> First it needs an incentive to lie ... or the model architecture allows it Given that one the largest problems with current LLMs is they will often "hallucin

> First it needs an incentive to lie ... or the model architecture allows it Given that one the largest problems with current LLMs is they will often "hallucinate" (i.e. lie or provide deceptive answers), it seems strange to phrase it this way.