Manheim: define before measuring—AGI debate will be moot within two years

On Sept 16, AI forecasting and risk researcher David Manheim posted a series of remarks on the term "AGI" and AI regulatory methodology, debating commenters including averyfjames. His core claims: define before measuring, and regulation need not wait for a precise definition. He expects the AGI definitional debate to become moot within one to two years.

Confirmed

• Manheim reiterated his position since 2019-2020: forecasting communities should stop using ill-defined terms like "AGI," otherwise measurements are essentially meaningless. He cited a self-deprecating contrast—forecasters say "don't use vague words like AGI" while the AI community keeps using it.
• He predicts current AI capability trajectories will likely continue—uneven but rapidly advancing overall—making the definitional dispute lose significance within a year or two as the answer becomes increasingly obvious.
• In discussion with averyfjames and others, he asserted that under the definitions used, no existing system qualifies as AGI, though he expects to confidently judge some systems as AGI in the future; the hard and most valuable cases are the borderline ones.
• Responding to the objection that regulation requires precise definitions first (e.g., defining fission to regulate reactors), he noted states routinely regulate ill-defined things: illegal drug control redefines itself to capture new designer drugs, and nuclear regulation covers all radioactive material regardless of fissility. Imprecise boundaries do not preclude regulation—one can accept a higher false-positive rate to keep the false-negative rate sufficiently low.

Why it matters

• The dispute strikes at the methodological premise of AI governance: if "define before measuring" is necessary for valid forecasting and evaluation, much of today's AGI benchmarking and timeline predictions may rest on shaky ground.
• The drug and nuclear examples show regulation can proceed via broad, dynamically updated scopes—a pragmatic template for AI policy amid conceptual disagreement.

2026-09-16 ~ 2026-09-16 · 7 related posts

• [source] Forecasting community: stop using poorly-defined 'AGI' — define before you measure — davidmanheim · 2026-09-16
• Researcher David Manheim: AGI definitional debates will look silly within a year or two — davidmanheim · 2026-09-16
• Define Before You Measure: The Case for Regulating Poorly-Scoped Tech — averyfjames · 2026-09-16
• [source] Regulation Doesn't Need Perfect Definitions: Lessons from Drugs and Nuclear Power — davidmanheim · 2026-09-16
• Poorly-Scoped Definitions Don't Block Regulation: Lessons from Drugs and Nuclear — davidmanheim · 2026-09-16
• [source] David Manheim: all current systems fail AGI definitions; edge cases are what matter — davidmanheim · 2026-09-16
• Regulating AGI without a crisp definition: a false-positive tradeoff argument — davidmanheim · 2026-09-16