Scientists generate hypotheses (analogous to LLM hallucinations) and test them against real-world image data; when scientific patterns match observed patterns they're verified as facts, so integrating image-processing AI with language models would allow AI to calibrate verbal output against image data and eliminate hallucinations.

causalpending

Speaker

Jordan Peterson

Evidence Quote

“they generate verbal hypotheses...equivalent in some ways to the hallucinations...test that pattern against real world images...if the pattern matches...verified and...approximating a fact”

Source

ChatGPT: The Dawn of Artificial Super-Intelligence | Brian Roemmele | EP 357— Jordan B Peterson
Created: 8/10/2026, 11:12:21 PM

Your notes

Loading notes...