That they all hallucinate? LLMs are not the borders of AI potential. SNET folks know this which is the primary justification for the architecture they are building.
They are pretty useful for basic research. Like google on steroids. Perplexity AI sites it’s sourcing too which is helpful for confirming your research
LLMs are nearly useless for research due to hallucinations and bias in output filtering
Обсуждают сегодня