The Economics of AI-Based Technologies: A Framework and an Application to Europe

Abstract

First, this paper discusses how AI-based technologies that are powering the “Fourth Industrial Revolution” are similar in their speed, effects and prerequisites to general purpose technologies (GPTs) that came before and how they are different. AI-based technologies are similar to the steam engine, electric power, and information technology in that (a) they change the nature of work but do not reduce it in aggregate; (b) they increase productivity when accompanied by innovations in business process and investment in human capital; and (c) they leave little room for leapfrogging because profitable adoption of new techniques requires having adapted to previous rounds. AI is different from earlier GPTs mainly in that the speed of its spread is likely to be faster. Second, it provides an economic classification of these technologies based on the channels through which they reduce costs, distinguishing between transactional, informational, and operational technologies. This distinction is relevant because they differ in their structural, social and spatial effects. Third, the paper provides a framework that brings together the economic consequences of AI-based technologies and common policy goals: competitiveness, cohesion and convergence. Finally, it illustrates how this framework can be applied to the European Union. EU member states do least well in transactional platforms where scale matters a lot, they do somewhat better in informational technologies where both market size and regulations matter, and some of them are global leaders in the development and use of AI-based operational technologies such as robots. As with earlier waves of technical change, creation and adoption of AI-based technologies will require good education and well-designed R&D incentives, regulatory capacity, and robust tax-and-transfer systems.