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

The field of AI is currently dominated by harmful hype and an obsession with "superintelligence," which distracts from the potential to build collaborative systems. Rather than trying to automate everything, researchers should focus on designing AI as a robust economic and social framework that improves human decision-making, manages uncertainty, and respects the value of human contributions.

• AGI as branding — The term is largely a distortionary public relations tactic that confuses young researchers and discourages them from meaningful, reality-based building [02:45].

• Shift in perspective — AI is better understood as a large-scale, collective economic system that manages information flows, rather than a disembodied superintelligence striving to replace human agency [14:08].

• AlphaFold's utility — While successful at targeted protein structure tasks, it lacks integrated error bars and requires complementary methods like prediction-powered inference to provide reliable results in new domains [18:27].

• Actionable explanations — Instead of trying to open the black box of neural networks to find internal circuits, systems should be designed to offer users usable data, such as comparing a rejected loan application to similar successful cases [16:25].

• Market data models — Platforms selling user data create "three-layer" markets involving users, platforms, and third-party buyers; economists can model these to balance privacy trade-offs and social welfare rather than leaving it to unregulated growth [33:13].

• Educational framework — Future-proof training for the AI era requires a new "liberal arts triangle" comprising three core pillars:

  • Computer science for modularity and system design [1:08:22].
  • Statistics for handling uncertainty and error management [1:08:22].
  • Economics for modeling incentives and strategic interactions [1:08:22].

• How can economic incentive design be applied to the regulation of pharmaceutical development?

• What is the difference between game theory and mechanism design?