimately unbounded in scope, while the prospect of large valuable economic rents from AI applications incentivizes strategic trade and investment policies.³

With AI and machine learning (ML), we are navigating largely uncharted waters. The Stanford 100-year project on AI (Stanford AI100) advised against premature regulation on the grounds this could prevent the development of potentially beneficial technologies, stifle innovation, and/or drive innovation to less restrictive jurisdictions.⁴ However, given the geopolitical AI arms race currently underway, and given the lure of large prospective economic rents, there is no likelihood of the pace of development and deployment of AI actually slowing down. By the same token, the terms and conditions under which AI accesses markets will be developed through a learning-by-doing process in which societies conduct natural experiments in allowing applications while “regulatory sandboxes” are used to develop the rules that in turn pave the way for international market access.

In this chapter, we discuss the rites of passage of AI as it enters the trading system. The next section discusses the challenge of getting AI applications to market and how they are being handled. Section III then discusses the hurdles that societal impacts may throw up, including national security, political choice, and income distribution. The final section ventures a discussion of how the integration of AI into international commerce might unfold.

II Getting Artificial Intelligence to Market: Navigating the Regulatory Framework

A The Artificial Intelligence Future Is Here

If we replace the term “AI” with “smart”, we realize immediately that AI is already all around us: AI applications power the smart assistants on cell phones, the range of smart home applications now widely in use, proliferating smart applications in business, and above all increasingly intelligent machines that combine a plethora of AI-driven functions to acquire increasingly flexible, human-like capabilities, up to and including humanoid robots carrying on conversations on stage⁵ and AI television news anchors reading the news.⁶

Stanford AI100 places widespread introduction of ML in software services and mobile devices as starting in 2000,⁷ even before the breakthroughs in technology that powered the development of modern AI. Kelly identifies these breakthroughs as follows:⁸ the development of “deep learning” based on stacked neural networks by Geoffrey Hinton in 2006 (which effectively industrialized learning); the application of parallel processing computer chips to neural networks by Andrew Ng and his team at Stanford in 2009; and the accumulation of big data, which greatly increased with the mobile revolution that followed the introduction of the iPhone in 2007. Agrawal, Gans, and Goldfarb⁹ place the commercial debut of AI only in 2012. Ciuriak and Ptashkina¹⁰ place the dawn of the data-driven economy circa 2010, more or less coincident with the breakthroughs that powered the commercial application of AI.

Well before these breakthroughs, the development of regulatory frameworks and quality assurance systems for AI were already underway, since the basic issues raised in developing standards for AI were already encountered in developing quality assurance for “expert systems”, which date back to the 1960s.¹¹ These systems were based either on data (encoded knowledge of a very specific area) or deep learning based on comprehensive structural knowledge of the subject matter, and used an “inference engine” that sought to mimic the decision-making process of a human expert.¹² The generic problems raised by these applications are as follows:

• The validation of an expert system requires human experts, who are in some sense more expert than the expert system itself. But leading human experts do not always agree, experts might not be available, and some might be biased; and the ethical contribution to a decision might be different from expert to expert.¹³ And how does one validate AI that performs at levels superior to humans?

• AI trained on data can only draw inferences within the scope and experience base of those data. But there is no way to definitively specify what is comprehensive coverage of the knowledge required to draw an expert inference. For example, humans often reason by analogy; how does one code the intuition that informs when an analogy is apt?

• Conventional validation requires precise testing of outputs. But definitive assessments are not possible with AI that will draw inferences from new information, even though the AI can be tested for repeatability and stability with given data inputs.

In the modern era, where AI is developed in non-deterministic processes through training on big data, in which the decision-making process cannot be broken down into sub-programs that can be individually tested, the problem becomes still more complex. While “black box” testing approaches have been developed, these are considered to be more “workarounds” than solutions to the problem of quality assurance.¹⁴ Notably, an AI chatbot trained on Twitter quickly became a foul-mouthed racist and had to be shut down,¹⁵ highlighting the issues raised for regulation by open-ended training data.

Notwithstanding these essentially unbounded concerns, use cases for AI through expert systems have proliferated and myriad applications have, as noted, already passed the applicable regulatory procedures and industry-established quality benchmarks without apparently encountering significant problems in terms of accessing international markets. How was this done? We turn to this question next.

B Horizontal Standards

The modern era of powerful AI emerged in a regulatory context informed by the experience acquired developing quality assurance for expert systems within the software engineering stream, under the auspices of the International Organization for Standardization (ISO) and the International Electrotechnical Commission (IEC). The ISO/IEC 90003 Software Engineering standards for expert systems date back to 1998. At the industry level, relevant quality assurance approaches include Total Quality Management (TQM), Six Sigma, and a number of others.

