Why would open AI hallucinate when it scouts information from the same web search or is it because the web itself has incorrect facts, knowledge. In such a case where does one really verify facts and figures or quotes.
Thank you for your support.
AI hallucinates because it is not “looking up facts” the way a search engine does. It is generating the most likely words based on patterns it learned during training.
So even if it sounds confident, it can still be wrong because it is predicting text, not verifying truth.
Temperature (an internal setting of LLM’s) just changes how random the answers are:
- Low temperature → more consistent answers
- High temperature → more varied answers
But neither guarantees correctness.
In short:
AI = pattern-based text generation, not a fact database, unless you use an AGENT within the LLM process.
That is increasingly difficult. You still have to do your own research.