Hinton: AI 'Hallucinations' Should Be Called 'Confabulations'
r0ck3t23 · x · 2026-07-06
Geoffrey Hinton argues that erroneous AI outputs should not be called "hallucinations" but rather "confabulations." He notes that the human brain does not store memories like a filing cabinet or computer memory, but reconstructs them on demand. The author comments that the tech industry's view of AI confidently giving wrong answers as a bug to be fixed actually measures intelligence by a "filing cabinet" standard, revealing a misunderstanding of its true nature. Hinton dismantles critics' arguments with a single word—because their brains are also confabulating.
More from AGI Musings
- Can closed frontier models support CBRN defense, or do we need open ones too? — olcan · 2026-07-27
- Steven Pinker says intelligence is bounded, not a limitless scalar like height — SydSteyerhart · 2026-07-27
- A poster imagines AI evolving into synthetic, collective, and beyond-intelligence systems — CurieuxExplorer · 2026-07-27
- AI surpassing human intelligence will be noticeable, as Claude's cryptic output proves — zetalyrae · 2026-07-27
- AI lab staff have gone strangely quiet about next-year capability predictions — ChrisGPT · 2026-07-27
- AI could erode science by flooding research with credible slop — rbhar90 · 2026-07-27