With technology rapidly advancing, many AI-specific standards are being developed at the national and international levels. For example:

• The US National Institute of Standards and Technology (NIST) has released a Plan for Federal Engagement in AI Standards Development,¹⁶ which lists nine areas of focus, including human interactions, performance testing, and trustworthiness. The US approach is generally “light touch”, relying on self-regulation by industry, and emphasizing commercial opportunity.

• China’s Standardization Administration of China (SAC) has released a White Paper to support China’s international engagement on AI standards for key technologies and interoperability, including on algorithmic transparency, liability, bias, and privacy, among other ethical and security issues.¹⁷

• The European Commission has, inter alia, issued a White Paper on AI; a report on safety and liability implications of AI, the Internet of Things (IoT), and robotics; and, through a High-Level Expert Group, ethical guidelines for trustworthy AI.¹⁸

• Japan has established an Advanced Integrated Intelligence Platform Project (AIP), which features a comprehensive programme on AI, including standards.¹⁹

• The United Nations has been active on the human rights aspects of AI, developing recommendations on ethical issues raised by the development and application of AI.²⁰

• The Organisation for Economic Co-operation and Development (OECD) Ministerial Council has agreed a set of high-level OECD Principles on Artificial Intelligence.²¹

• As regards the deeper issues raised by ML, international standards under development include the ISO/IEC CD 23053 (“Framework for Artificial Intelligence Systems Using Machine Learning”) and the ISO/AWI TR 23348 (“Statistics – Big Data Analytics – Model Validation”). These may provide a common approach for assessing compliance of AI software in high-risk applications in regulated industries.²²

Trustworthiness standards are of particular interest as they cover a gamut of difficult issues, including accuracy, explainability, resiliency, safety, reliability, objectivity, and security.²³ The ISO technical committee on AI published its first overview of trustworthiness in AI only on 28 May 2020.²⁴ While this document discusses these various aspects of trustworthiness, the specification of levels of trustworthiness for AI systems remains beyond the scope of the ISO process. And, of course, it is precisely the level of trustworthiness where social and political choice is decisive, as demonstrated by the heated debate over the use of facial recognition by public authorities.²⁵

Progress in these areas is being driven by necessity because AI is being deployed commercially and regulation cannot wait. For example, the European Union’s (EU’s) General Data Protection Regulation (GDPR) establishes explainability as a right: under the GDPR, individuals have a right to ask businesses that use their personal data for automated processing how decisions that affect them were made – and businesses must be able to explain to be compliant. Moreover, the GDPR establishes the right to request human intervention for review of an AI decision, and grants new investigatory, advisory, corrective, and punitive powers to the EU’s data protection authorities, putting firms on notice.²⁶ Explainability has also engaged the attention of the military in developing protocols for military use of AI.²⁷ “Explainable AI” has thus become an important frontier for research²⁸ – and, indeed, has acquired its own acronym, “XAI”.

In short, while horizontal AI-specific regulations were largely missing in action in the early phase of integration of AI into the economy and society, this gap is fast being filled.

C Vertical or Industry/Product-Specific Standards – Mechanical Functions

The largely unimpeded commercial progress of AI to date has arguably reflected several characteristics of the market, in addition to the general absence of restrictive horizontal standards:

• Industrial applications were developed by highly sophisticated companies working with sophisticated clients, including government agencies, with the AI embedded in machinery that was subject to an industry- or sector-specific (hence “vertical”) regulatory framework.

• Consumer-facing applications were embedded in products marketed mostly by “superstar” firms (e.g. cell phones with “smart” assistants and other AI-powered applications) and subject to product-specific standards and regulations administered by designated agencies with deep expertise in regulating on behalf of unsophisticated households.

The least problematic applications from a standards perspective are those where AI performs purely mechanical functions; performance in these types of functions tends to be measurable and the behaviour of the AI, even with learning, converges to an observable standard. AI applications that replace human cognitive/decision functions and involve agency on the part of the AI (i.e. where the AI makes autonomous decisions with real-world impacts) attract more regulatory attention. Applications can of course combine mechanical and cognitive functions. Accordingly, certification for domestic markets of particular AIs may involve a multiplicity of approvals.

One of the most straightforward uses of AI is to automate routine business or production processes or to reassign specific human functions to machines for accuracy. These types of applications have been adopted rapidly and widely and spread globally, without seemingly encountering barriers.

