LLM optimizer tuning is planning, argues rebuttal to 'planning is overrated'
HanchungLee · x · 2026-10-19
Grant Slatton argues planning is overrated — just take the shortest-term action pointing roughly at your goal, since reality is a super high-dimensional space. Hanchung Lee quotes and counters: planning is underrated; LLM optimizer innovations are precisely about finding the right step size/learning rate, how to taper it, and determining direction — "that's planning."
More from AGI Musings
- Ben Goertzel: which minds have beneficial 'basic AI drives'? Only non-maximizers — bengoertzel · 2026-10-19
- If AI records every conversation by default, society changes dramatically — generativist · 2026-10-19
- Davis Patterson: AI and Robots Will Cut Home Building Cost and Time 10-100x — davidpattersonx · 2026-10-19
- Indie dev of 17 years: AI cuts game iteration from days to minutes — yunta_tsai · 2026-10-19
- Professor caught hallucinating citations as post claims humans now hallucinate more than AI — ColinWPLewis · 2026-10-19
- MIT's AI Decision Matrix: allocating human vs AI decisions by ambiguity and risk — mcraddock · 2026-10-19