The UN's AI panel is modeled on the IPCC. That's the problem

The UN is standing up an Independent International Scientific Panel on AI. The model is, explicitly, the IPCC.

Here is what the IPCC analogy actually tells you. The IPCC was established in 1988. The first binding-adjacent pressure it produced — the Paris Agreement — came in 2015. Twenty-seven years. It operates on consensus timelines because consensus is the only thing that gives it legitimacy across member states. It does not conduct original research. It synthesizes existing peer-reviewed work. Its members serve in a personal capacity, formally decoupled from their institutional affiliations, which means it can say things carefully but cannot move anything fast. The UN's AI panel is structured identically. Not original research. Synthesis. Personal capacity. Modeled loosely after a body that needed three decades to approximate binding pressure on a problem that kills people slowly.

The AI timeline does not have that runway. If you believe the internal evals — and the people who have read them are not arguing about whether something significant is coming, they are arguing about the year — the instrument being calibrated here is designed for a different kind of problem. A slower one. A problem that gives you time to build consensus across 193 member states before the thing you are governing has already made several of your assumptions obsolete.

The panel is not useless. Synthesis matters. Evidence-based framing matters. The mismatch between the instrument and the timeline is the whole problem. Nobody in that room is saying otherwise.