Industry is already familiar with industrial robots. Integrating AI into an industrial robot makes the robot more intelligent in the sense of being able to perform more complex functions. In such traditional industrial robotic applications, robots can substitute for particular human roles entirely and even work in isolation from humans. A quintessential example is provided by the role of AI in supply chain management automation. The integration of AI, improved sensors, sophisticated warehouse management software, IoT telecommunications systems, and automated robotic technology effectively allows warehouses to operate autonomously on a literally “lights-out” basis.²⁹

More commonly, AI applications in workplace settings support human–robot interaction within a shared workspace. Instead of replacing people with autonomous modules, such collaborative AIs (so-called cobots), trained with ML techniques and big data, work with humans, providing extra precision, speed, and consistency without fatigue in routinized tasks, while leaving the less routine aspects to humans. There are many examples of cobot applications already in use.³⁰ One example is “pick and place” functions, which involve mundane repetitive tasks that require cognition and result in errors due to boredom; such jobs can be more efficiently (and more safely given the propensity for repetitive strain injuries) done by robots with advanced vision systems and trained by AI, while the human member of the team focuses on aspects that require decisions. Another is “packaging and palletizing”, which includes a range of functions from shrink-wrapping and boxing to placing products on a pallet for shipment.

Routine quality inspection functions are also being turned over to cobots that inspect finished parts by comparing images from multiple high-resolution cameras that capture all angles of a product simultaneously and are not prone to mental fatigue. More sophisticated cobot applications under development include an aircraft inspection assistant cobot in the “Hangar of the Future”, which automates aircraft inspection as part of maintenance, repair, and overhaul operations.³¹ Trucking is likely to go down this route with AI systems taking over the long-haul highway portions, leaving the first and last mile which involve more complicated environments to human drivers.

While many (if not most) of these tasks involve AI enabling the replacement of physical labour by robots, there are other cases where the AI replaces the skilled function. It is typically the case in these instances that the AI is hyper-competent and the AI’s work is superior to the human’s. This is likely the future for much assembly-type manufacturing that requires precision work such as automotive and aircraft assembly – see, for example, the use of AI and ML techniques to refine the installation of aircraft skins by Boeing.³² Healthcare has emerged as a major use case for cobots where the AI is hyper-competent in this sense, particularly surgery-assisting cobots that use AI to improve the precision of surgical procedures.³³

Other interesting examples of this include Sony’s Hawkeye in tennis, which uses AI to make line calls. In tennis, the AI over-rules the human line caller in a challenge. In the 2020 US Open, AI made all the line calls on fifteen of the seventeen courts;³⁴ meanwhile, in the 2020 French Open, the failure to deploy the AI line-calling system was decried following an apparent mis-call at a critical moment in the match between Canada’s Denis Shapovalov and Spain’s Roberto Carballes Baena, leading to a rising tide of sentiment within the professional ranks in favour of the system. A retail market version (“InOut”) is already in use.³⁵ Similar applications have been developed for goal-line decisions in football.³⁶ Baseball is experimenting with turning over the ball-strike calls to AI based on analysis that human umpires incorrectly call pitches (e.g. Chen et al. find that umpires call only about 60 per cent of close pitches accurately and show systematic bias due to effects such “anchoring” or the “gambler’s fallacy”³⁷). In its first baseball application, the human is advised by the AI and it is the human that makes the definitive call.³⁸

Clearly, such AI applications have navigated complex sector-specific regulatory systems to get to market (such as those for medical devices or civil aviation) – or none at all (such as those for sports). From a trade perspective, the technology typically enters a new market either through foreign direct investment or through a transaction between a sophisticated supplier and a sophisticated buyer with considerable tailoring of the application to the specific circumstances and needs of the buyer. Accordingly, the future for the international dissemination of such AI applications does not appear to be any more problematic than its experience to date has been.

D Vertical or Industry/Product-Specific Standards – Cognitive/Decision Functions

AI that performs human cognitive/decision functions, in contexts where agency is involved and the decision criteria are less clear-cut and the consequences more significant than making a ball/strike call in baseball or a line call in tennis, will likely face substantially higher hurdles to achieve acceptance. The essential analogue would be competence regulation for human experts. Depending on the nature of the judgements the AI would be called to make, how it is trained might come into play.

In the legal domain, for example, the amount of unstructured data mobilized for legal cases is enormous. It is no surprise that natural language processing (NLP) and image recognition techniques lend themselves to extract efficiencies in the preparation of legal cases. As the marketing of these tools is between sophisticated businesses, there are no apparent issues.

At the same time, deploying advanced algorithms in actual legal procedures raises concerns related to the core principles and guarantees of judicial systems. In this regard, the European Commission for the Efficiency of Justice (CEPEJ) adopted the first European Ethical Charter³⁹ on the use of AI in the justice system in 2018. The charter outlines principles to guide policymakers, legislators, and justice professionals to help them to embrace and, where needed, confront the spread of AI applications in judicial systems. These principles aim to ensure compliance with fundamental rights, non-discrimination, quality and security, transparency, and controllability.

