Data on search engine market share is available, but I wonder what that looks like for Lemmy users in particular, who I would assume lean more technical than the average user, so probably use DuckDuckGo and alternates more than Google.
I use a mix of DuckDuckGo and Kagi. I’ll also use ChatGPT, which can be good if you’re careful to verify the answers it gives you as a check against hallucinations. It’s useful for short, direct answers without ads or SEO bullshit.
This article on Ars (and if you’re not a subscriber, you absolutely should be, as they are the best tech journalists out there) inspired the question: https://arstechnica.com/gadgets/2023/06/google-admits-reddit-protests-make-it-harder-to-find-helpful-search-results
Fucking Reddit. Enshittification ruins everything.
I can see a usecase for where you don’t know where to start or search with, and then verify with actual searches.
I recently used it to explain for a friend what is the difference between wheat and ale beer, and it gave a very good summary. With DDG I might not get a direct explanation and would need to read a few articles and then word them in a comprehensive way.
I think we’re making some confusion here.
If you need answer to a question, you can do it by searching for sources, using a search engine, or you can do it by asking a knowledge base (e.g. reading and Encyclopedia).
If that’s the usecase, you can successfully use a generative AI to obtain an answer (providing the fact you are willingfull to accept a possibly bad answer). You are not using the AI as a search engine.
If you need a search engine, if you need to search for places and sources into the web to access it directly, you should not use an AI, because it’s not how it works. It does not search and it does not provide you searching results, it “simply” generate answer based on training, Which is a completely different task.