 | , - |
 | Dauphine, Salle Raymond Aron (2e étage) |
Cette conférence est organisée en partenariat avec l'Institut Gaia-X, avec le soutien de la Caisse des Dépôts.
Événement en anglais.
Abstract
The development of vertical artificial intelligences —that is, AI engines specialized by value chains or use cases—is a crucial challenge for all value chains around the world. The integration of artificial intelligence into various products, including autonomous vehicles and medical equipment, as well as along value chains, from the aeronautical to tourism industries, is poised to catalyze significant productivity gains, enhanced capabilities and innovation. This phenomenon is likely to contribute to the overall socio-economic growth. Also, the ability to develop and implement vertical artificial intelligence will be pivotal in reshaping the competitive positions of firms and national economies. Inter alia, the reindustrialization of Europe will be depending upon the capacity to reconfigure value chains through the implementation of such technologies. More generally, vertical artificial intelligence constitutes a pivotal element in the prevailing competitive dynamic between the United States, China, and Europe. It is therefore imperative to cultivate a comprehensive understanding of the economic principles and technological facets of vertical AI.
Vertical AI is predicated on a distinct technological foundation from general AI, the latter of which is popularized by LLMs and generative AI. Its approach is less agnostic because its specialized purpose allows its developers to combine scientific and technological knowledge with machine learning, which in turn leads to smaller models. Furthermore, it is trained on industrial data generated by the captors embedded in products and processes, as well as by the information systems of organizations delivering products and services to professional and non-professional users alike. Accessing these high-quality data sources necessitates the cooperation of the corporations or governmental bodies that generate them. These entities are also the primary potential users of these vertical AI engines. The combination of these two technological characteristics results in a contrast between the economics of vertical AI and of general AI.
The development of general/generic AI has been predominantly propelled by the emergence of prominent entities capable of mobilizing substantial financial investments to access voluminous data sets and massive computing capabilities to train models. Operating the resulting very large models subsequently demand significant resources and are characterized by substantial economies of scale. Consequently, this has engendered a winner-takes-all dynamic, as the return on investments necessitates the provision of services to a substantial user base.
In contrast, vertical AI demands a symbiotic relationship between users and developers, a balanced consideration of economies of scale and specialization requests (fit for purpose), resulting into AI engines characterized by a much lower capital intensity. These characteristics may potentially lead to a more fragmented market, characterized by a differentiation-based competitive environment, and a coopetition between AI engine designers and the entities and communities generating data and implementing AI based solutions in their operations and products and services.
The conference will convene a group of experts from academia, government, and industry to deliberate on the technological, strategic, and politico-economic challenges associated with the development and implementation of vertical artificial intelligence.
Programme et intervenants
Morning | The economics and technology of Vertical AI
- Jakob Rehof – Lamar Institute Dortmund
- Michael Kamp – Lamar Institute Dortmund
- Eric Brousseau & Joëlle Toledano – Governance & Regulation Chair
- TBC – AI Industry | Keynote Address
- Laurent Lafaye – Dawex
- Art Langer – Columbia University
- Additional speakers to be announced
Afternoon | European Challenges for vertical AI
- Yvo Volman – European Commission, DG CONNECT (TBC)
- Bert Verdonck – Luxembourg National Data Service
- Jean-Pascal Riss – Schneider Electric
- Michael Ziegelmeir – STACKIT
- Jean-Baptiste Burtscher – Valeo
- Boris Otto – Fraunhofer Institute
- Bruno Sportisse – Inria
- Additional speakers to be announced
Conclusion
- Ernst Stöckl-Pukall – Digitisation and Industrie 4.0 Unit, German Federal Ministry for Economic Affairs and Energy (TBC)
- Thomas Courbe – Direction Générale des Entreprises (DGE) (TBC)
