2 ms·
It doesn't really seem that the situation would be improving or could be even solved: "While hallucinations are reduced relative to general-purpose chatbots (G
by jruohonen 2y ago
It doesn't really seem that the situation would be improving or could be even solved:
"While hallucinations are reduced relative to general-purpose chatbots (GPT-4), we find that the AI research tools made by LexisNexis (Lexis+ AI) and Thomson Reuters (Westlaw AI-Assisted Research and Ask Practical Law AI) each hallucinate between 17% and 33% of the time."
Ref.:
https://news.ycombinator.com/item?id=43814663 https://news.ycombinator.com/item?id=43814663
- techpineapple 2y agoIs there indexable database of references you could compare hallucinations against? How does perplexity provide reliable references
- jruohonen 2y agoYes: there are databases. However, as also elaborated in the paper, the issue is not merely about providing wrong (including nonexistent) references but also about factual correctness of the references or not giving references when needed. I don't know whether these and other aspects of Perplexity's (or CoPilot's) references have been evaluated.