Ask Lemmy
A Fediverse community for open-ended, thought provoking questions
Rules: (interactive)
1) Be nice and; have fun
Doxxing, trolling, sealioning, racism, toxicity and dog-whistling 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 spam
Please do not flood the community with nonsense. Actual suspected spammers will be banned on site. No astroturfing.
4) NSFW is okay, within reason
Just remember to tag posts with either a content warning or a [NSFW] tag. Overtly sexual posts are not allowed, please direct them to either !asklemmyafterdark@lemmy.world or !asklemmynsfw@lemmynsfw.com.
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 info@lemmy.world. For other questions check our partnered communities list, or use the search function.
6) No US Politics.
Please don't post about current US Politics. If you need to do this, try !politicaldiscussion@lemmy.world or !uspolitics@lemmy.world
7) No Hit-and-Run questions.
Please don't delete your post for no apparent reason. If you plan on deleting a question later, say so in the post, or if you feel that you have a good reason to remove it, message a mod beforehand. It's not fair to the ones who took their time to answer, and it's not in the spirit of the community.
8) No Bots.
Posts or comments from bots, LLM's, AIs, Neural Networks, Transformers, or Marvin the Paranoid Android are not welcome in AskLemmy. Real humans only please.
Reminder: The terms of service apply here too.
Partnered Communities:
Logo design credit goes to: tubbadu
view the rest of the comments
I can't speak for the community, however in 2022 when ChatGPT was in development, lots of us were excited about it. When ChatGPT came out, and the image AIs came out, I tried them and was pretty impressed at first. However as I learned more about them and time went on, I learned to not like them, and avoid them. This is what happened:
So that's why i don't like it and try not to use it any more.
That is a reasonable list I 100% agree with you. Many of them are part of my experience. Especially as using AI as an authority. You can't even convince them otherwise.
However, I've got a question for you: have you noticed any good or correct usage of AI.
Recently I've been looking up ethically trained models and there do seem to be a handful out there. I haven't tried them yet.
I'm pretty ok with the ones that help subtitlers do their work but that's about it that I've found so far.
First point is quite important. In some cases it's merely nuisance, but in the extreme it could cause a fatal error. Reminds me of a parable:
An astronomer keeps a subscription to a newspaper, and uses it to keep informed about politics, world issues, etc. One day the astronomer is excited to see that their paper is running an article on space: planets and stars and black holes and shit. Excited, they flip over to the article to see what they have to say and how they're educating people. They're shocked at the contents. They had expected that the paper would dumb things down with metaphor or layman's terms but it's worse than that - key parts of the article are factually, demonstrably, and completely wrong. Maybe not "Sun revolves around the Earth" stuff, but easily verifiable things and common misconceptions.
Realizing how badly the paper does within a field they know intimately, the astronomer cancels their subscription. If the newspaper reports about special interests are so wrong, they clearly aren't a good source of truth for the things that matter.
AI provides the output of that newspaper dialed up to a million percent by volume.
My boss used to work in auto repair and loves sharing stories about how ChatGPT is very wrong about error codes for certain cars, many times in a way that's very easy to look up.
He then nearly spent $1000 on parts that would have been totally useless to the company because he couldn't be bothered to read the spec sheet and trusted ChatGPT to tell him which part we needed rather than the person actually working on that project.
Your first bullet point is why I encourage people if they are going to use it to use it for something they know really well or will see the effects where mistakes are not going to be that bad with consequences (video games). I don't think anyone should use it for things they don't know till they get a good idea of how it performs on things they do.
No, it's not about knowing the llm that would ever make it OK to use on a subject you do not know...
The entire point is you have to know when it is spitting out jibberish.
What I meant is then they can see its limited and be aware that you have to evaluate its response. Its pretty much the same as searching the web was but we all know many people just found something and took it as gospel. Its kinda the same thing.
Yea, and if they're not a master in a field, they cannot validate its output.
Just because an LLM didn't output garbage for one query does not magically make it less likely to output garbage on a different topic. That's one of many reasons people in the know really hate this. You cannot just trust the output ever, which begs the question, "why do these rich goons all trust it enough to replace jobs?".
Because they're idiots that cannot validate any part of it.
You don't need a masters anymore than you do to read a paper in the field. I mean you can say the exact same thing about web searches. You can use logic and reason combined with some double checking to validate its output. I personally have instructions to provide references for all answers it gives so that I can double check and get a better idea about the thing im looking into. It is disturbing though how much of its output is hidden behind its learning and it can't give you a reference.
Tell me you've never designed and built something truly complicated without telling me.
Some things don't show up before you see the crash unless you know what you're doing intimately in many fields.
yeah but that is not what people are using llms for.
People misuse tools all the time. Why defend ignorance with ignorance?
where are you going with this. you said people need a masters degree for what they where going to validate and its just not what people on doing with it. well normal people who knows with corporate stupidity. we are talking about typical folks here though.
You're the one that brought up master's degree. I'm just saying people need to know how to validate the output.
Just because idiots will misuse something doesn't magically make it ok for anyone else to misuse a tool, too.
you said "master in a field" that implies a masters degree.