LLMs... Do you use them?

[ThunderPerfectWitchcraft]

⚑ Moderators Note ⚑

Hi there.

The posts that were held back for the moderation discussion have been restored; we found that they are - by and large - topical enough. However, we would like to point out that the OP asked about personal experiences and reasons to use (or not use) LLMs - in that regard it is okay to bring up ethical and environmental aspects (and even pure matters of taste) but please try to focus your post rather on your personal experiences than on the perspective of the society as a whole. If you want to discuss the latter, there is a thread about this here.

Thank you.

[st3phvee]

I'm glad the discussion was restored! I think it's an important discussion to have, especially at this time when every major tech company out there is doing its utmost to force LLMs upon us all, in all aspects of our lives -- whether we want it or not.

For me, at least, the societal impact and ethical problems / dangers associated with LLMs are inextricable from my personal reasons to use / not use this technology. I tinkered with it a little when GPT-4 was released to see what all the fuss was about (I wasn't impressed), and I've occasionally used it to demonstrate how bland and repetitive the text generation is, but that's the extent of it. The last time I used it, I just felt gross. It didn't feel worth it at all, given what I've learned about it.

[ThunderPerfectWitchcraft]

...especially at this time when every major tech company out there is doing its utmost to force LLMs upon us all, in all aspects of our lives -- whether we want it or not.

I guess that over 99% (all except the obligatory "Let me check this out"-initial usage, I did with some image generator) of my "AI uses" happened when I used other peoples hardware, searched something with the default search engine and ended up on google with their AI assistant. The only software I personally use that includes AI-stuff is Firefox - since I set another default search engine, I was not affected, but still, if there would be a viable alternative, I'd change over for this - so all in all, it was rather easy for me to dodge AI usage (using nearly exclusively free/libre open source software seemingly helps a lot in this aspect) - currently it still works out well. But this seems to be a privileged situation - checking through this thread, there is seemingly a good portion of involuntary AI-users already, who are forced to use it at their job.

For me, at least, the societal impact and ethical problems / dangers associated with LLMs are inextricable from my personal reasons to use / not use this technology.

I've similar feelings - for me, it is primarily the environmental impact; even if the results would be somewhat good or handy (and I don't think they are) the prize would feels too high for me. And the social aspect and effects, and the underlying ideology of the AI-technology (that is inherently capitalist; and I believe that it is partly for this reason that the problems of capitalism - for the social and environmental problems of AI that are often mentioned here are truly problems of our economic system- are accumulated in it, and come to a head) would be also relevant reasons for me enough not to use this tech even when seen isolated.

However, individual decisions have only a small impact - similar to other problematic fields, policital actions seem to be the only viable way to deal with these problems - but that is out of this threads scope.

[devils]

OP here once again. I'm personally glad we've debated the ethics of generative AI and LLMs here- I'm certainly not super into these things, and I wanted to see others' opinions (thank you for sharing them btw! ). That being said, I understand why moderators were struggling a bit with this one, ahah... sorry about that!

Anyways...

The mock-friendliness Reddit-speak ChatGPT and Grok use is twee, but it's no surprise why people cling to them during the loneliness crisis. Search engines are enshittified, and using AI to search "feels like" asking someone a question and having them work with you. It's really fucking sad to skim Reddit or other sites and see people confess to using ChatGPT to vent or as therapy because they have no one else to talk to. I used to be in a subreddit for people who have trauma from negligent therapists and mental health professionals, and a lot of people report AI therapy being better because it's very 101 and won't have human biases. Professional therapy made my PTSD worse to the point I haven't seen one in 4 years, so I can understand how that's tempting.

I really like your reply, and the issues regarding mental health services and how they can make AI chatbots tempting are something I hadn't thought about. I will say that «AI therapy» is something I completely disagree with, as most chatbots tend to just confirm what the user already believes- I don't think therapy is for everyone, but AI can be especially dangerous to those who are prone to magical thinking.

