Hallucinations in large language models—where they generate false information—are not evidence of lack of understanding, because human memory works the same way: we reconstruct plausible memories from learned patterns rather than retrieving stored facts, and these reconstructions are often false while still conveying the gist of what happened.
causalpending
Speaker
Geoffrey HintonEvidence Quote
“another argument people use is that hallucinations show these things don't really understand what they're saying um they occasionally just make up stuff that's not true well actually that's what people do all the time...when you remember something you don't get it out of some file storage somewhere you just make up something that sounds plausible given the context”
Source
Geoff Hinton - Will Digital Intelligence Replace Biological Intelligence? | Vector's Remarkable 2024— Vector InstituteCreated: 8/11/2026, 7:34:33 AM
Your notes
Loading notes...