David Manheim: all current systems fail AGI definitions; edge cases are what matter

davidmanheim · x · 2026-09-16

AI safety researcher David Manheim asserts that every system existing today fails any reasonable AGI definition, but he expects to confidently identify AGI at some future point — the hard part being edge cases, which are precisely what matters. The thread debates measurability of AGI via the famous Potter Stewart obscenity jurisprudence analogy, with Manheim warning against conflating 'enforceable' with 'easily enforceable'.

Related event: Manheim: Define Before Measuring — AGI Debate Will Resolve Itself Within Years(7 posts)→

Original post →

More from AGI Musings

AGI Musings channel →