Replying to Avatar Jacob | Five Eye Tea

I'm a latecomer to the AI game but so far, I've found Grok and Perplexity to be the most useful overall. Grok's LLM is advancing exceptionally fast, but Perplexity's formatting is a lot better to my eyes and the way it sources its results is so much more useful.

At the moment, Grok is probably my most-used AI, but I use Copilot for general queries (mostly tech-related), Perplexity for anything I need deeper research on, and then I fall back on ChatGPT for anything miscellaneous. That said, I use Brave's Leo for quick page summaries.

I also use Venice in the same way I use ChatGPT, but for queries that are slightly less "public friendly". I'm also aiming to test out Proton's Lumo in the near future but when I first tried it, it seemed like it had a little way to go before.

Eventually, once my home lab is set up the way I want it to be, I also plan to experiment with self-hosting my own LLMs because that'd definitely be ideal for privacy purposes, but I'm comfortable using the online options for now for generic queries that aren't really private. #ai #artificialintelligence
Avatar
Kayne 5mo ago

All this instead of just looking things up and reading it for yourself?

Reply to this note

Please Login to reply.

Discussion

Avatar
Neigsendoig Cocules 5mo ago

AI is useful for strengthening knowledge on what one already knows.

It isn't useful, however, for learning new things.

Avatar
Jacob | Five Eye Tea 5mo ago

I still do that as well. My first resort is usually to search with Brave Search. I have and use search engine skills on the regular. However, there's no denying that using AI for more complex research is a lot more efficient and useful than scrolling through thirty Reddit threads, processes of troubleshooting on Tom's Hardware/Stack Overflow, and so on. Saves a lot of time if you can just run it in an AI.

I think there's still a use case for classic search engines (and internet research skills), as well as a use case for AI. I don't rely on it for everything. I also don't use AI for anything remotely private or sensitive, either.