In the latter regard, international practice already shows the wide range of possibilities in how societies might act: China has established an AI Internet court presided over by an AI judge for cases involving legal disputes in the digital domain;⁴⁰ Estonia has launched a project to build a robot judge to preside over small claims disputes involving sums of less than € 7,000;⁴¹ the United States allows a limited yet still controversial⁴² use of AI in informing legal decisions concerning whether to incarcerate defendants pending trial; but France, on the other hand, has banned the use of AI in legal proceedings.⁴³ This effectively spans the waterfront of possible positions on AI’s role from full agency, to supporting role, to outright ban.

Healthcare is also witnessing pioneering developments of AI applications, given the availability of enormous amounts of data that greatly exceeds human cognitive capacity to effectively manage,⁴⁴ increases in computational power, and the development of ML techniques to retrieve information from unstructured data as well as in imaging and signal detection tasks. As a result, the healthcare system provides many examples of AI that are already widely deployed in areas such as radiology, oncology, and ophthalmology,⁴⁵ and even general medical decision-making, such as triaging patients in a hospital setting;⁴⁶ and AI-powered chatbot triage services as an alternative to telephone helpline services to dispense healthcare advice and direct patients to local and out-of-hours medical services.⁴⁷

Not all AI products for healthcare face significant regulatory oversight – for example, consumer-facing platforms or assistants that dispense conventional advice (e.g. guiding patients in their preparation for surgery or through the recovery process). Such applications that are already widely distributed involve modern versions of expert systems that are embedded in online products and that are relatively simple in terms of the understanding of terminology, data protection, human involvement, safety, and risk management. The level of trustworthiness can be decided by market competition which fosters industry standards as regards accuracy, robustness of technical capabilities, and other application-specific criteria. Standards can be overwritten by authorities if any concerns arise.

The US Food and Drug Administration (FDA) has taken the lead in developing a regulatory framework for approval of AI/ML medical devices in more critical applications.⁴⁸ It has established three levels of clearance for AI/ML-based medical applications, namely:

• a 510(k) which clears Class I or II devices for market if they can be established to be at least as safe and effective as another similar, legally marketed device;

• pre-market approval for Class III devices that require greater regulatory evaluation of the scientific evidence because of potential risks to health (e.g. pacemakers); and

• a de novo pathway for novel medical devices for which there are no legally marketed counterparts, for which the FDA performs a risk-based assessment to establish safety and effectiveness.

Already we can see the potential for differing conclusions across major regulatory jurisdictions as to what is sufficiently safe and effective to be put on the market, given the scope for differing risk tolerances, including for devices requiring pre-market clearance where there is potential for different regulatory agencies to reach different conclusions; and even more so for de novo devices.

An example of embedded AI that provides a glimpse into the regulatory framework through which it moves is provided in the aviation sector, where aircraft incorporate a myriad of systems that co-share flying operations with human pilots,⁴⁹ performing both mechanical and cognitive functions. In the Boeing 737 Max case, a faulty sensor resulted in incorrect information being fed into an AI system (the automated flight-control system, Maneuvering Characteristics Augmentation System, or MCAS), which resulted in two crashes.⁵⁰ An international panel of experts⁵¹ was formed to review the causes of the breakdowns in Boeing’s internal safety disciplines and the US Federal Aviation Authority’s certification and oversight procedures. The panel made a dozen recommendations,⁵² which established de facto conditions for the re-entry into service of this Boeing aircraft around the world.

E The Locked Versus the Unlocked

Since the first AI/ML device received FDA approval (a wearable-tech monitoring system introduced in 2012),⁵³ some sixty-four AI/ML-based medical devices and algorithms have received FDA approval and been put on the market. While this early experience is encouraging, a still more complex issue has been encountered in this area. The current regulatory approach for medical devices was designed for devices that are “locked” (i.e. devices that give the same answer each time the same inputs are presented) and feature only discrete modifications from time to time. It is now recognized that this needs to be adapted for algorithms that learn with each application.⁵⁴

In this regard, the FDA has put out a discussion paper setting out a proposed regulatory framework for modifications to AI/ML-based Software as a Medical Device (SaMD), which involves a total lifecycle approach to regulation, based on four principles:⁵⁵

• establish clear expectations on quality systems and good ML practices (GMLP);

• conduct pre-market review for those SaMD that require pre-market submission to demonstrate reasonable assurance of safety and effectiveness and establish clear expectations for manufacturers of AI/ML-based SaMD to continually manage patient risks throughout the lifecycle;

• monitor the AI/ML device and incorporate a risk management approach and other guidance in development, validation, and execution of algorithm changes; and

• transparency to users and FDA using post-market real-world performance reporting for maintaining continued assurance of safety and effectiveness.

