Cambridge AI researcher: 'understand' and 'think' are hazy words, stop demanding hard definitions
dioscuri · x · 2026-09-30
Cambridge AI researcher dioscuri weighs in on the 'do models understand/think' debate: part of the problem is that words like 'understand' and 'think' are extremely hazy and shouldn't be expected to map onto well-defined processes.
He draws an analogy to sports punditry: calling a player 'gritty' or 'tenacious' is somewhat informative but doesn't latch onto discrete measurable quantities — what units measure grit? Pushing those questions too far misunderstands their purpose: these words set rough expectations about behavior, not physical quantities like mass or charge.
More from AGI Musings
- Kevin Xu on Chinese open source: how China stumbled into open weights, from Linux floppies to DeepSeek — kevinsxu · 2026-09-30
- evilsocket: Non-Transformer 'Old School' Models Are Massively Under-Explored — evilsocket · 2026-09-30
- Allie Miller: Ultrafast AI Will Turn Meetings From Discussion Into Instant Decisions — alliekmiller · 2026-09-30
- From deer videos to a 4D time machine: the 'spatial memory palace' for physical reality — bilawalsidhu · 2026-09-30
- AI won't take jobs, they'll just go from fake to extremely fake, researcher quips — kevinnbass · 2026-09-30
- "Stop confusing obedience with intelligence": a pithy take on AI — PierceLilholt · 2026-09-30