[fablefound]

As someone who's been in therapy for over a decade (thanks, getting medically acknowledged OCD at 6! /s), I do worry about what happens if people seek out LLMs as therapy given their syncophantic habits. I've had HUMAN therapists do that, and considering AI's not really the best at saying "no" or making people question their own cognitive distortions, it's a real concern.
Like, AI psychosis aside (a PRETTY BIG thing to put aside but i digress), what's the point in a therapist that'll just smile and nod the whole time you rant to it?

Also, gonna vouch a lot of the other stuff people have said about AI already: it's diabolical for the environment, it makes my stomach turn to think about how much work from real people has been stuck into a blender and then mashed together into slop, and I do worry about the consequences of long-term use. I know people who pretty much use it the way they use Google, and that's kinda terrifying.

[brisray]

I don't use any AI and I doubt I ever will. I think I know it can be very useful but it's my preference not to.

I see programming as a puzzle, you can find all the bits, but how do you put it all together? I know AI can write programs very quickly, but I worry that its taking away people's problem solving qualities. I don't see anything wrong with using it for some projects but several people have found it's not improving security for public facing projects at all.

From the logs, I know the LLMs scan my sites fairly regularly. So much so I've taking steps to stop them visiting the sites at all. I've never minded people using them for odd bits and pieces but I've noticed a lot of sites are written using nothing but LLMs and I find them boring to read - there's no humanity in them.

Advertising blurbs have always been odd. most of them are just like detergent adverts, best ever, new improved, superlative, fantastic etc. and I stopped reading most of them years ago.

I had a weird thought a while ago, if AI learns what to write from everything it can get to, what happens when nearly everything is written by AI?

[ThunderPerfectWitchcraft]

I had a weird thought a while ago, if AI learns what to write from everything it can get to, what happens when nearly everything is written by AI?

This is not a weird thought, but a existing problem. The omnipresence of AI-generated text will reduce the quality of the models, making AI-free samples a valuable good.

As someone who's been in therapy for over a decade (thanks, getting medically acknowledged OCD at 6! /s), I do worry about what happens if people seek out LLMs as therapy given their syncophantic habits.

There is a very, very good article at the Guardian about the first developer of a bot that imitated the use of human language. Since the easiest thing to do was to make it ask questions about what the other person just said, he styled it after a therapist - he had a lot of experience with them, since he was affected from depression himself. He was filled with horror when he noticed that not only normal users thought they could discuss their problems with the program, but also therapists told him that they believed it could work as a tool in therapy. Over this, he became a vocal critic of the digital revolution and its effects on society altogether.

[fablefound]

He was filled with horror when he noticed that not only normal users thought they could discuss their problems with the program, but also therapists told him that they believed it could work as a tool in therapy. Over this, he became a vocal critic of the digital revolution and its effects on society altogether.

ohhh lordy yeah that is kinda mortifying. straight-up psychological thriller in the making: everyone thinks the ai is so helpful, but only people who know how the darn things work recognize the danger. there have been some really tragic incidents already, including loss of life, attributed to ai use, but everyone thinks that they'll be different, that they can use it "correctly."

[Seren]

I have used ChatGPT less times that I have fingers, and that was really only to experiment and see what it was like, and one or two times where I wanted to look something up, but couldn't for the life of me put into words what exactly the thing I wanted to look up was called.

I don't really see LLMs as very useful to me, I can do most things a LLM can do by learning it myself. I don't like to outsource my brain to something like that unless I literally cannot do it after trying it myself from multiple angles.

I also hate how fake it feels. It makes me feel gross when the LLM tries to validate me or compliment me. I just want an answer, I don't want to be flattered and pandered to. It feels like a love bomb lmao,  I    h a t e    it.

So I stay away for those reasons. Plus I just feel better when I can put pieces of information together myself instead of asking the 'all knowing answer machine'. Especially when I have to go and then fact check it anyway to ensure it isn't just making stuff up. I find it's faster to just figure it out myself.

I can definitely see it being useful in some instances for people, but I just don't want to get too comfortable using one to the point I stop thinking for myself, because I know myself well enough that I would take the easy way to just use it if I allowed myself to.