These principles – in particular the third, which requires a continual programme of monitoring and validation – highlight the issues posed by the inherent fluidity of deployed AI/ML devices and algorithms that are undergoing continuous modification with acquired experience. Coupled with the ubiquitous concerns about bias and data security, this fluidity underscores the need to establish and maintain a high-trust environment between the creators of the AI, the user community, and the regulators. Similar levels of confidence and transparency will be required between national regulatory bodies to ensure international market access. However, as AI will rely heavily on trade secrets to protect the intellectual property in AI applications (e.g. algorithms and data), the issues concerning the quality and biases inherent in the data used to train AI algorithms may prove to become points of friction in international trade.

III Getting Artificial Intelligence to Market: Navigating Societal Choice and Insecurity

While the integration of AI into the trading system has been more or less seamless at the technical level, as it begins to have systemic significance, new hurdles are likely to emerge. Three of these in particular loom large as potential points of friction: societal impacts, national security concerns, and the question of the impact of AI on jobs. We address these next.

A Societal Impacts

The nexus of AI/ML/big data not only impacts at the micro level on individuals and firms but also drives a complex co-evolution of technology, the economy, and society that takes on its own dynamic, as captured in the title of Kevin Kelly’s 1994 book, Out of Control: The New Biology of Machines, Social Systems, and the Economic World. The pace of evolution in machine space is dictated by the resources committed to innovation and thus is almost arbitrarily fast. The technological instinct is indeed to move fast and disrupt; however, with disruptive technological change, the co-evolution of societal structures and of the economy ensures that, along with all that is gained, there is also much that is lost. Moreover, governance systems that evolved in an age of much slower technological change are not well equipped to get out in front of the implications of new technologies. The result is system friction:

The shift of our economy and society online is taking place without referendum. What could go wrong? As it turns out, plenty.⁵⁶

This friction surfaced in the “techlash” that flared in the second half of the 2010s.⁵⁷ There were numerous contributing factors beyond the pace of change. For example, there was widespread apprehension about the potentially dystopian directions of change,⁵⁸ many of which were popularized by the television series, Black Mirror, and even amplified by Elon Musk who said in an interview, “With artificial intelligence we’re summoning the demon”.⁵⁹ The fragility of democracy in silico was underscored by the revelation of manipulation of electorates in historical events such as the Brexit Referendum and the 2016 Trump presidential campaign by firms such as Cambridge Analytica using Facebook data and applying AI-driven quantitative social psychology tools.⁶⁰

Even more fundamentally, the concentration of wealth enabled by the data-driven economy irrevocably altered the balance of power within modern societies. This is underscored by the fact that a company like Facebook has 2.5 billion clients for its applications⁶¹ – more than the populations of the United States, the EU, and China combined. This change in power relations was evidenced in the behaviour of the technology CEOs who did not fail to sense their new status:

By displacing the print and broadcast media in influencing public opinion, technology is becoming the new Fourth Estate. In our system of checks and balances, this makes technology co-equal with the executive, the legislature, and the judiciary. When this new Fourth Estate declines to appear before [the International Grand Committee] – as Silicon Valley executives are currently doing – it is symbolically asserting this aspirational co-equal status. But it is asserting this status and claiming its privileges without the traditions, disciplines, legitimacy or transparency that checked the power of the traditional Fourth Estate.⁶²

These factors combined to generate pushback on the technology companies, their CEOs, and indeed the practical implementation of the technology nexus of AI/ML and big data.

At the national level, the likely source of issues for international trade will be invocation of the precautionary principle to exclude certain uses or technologies altogether based on societal preferences. The international community has some practical experience with this. Generally, under the World Trade Organization (WTO) Agreement, in particular the Technical Barriers to Trade (TBT) Agreement and the Agreement on the Application of Sanitary and Phytosanitary Measures (the “SPS Agreement”), countries have the right to set higher standards than accepted international standards,⁶³ although they are subject to general tests of reasonableness such as avoiding arbitrary or unjustifiable distinctions in risk tolerance across different situations (including, of course, not discriminating against imports compared to domestic products). At the same time, where relevant scientific evidence is insufficient, a WTO member may provisionally apply restrictive measures based on available pertinent information subject to the requirement that a more objective assessment of risk is made within a reasonable period.⁶⁴ While not directly referencing the precautionary principle that is formally incorporated in multilateral environmental agreements such as the Cartagena Protocol on Biosafety, the WTO Agreement thus does allow for precaution in setting rules.⁶⁵

This base of experience, particularly the extensive debate concerning the precautionary principle,⁶⁶ helps prepare us for the challenges of carving out legitimate policy-based derogations for trade in AI from the freedom of commerce that international economic law defends.

