IMD World Digital Competitiveness

coordination, there will be individual country incentives to protect one´s interest and benefit from other coun tries´ goodwill. The European Union, in turn, has decided to regulate AI from the top down through the EU AI Act. 11 The European Commission’s proposal for an AI framework, the first of its kind in the world, was published on 21 April 2021. GDPR (General Data Protection Regulation) was the seed that set the principles under which AI is regulated in Europe –namely, that data belongs not to the government or the private sector, but to individuals themselves. The EU distinguishes between “Limited-Risk” and “High-Risk” AI systems and foresees different degrees of transparency for each. AI policies in the US are established in the Algorithmic Accountability Act of 2022, under which companies are required to assess the impact of their systems on privacy and transparency. But this regulation is, as of today, still not enacted. Any global system to come should build on the fruitful and successful coordination of EU policies that helped make the EU AI Act a reality. It is, of course, difficult to foresee the interest of AI powers such as China and India to agree to regulations that curtail the power of the state in favor of individual rights and privacy. Unlike their approach to general digital infrastructure, governments today rely on the private sector to generate AI projects and solutions. At the end of 2019, for instance, privately held AI companies in the US attracted nearly USD40bn across more than 3,100 transactions 12 (Arnold et al, 2020). In China it is the state who replaces private capital and with similar outcomes. The role of the state is to provide adequate regulation and talent (see the previous section), but sometimes it takes the direct participation of AI companies. The German government, for instance, plans to invest one 4. AI infrastructure investment

billion euros in AI during 2023. 13 Relative to China and the United States, figures in other countries are modest by comparison.

5. Job creation with or without AI

A recent study by McKinsey estimates 14 that in the coming years, and at least in the United States, more people will move toward high-wage jobs–and that fewer workers will be willing to take lower-wage service jobs. This means that routine, low-skilled, and low-paid jobs can easily be done by robots or AI applications. Gener ative AI in particular will take over 30% of the hours currently worked by humans, according to the study. Two questions that the report does not answer is (1) whether our economies can generate enough high-wage jobs for those whose tasks are becoming automated, and (2) whether AI will cause a reduction in the salaries of those who will willingly work less once part of their tasks are taken over by machines. These are particularly important questions for emerging markets. Additionally, we do not know how the intro duction of AI applications to the performance of tasks currently outsourced to countries with cheaper labor will impact the latter countries. For instance, apparel manufactures like Inditex and H&M outsource manu facturing to lower-wage countries like Bangladesh and Turkey. In a new era of smart automation operating at near-zero marginal cost, the damage to employment in such developing economies can be severe. We should also be concerned about how AI could impact the ability of developing economies to compete with “AI-advanced” economies once AI has reduced their cost advantage. There is a risk of the world becoming more fragmented in terms of trade in both goods and services. This will ultimately impact unemployment levels everywhere. Data protection is also essential for developments in artificial intelligence, and our 2023 findings on this topic –via data we gather on cybersecurity–will be presented by my colleagues in the report that precedes this one.

11 See https://www.weforum.org/agenda/2023/06/european-union-ai-act-explained/ 12 Arnold Zachary, Ilya Rahkovsky, and Tina Huang (2020), “Tracking AI Investment: Initial Findings From the Private Markets,” Center for Security and Emerging Technology. 13 https://europepmc.org/backend/ptpmcrender.fcgi?accid=PMC1834265&blobtype=pdf 14 McKinsey 2023, “Generative AI and the Future of Work in America,” available at https://www.mckinsey.com/mgi/our-research/generative ai-and-the-future-of-work-in-america

23