> Sure ChatGPT can hallucinate but Google's results rely on some random person somewhere to have properly answered something.
i don't know about your experience with chatgpt, but to me it's been untrustworthy. so much so that every answer i get needs to be double checked against google and that "random person".
I think Reddit has a golden opportunity here, considering how often I prefix searches with "site:reddit.com" to filter out SEO bullshit and usually get an answer from a subject-matter expert in the top ten results. They could even tie an LLM into it and regurgitate the highest-voted comment, and it would likely be more effective than Google or GPT.
I've been saying it for a while, you could basically replace entirely google with a 10k websites index and get above their quality of results 90% of the time.
That's fair I can definitely agree that but my point was more that neither is a very trustworthy source anyway. It's the same way I'd double check one source versus another. It's just easier to start with GPT at this point because at least I know it'll just answer based on what it has been trained with versus special interests. One example I've used is the question "Is Fix A Flat ok to use in tires?" the results are from the company that makes fix a flat saying basically "Yes obviously" while you then see tire companies and repair shops giving you mixed responses. Then you have the third group being just people who have heard from one of the first 2 groups. People are worried about AI degrading the quality of information over time, learning from itself. We're basically there thanks to people caring more about being right than being correct and never doing their own leg work.
i don't know about your experience with chatgpt, but to me it's been untrustworthy. so much so that every answer i get needs to be double checked against google and that "random person".