A likely more challenging aspect of the pushback is at the sub-national level. A quintessential example of this, given the breadth of issues raised, was the communitarian response to the ambitious, futuristic smart city proposal for the Toronto waterfront Quayside district put forward by Sidewalk Labs, a subsidiary of Alphabet/Google, which aimed to essentially “disrupt the neighbourhood” in multiple dimensions.⁶⁷ This proposal was eventually withdrawn after a concerted battle by community activists.⁶⁸

Governance flashpoints in the Sidewalk Toronto case included the proposal to claim a share of property taxes (essentially privatizing municipal governance); privacy concerns about the capture of the enormous flow of data that the district would generate through ubiquitous sensors (concerns which led to the resignation of the privacy adviser, Ann Cavoukian);⁶⁹ and more general governance concerns given that the administration of the smart city district would involve a private firm replacing regulations established through democratically accountable processes with its own frameworks⁷⁰ and digital incentives (e.g. one element of the plan was to grant residents access to certain spaces based on how much data they provide, or rewarding them for “good behaviour”⁷¹).

Another set of objections focused on the financial aspects of the proposal, starting with the inside track that Alphabet/Google appeared to have had for the project,⁷² which evoked the sense of overweening influence wielded by “big tech”; the vast asymmetry in information between the Canadian government bodies negotiating the deal and Sidewalk Labs, in particular concerning the ownership and ultimate monetization of the intellectual property and data that the smart city would generate;⁷³ and the economic power that the administering company, a multinational digital “superstar” firm, would have had over the district, which raised the omnipresent sceptre of market failure to which the data-driven economy is inherently susceptible.⁷⁴

The Sidewalk Toronto example highlights the likely role of cities and communitarian activism in mediating social acceptance of AI. We have already seen communitarian activism drive policy on single-use plastics and Styrofoam products, with some US states and cities banning their use; and, highlighting the frictions, we have also seen some states imposing pre-emptive laws to prevent their cities from banning such products.⁷⁵ The use of AI for facial recognition has similarly met with divergent policies, with embrace in some states and bans in others⁷⁶ – and even international sanctions for alleged human rights abuses.⁷⁷ Reflecting the reading of public opinion, Microsoft, Amazon, and IBM publicly committed not to sell facial recognition to police departments because of human rights concerns over surveillance and racial profiling in the context of Black Lives Matters protests, until there is federal legislation that regulates its use and takes into account human rights issues.⁷⁸

The scope for sub-national variance of treatment is also illustrated by regulations being developed for autonomous vehicles. Husch and Teigen highlight the many differences in the rules frameworks that have been adopted in the United States, where regulation of autonomous vehicles falls to the states.⁷⁹ Since 2012, there has been inconsistent acceptance, with some forty states having enacted legislation related to autonomous vehicles, implemented an executive order, or both.⁸⁰

With urbanization growing steadily and expected to raise the share of the world’s population living in cities from over 55 per cent in 2020 to 68 per cent by 2050,⁸¹ cities will gain increasing clout in governance and will be looking for technological solutions to the infrastructure and administrative challenges posed by newly highlighted pandemic risks, environmental sustainability imperatives, and income inequality. They will thus be both the demandeurs for AI technology and the battlegrounds for its acceptance.

B National Security

The digital transformation, the advent of the data-driven economy, and particularly the coming implementation of fifth-generation telecommunications networks (5G) and IoT applications, which 5G will power, combine to fundamentally transform the concept of national security. This reflects in the first instance the proliferation of vulnerabilities to cyber attacks, whether from state actors, from criminal elements (e.g. ransomware attacks on cities and public institutions), or even from university students gaming the system (e.g. the infamous Mirai bot event that crippled the Internet in 2016 was initially thought to be the work of a state actor before being traced to US college students).⁸² As 5G and growing AI applications transform the backbone infrastructure of an economy (i.e. transportation, telecommunications, energy, and finance) from a passive utility into an interactive “central nervous system”,⁸³ national security principles have to be updated quite fundamentally.

Importantly from a trade perspective, these vulnerabilities are fundamentally different from those that informed the crafting of the current WTO national security exception as set out in the General Agreement on Tariffs and Trade (GATT) Article XXI. The framers of the GATT had World War II and the use of nuclear bombs in mind when providing examples of issues that might reasonably trigger the Article – circumstances that relate to fissionable materials (that is, nuclear weapons), traffic in arms, or measures taken in time of war or other emergencies in international relations.

