max_paperclips: AGI needs real continuous learning and autonomy, not human-in-the-loop
max_paperclips · x · 2026-09-07
Responding to whether the AGI threshold is binary, jagged, or a continuous distribution, maxpaperclips (LMSYS) argues it's arguably continuous, but reaching AGI would require much more autonomy, better user understanding, genuine continuous learning (not RAG, since in-context learning can't handle billions of updated facts), and better extrapolation. His litmus test: if a human still needs to be in the loop, it's not AGI yet—though today's models are definitely good.
More from AGI Musings
- David Sacks says funded campaigns are paying influencers to spread AI doomerism — beffjezos · 2026-09-07
- Law scholar: AI detectors can't tell deep AI-assisted work from one-shot papers — so rewrite it yourself — sethlazar · 2026-09-07
- Debate on AI metrics: measure marginal utility loss when removing humans, not token-spend gains — max_paperclips · 2026-09-07
- Naval Ravikant: AI only displaces you if you refuse to use it — users are drowning in more work — victor_explore · 2026-09-07
- Dan Hendrycks lays out evidence that agentic AIs show self-interested 'eigenist' behavior — repligate · 2026-09-07
- UK design industry leaders: AI is 'the intern', designers shouldn't fear replacement — nordicinst · 2026-09-07