AI, Philosophy, and Governance Take the Stage at WAIC

量子位 · wechat · 2026-07-19

This article covers a WAIC forum on "Mind and Intelligence," bringing philosophy, scientific intelligence, social simulation, research credibility, and industrial application to the same table. The core issues shifted beyond model parameters and leaderboards to whether AI truly "understands," if it can act as a scientist, and how it should be validated and governed in the real world.

The article highlights various guest perspectives: some emphasized that "the world exists before intelligence," drawing on symbolic, physical, and "body-environment-others-history" foundations. Others showcased self-evolving research models where AI independently generates ideas, writes code, runs experiments, accumulates experience, and even produces peer-reviewed papers. Discussions also covered mathematical proofs, protein dynamics, social simulations, and trustworthiness in humanities, noting that while models can mimic real society in macro statistics, they show significant biases in causality and specific demographics.

The latter half shifted to governance and industry: a bluebook introduced frameworks like STRIDES, stressing that human-in-the-loop must have actual intervention, correction, and explanation rights rather than just acting as a rubber stamp. In biomanufacturing, lab automation, and synthetic biology, AI must integrate with equipment, workflows, approvals, and production loops. The real challenge lies in the "last mile" and the distribution of benefits and responsibilities. The article concludes that as AI gets better at answering, humans must redecide what to ask, what to believe, and how to bring technology into reality.

Original post →

More from AGI Musings

AGI Musings channel →