Intelligence is collective, not artificial — Prof. Michael I. Jordan (UC Berkeley / Inria)

Artificial intelligence research currently suffers from a lack of deep economic and social scientific thinking, prioritizing hype-driven "AGI" terminology over actionable, human-centric engineering. Michael Jordan, a prominent computer scientist and statistician, argues that current large language models are merely statistical prediction engines rather than intelligent entities. Instead of pursuing science-fiction goals like superintelligence, researchers should focus on building robust, multi-agent systems that integrate human values, economic incentives, and uncertainty quantification. By treating AI as part of a broader ecosystem—similar to how markets or supply chains function—developers can create tools that aid human decision-making rather than displacing it. This approach requires shifting from simple gradient-based optimization to designing mechanisms that respect human agency, privacy, and the ephemeral nature of social knowledge, ultimately fostering a more collaborative and productive relationship between humans and technology.