definition

LLMs give Wikipedia-level knowledge by predicting next word

Large language models work by a stochastic process of sampling from their training data to predict the next word, producing fluent, coherent, non-plagiarized text at roughly Wikipedia-level knowledge, but this is fundamentally a backward-looking exercise with no mechanism for anticipating new data.

definitionpending

Speaker

Teppo Felin

Evidence Quote

“it's very good at predicting the next forward. Right? And so, it's good at that.”

Source

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

Your notes

Loading notes...