causal

An LLM in 1633 would have rejected Galileo

A large language model in 1633 trained on all existing scientific text would only parrot back the frequency of statements it had seen—and since geocentric, Earth-stationary statements vastly outnumbered heliocentric ones, plus everyday observation and 'big data' would confirm the Earth feels stationary, the model would have rejected Galileo's heliocentric view because it has no forward-looking mechanism for new ways of seeing.

causalpending

Speaker

Teppo Felin

Evidence Quote

“it can only parrot back what it has most frequently seen in terms of the word structures that it has encountered in the past. And so, it has no forward-looking mechanism of anticipating new data and new ways of seeing things.”

Source

When Prediction Is Not Enough (with Teppo Felin)— EconTalk
Created: 6/13/2026, 12:09:11 AM

Your notes

Loading notes...