LLMs vs Search Engines

LLMs are a complement to search, not a replacement.

No LLMs are not 'search'. Search as in google or a database query is deterministic. Are there results for x query? If there are we can return them. If there aren't we can't return anything.LLMs do not work that way. LLMs do not have a conception of facts. Any query you make to an LLM has an output. The quality of that output depends on the training data. For high probability output you might think the LLM is returning the correct 'facts'. For low probability outp

In the current LLM tech if you need sources reliably, you need to start with search first and then summarise that with LLMs.ChatGPT without web browsing skips the search part and ends up hallucinating URLs

Any current question to an LLM is just a textual interpretation of the search results though; the use the same source of truth (or lies in many cases)

Pretty much. Its just that most people are too lazy to look through results and synthesize information themselves. Also it would be helpful if the LLM could identify outdated or conflicting information, but I don't know if that is within their capabilities.

Because search engines give you results you can look through. LLMs just make up garbage and pretend its true.

Yes, please. Do not use LLMs as a substitute for Google search. If you are looking for factual information, just use google, bing, or duckduckgo. You should only use ChatGPT for things that you are able to review it's work.Technology is supposed to make us smarter. Blindly believing an AI that we know can hallucinate makes us dumb with confidence.

That's why you are better served asking LLMs for factual information when they are connected to tools like search. ChatGPT with search enabled, Bing.chat or https://www.phind.com/ will give you sourced and much more reliable output.

I have repeatedly tried to use LLMs as search engines, both general (like ChatGPT) or more focused on specific domains.I have not been impressed by the results. In my experience, LLMs used this way generally output confident-sounding information but have one of two problems the majority of the time:- The information is blatantly wrong, from a source that doesn't exist.- The information is subtly wrong, and generated a predictive chain that doesn't exist from part of a source.<

I think the trick is not so much which model you use, but rather how you ask the questions. Make it very clear you want sources front and center, and use the LLM to try to cut through all the SEO and find the information you are looking for. Don't rely so much on what the model tells you, but rather where it leads you.