this post was submitted on 13 Nov 2024
171 points (97.8% liked)

Ask Lemmy

26875 readers
2809 users here now

A Fediverse community for open-ended, thought provoking questions

Please don't post about US Politics. If you need to do this, try [email protected]


Rules: (interactive)


1) Be nice and; have funDoxxing, trolling, sealioning, racism, and toxicity are not welcomed in AskLemmy. Remember what your mother said: if you can't say something nice, don't say anything at all. In addition, the site-wide Lemmy.world terms of service also apply here. Please familiarize yourself with them


2) All posts must end with a '?'This is sort of like Jeopardy. Please phrase all post titles in the form of a proper question ending with ?


3) No spamPlease do not flood the community with nonsense. Actual suspected spammers will be banned on site. No astroturfing.


4) NSFW is okay, within reasonJust remember to tag posts with either a content warning or a [NSFW] tag. Overtly sexual posts are not allowed, please direct them to either [email protected] or [email protected]. NSFW comments should be restricted to posts tagged [NSFW].


5) This is not a support community.
It is not a place for 'how do I?', type questions. If you have any questions regarding the site itself or would like to report a community, please direct them to Lemmy.world Support or email [email protected]. For other questions check our partnered communities list, or use the search function.


Reminder: The terms of service apply here too.

Partnered Communities:

Tech Support

No Stupid Questions

You Should Know

Reddit

Jokes

Ask Ouija


Logo design credit goes to: tubbadu


founded 1 year ago
MODERATORS
 

It would be amazing if it doesn't disintegrate if it rains, too.

you are viewing a single comment's thread
view the rest of the comments
[–] Benjaben 2 points 1 day ago* (last edited 1 day ago) (1 children)

Just throwing this out there, but the problem you're describing sounds like a good fit for an LLM I've been using for similar purposes, Claude.

I've found it to be really good at helping me slog through what would be a burdensome and wasteful amount of reading, in order to answer specific questions OR to get a baseline understanding of a thing.

It's a bit hard to know how much value comes from my engineering background and my tendency to "know what I don't know" and thereby ask focused questions, but it's definitely worth a shot. I have found it to be surprisingly sophisticated and much better than slogging through the wasteland of bad search results + too much unrelated but real info.

A topic like this where there's a tremendous amount of legit docs, articles, and forum activity - it's really the exact use case where it's very difficult for a human, and very easy for an LLM to effectively digest that info.

Some caveats I've noticed:

  • it sometimes is overly agreeable / "friendly" when it should be more direct
  • it does sometimes hallucinate or say BS with casual confidence, which sucks because the more you need the info the less well you can spot that. It hasn't hampered usefulness too much for me, but then again I'm usually able to spot the mistakes even in ~unfamiliar subjects
  • they've moved the free tier back to a less capable model at the moment...most of my good experiences are with Claude 3.5 Sonnet, but Claude 3 Haiku (present free tier) is still good

If you're really curious but the volume of reading and documentation to get started is presenting a big barrier, try using Claude to see how quickly you might be able to clear that obstacle. It's been removing those exact barriers for me very effectively lately.

Edit to add: a particularly useful way I can imagine folks in your shoes using this - as a "companion" while you try to follow a guide in an article somewhere. It can answer questions about terms you don't understand, even reasoning behind doing certain steps or what to do if it goes wrong. In fact, you could almost certainly just feed it the written procedure itself (telling it that you're doing so) and really get it to reason about the process with you. Just help get you through whatever implementation.

[–] [email protected] 3 points 1 day ago (1 children)

I genuinely appreciate you taking the time to make this comment, thanks.

I do see how this can be helpful for a lot of things, but I think in my situation (namely a bunch of neurodiversity and brain fog, and no existing interest in or experience with LLMs), having to learn how to deal with an LLM to a point where I get results I'd be confident enough in to trust without having to double or triple check, is probably a bigger hurdle than just diving in to figuring out security camera set ups. It's putting one more thing in the way of the information I actually need, which means my brain is much much less motivated than it already is to get the information, if you see what I mean..

[–] Benjaben 2 points 1 day ago (1 children)

I do understand what you mean, but I think you're probably significantly overestimating the difficulty of using the tool. One of its major strengths is its ability to just understand you, like you'd talk to anyone human, with the benefit that you can even instruct it to use a style you prefer. Just say "I'd like your answer to be terse, let's see if we're on the right track before getting into details". Just as an example.

With all that said you know what you want and need better than anyone else, that's all I've got to say on it, cheers!

[–] [email protected] 2 points 1 day ago

Thanks again 👍