[sometdog]

That is true, but you can combine LLMs with other tools to get sourcing and more accurate answers overall. Instead of using the LLM to directly answer a question, you can use the question to search for relevant text in a particular knowledgebase, and then use the LLM to summarize those results.

[DebtDeflation]

>use the question to search for relevant text in a particular knowledgebase, and then use the LLM to summarize those results

As a consultant, based on personal experience, I can say that what you wrote above constitutes >90% of current enterprise use cases for ChatGPT. What clients REALLY want to do is be able to take a pre-trained LLM and then train it further on their own corpus of documents, but given limitations around token window size, the above is probably the best way to fake it for now.

[gibsonf1]

If you use the LLM to summarize the results, it makes things up by design. As soon as you introduce the LLM into search, you lose.

[raincole]

I've personally replaced Google with Bing Chat for technical things (like searching for specific API). Does it make things up? Maybe. But in my experience it never happes even once in past whole week (>100 times of searchs).

It's not "it happened but you just didn't notice..." if it uses a function call wrong I'd have noticed. My code won't compile. My test won't pass.

So far it either gives me 100% correct result, or completely fails. But it doesn't generate "seemingly correct but actually wrong" things even once, unlike ChatGPT.

[autonomousErwin]

I find this the killer application for ChatGPT (at least for now). Answers you can very quickly verify and care little about the sources because a significant no. of answers on Stack Overflow make ChatGPT look modest in confidence comparison

[throwuwu]

Biased much?