By contrast, cyber attacks are high-frequency and relatively low-cost events, mostly carried out by bots with limited attributability to anyone, including to state actors. Security firm F-Secure, which deploys decoy servers to attract such attacks (so-called honeypots), recorded 5.7 billion attacks in 2019, up from 1.0 billion in 2018.⁸⁴ Sacramento-based Sutter Health reported 87 billion cyberthreats encountered in 2018.⁸⁵

The cyber context resembles that of a biological immune system in a biosphere full of viruses, mostly fighting them off, but sometimes catching a cold – unpleasant but with consequences that fall far short of those associated with kinetic war (let alone nuclear war). The first suspected death attributable to a nonstate cyber attack occurred in 2020 in Duesseldorf, when a ransomware attack on a university hospital forced redirection of emergency cases elsewhere, delaying critical care.⁸⁶ The financial costs of such attacks are estimated in the millions of dollars but the overall cost at the economy level for the United States in 2019 amounted to only perhaps USD 7.5 billion or 0.036 per cent of US GDP.⁸⁷

To be sure, the costs of disruption of infrastructure by state actors could be substantially higher – for example, a “kill switch” on an electrical grid being triggered. This possibility appears to have been established by infiltrations by governments of rivals’ systems.⁸⁸ However, given the multiple sources of risks (including human, software, and hardware), it is far from clear that these concerns (or related concerns of cyber espionage) warrant extreme measures that preclude trade, such as the US’ “5G Clean Path” programme that aims to freeze Chinese telecommunications equipment suppliers out of 5G systems outside of China.⁸⁹

The WTO has little experience in dealing with national security issues as an exception.⁹⁰ One reason is that “trade restrictions during the Cold War period mainly related to non-Members, and there was no great need for justification under GATT”.⁹¹ Another is that countries were reluctant to set precedents that might be used against them, and thus figuratively opening Pandora’s box.⁹²

Not surprisingly, the framing of national security exemptions in trade agreements is evolving. For example, the recent update of the North American Free Trade Agreement (NAFTA) – the US-Canada-Mexico Agreement (USMCA) – included a GATT Article XXI-type exception but dropped the examples. Unfortunately, it provided no alternative language, leaving fully open the question of what kinds of national security risks in this digital age would support an abrogation of trade commitments. This gap is especially problematic given the evolution of the global system of production and trade into a “made in the world” system of global value chains.

Decoupling and repatriation of international supply chains comprise one possible solution to national security concerns, but this would come at some considerable economic efficiency cost, would not actually remove the vulnerabilities from the IoT framework, and would in any event not be a realistic option for any economy other than perhaps the United States, the EU, or China.

AI finds itself in the eye of this particular storm. It is central to the national security frameworks of the major powers. As a practical example, China has indicated it would block the transfer of the AI algorithm underpinning the ByteDance TikTok operation.⁹³ The problem in this instance is not that the AI cannot get into a market, but rather that it cannot leave a market. This risk will hang over other companies – will Tesla, for example, be allowed to transfer its Chinese-developed technology to the USA if US bans on transfer of US technology to China continue? At the same time, control of AI that is in a position of influence over popular opinion in a country clearly will not be allowed for companies from countries that are considered strategic competitors.

Accordingly, national security could be a conversation killer for AI when market access comes up in an international trade context.

C Labour Markets and the New “Guilded Age”

AI can be thought of as a new form of productive capital – machine knowledge capital. As such, it is likely to complement human skills in some tasks and compete with them in others. If we think of “jobs” as packages of “tasks”,⁹⁴ automation of tasks results in partial automation of all jobs. Consistent with the experience of skill-biased technological change over the past several decades,⁹⁵ income inequality is likely to increase as workers whose skills are mainly complemented by AI will realize rising returns to their human capital, while those whose skills are mostly substituted by AI will face job loss or strong downward competitive pressure on wages.⁹⁶

Various scenarios have been suggested for the impact of AI on labour markets. Pessimistic scenarios⁹⁷ conclude there will be heavy job destruction. Less pessimistic scenarios⁹⁸ conclude that automation will mainly transform jobs rather than destroy them, but that low-qualified workers will likely bear the brunt of the adjustment costs since a greater proportion of their tasks can be automated compared to highly qualified workers. The main challenges in this scenario are facilitating job/task transition with training and addressing income inequality. Meanwhile, technology optimists conclude that AI will create jobs.⁹⁹

Regardless of which scenario ultimately obtains, it seems clear that a new factor of production will claim its share of national income – and since this new factor primarily competes with human brain work, it follows that this share of income will be clawed away from today’s white-collar work force. The current organization of society and economy in advanced countries in terms of status and income is based on human capital. People invest heavily to acquire both the knowledge capital and the credentials. Even though student debt is often crippling, the overall returns to a university degree are still very substantial: an estimate of the net present value of a university degree in the United States in 2018 was, on average, USD 344,000.¹⁰⁰ At the same time, at a price point where the annual cost of college equals USD 50,000, the odds of the investment in a college degree paying off fall to about 50–50.¹⁰¹

What happens in this context when the rents to higher education are eroded – that is, when the incomes that drive the net present value of a degree fall? The answer is, of course, structural adjustment along many margins – demand for higher education falls, prices fall, and the supply of this service contracts. Universities and colleges are pillars of their local economies. So these college towns would suffer as well from the multiplier effects. In this regard, the AI shock to white-collar work and the social organization around it in the advanced economies would resemble the China shock to industrial work and the social organization around it in the advanced countries in the first decades of the twenty-first century¹⁰² – except that the AI shock will likely be larger and likely come faster.

