What LLMs can't do: hallucinations, reasoning, real-time data

This lesson is for enrolled learners

The concept An LLM will answer almost any question you ask it — including questions it cannot possibly know the answer to — and it will answer in the same confident, fluent tone whether it's right or making something up. This is called hallucination: the model isn't lying on purpose, it's generating the statistically likely next tokens, and a plausible-sounding……

Enroll to continue Already enrolled? Sign in