The political ramifications of this in the advanced countries can only be guessed at; however, the best guide perhaps is what happened with the China shock to industrial jobs and incomes – protectionism of all sorts. AI should expect a similar welcome as it starts to make serious inroads into the rents currently captured by white-collar work and to undermine the social edifice built on those rents.

In pre-industrial times, the protection of rents flowing to skilled artisans was through craft guilds. In their day, these acted as professional associations, restricting entry to capture rents, but also enforcing quality standards, preserving and transferring knowledge inter-generationally through the apprenticeship system, and providing financial support for their members.¹⁰³ Modern professions such as law, medicine, accounting, and architecture replicate guild practices by requiring a licence, passing a qualifying exam, or acquiring a diploma from a formal programme of study.¹⁰⁴ The modern guilds have been able to resist international services trade liberalization and may be expected to mobilize to moderate the entry of AI into their functions to protect the rents that flow to knowledge credentials. From this perspective, the age of AI – at least in its early years and decades – may be a new “guilded age” in which the professions find ways (which trade economists would see as non-tariff barriers) to restrict market entry.

IV Discussion and Conclusions

AI has made impressive inroads into our economy and society, but this was far from an overnight success, as it struggled through many decades and several AI winters, disappointing many hopes and prognostications along the way. With the emergence of the data-driven economy, the technological conditions for AI to blossom were finally in place – and blossomed it has. AI is now all around and contributes importantly to the value of internationally traded goods and services.

For the most part, AI has navigated the regulatory path to market entry without problems. However, as AI has become more powerful, high-level concerns have started to mount about its impact on society, national security, and the livelihoods of those who will compete with it. Based on the experience to date, regulatory concerns that could create market barriers to AI in the future are likely to align with the Pareto principle (the “80–20” rule), whereby most of the issues will prove to be easily handled at least at the technical level, allowing the integration of AI into economic and social life to proceed apace, while a smaller subset of cases that generate cross-cutting societal impacts and raise security and economic distributional concerns will generate most of the headaches.

The transition of AI from executing instructions to exercising agency, which raises thorny issues for legal doctrines,¹⁰⁵ still lies largely ahead and raises rather open-ended questions about social acceptance, alongside the already thorny issues raised by its use as a tool for political influence and social regulation. Also mostly ahead are the impacts of AI on the job market – in particular on white-collar work and the social structures built around human capital in the advanced economies (although blue-collar work will not be entirely spared either, as AI combined with robots will make the latter more flexible and more competitive with blue-collar workers).

A complicating factor (as if the above were not complicated enough!) is that AI is being developed at a pace that exceeds the ability of regulators to regulate it. This has stalled deployment of AI in domestic contexts (e.g. several major US firms have declined to supply AI for facial recognition until federal regulations are established) and promises to be still more problematic internationally, given that trust is at a nadir internationally – particularly between China and the United States, the two leading AI/ML centres. While this state of affairs seems unpromising for future collaboration, it might be noted that professional exchanges between the Chinese and US epidemiological communities during the COVID-19 crisis were as cordial and forthcoming as the political relations were not. Science transcends national boundaries and with AI/ML we will be dealing with truly cutting-edge science. Moreover, the issues of trust between humans might become rather moot when AI clearly surpasses individual human expertise. The path for AI into practice has generally been cleared by simple demonstrations of its capacity to do better.

The potential difficulty of untangling these issues is well illustrated by the US ban on the TikTok app based on its ownership by China’s ByteDance. This case has triggered commentaries focused on the societal risks of the app itself,¹⁰⁶ the alleged national security risks posed by the data it collects,¹⁰⁷ and the value of the company (as much as USD 50 billion¹⁰⁸).

History has been described as one damn thing after another. The first decade of the data-driven economy proved to be one of increasingly dense history, with the year 2020 serving up a perfect storm of historical developments. The technology nexus of AI/ML/big data played a not insignificant role in generating that history and also found itself an increasingly divisive bone of contention. As new applications proliferate, the discussion in this chapter suggests that the path of AI to international markets will become more complicated.