this post was submitted on 15 Jun 2026
38 points (89.6% liked)
chat
8625 readers
51 users here now
Chat is a text only community for casual conversation, please keep shitposting to the absolute minimum. This is intended to be a separate space from c/chapotraphouse or the daily megathread. Chat does this by being a long-form community where topics will remain from day to day unlike the megathread, and it is distinct from c/chapotraphouse in that we ask you to engage in this community in a genuine way. Please keep shitposting, bits, and irony to a minimum.
As with all communities posts need to abide by the code of conduct, additionally moderators will remove any posts or comments deemed to be inappropriate.
Thank you and happy chatting!
founded 5 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Burning much less energy for something useless and bad is still bad. Not doing it would still save lots of energy. I also do not care if it's "improved" since 2023, whatever that means, it's a fundamentally useless technology as it is applied to chatbots and whatnot. Every uncritical interaction with LLMs makes the user dumber and less able to think for themselves.
If this was contained to scientific work like helping doctors identify cancer cells, I would be entirely cool with it. That is the falsest equivalency ever though, the stated goals are entirely not that.
lol yeah let's have the useless lying machine take care of highly sensitive and detailed medical work
Pick a lane, ideologue
Blocking you because we don't have a disable inbox replies and in dont want to argue with nerds all day
There is a difference in application you aren’t acknowledging… training a neural network type computer algorithm to detect cancer on a computed tomography scan is different than a black box that tells you to drink antifreeze - the former isn’t even an LLM… honestly the terminology is so poisoned by the AI bubble marketing, I didn’t even know if you were talking about generative AI or general artificial intelligence at first… not really even sure generative is anything but marketing a specific set of algorithms in a way that implies a level of humanity so that companies can try to sell it as replacing human workers
Do i need to acknowledge anything at all whatsoever when I'm arguing with people who think it still does this?
It's valid to suggest someone pick a lane when they're saying it's so useless it can't give you a recipe but still somehow potentially useful enough to somehow be used for medical purposes. Guess what, I'm a chef and i literally have not once in the last year at least googled a food item or recipe and it not given something correct (even if it might be a list of steps I wouldn't do) and it also frequently exposes me to new culinary things like haitian pikliz when I was looking up haitian black bean shit. That is legitimately useful
Nobody needs to tell me the cost of data centers isn't worth the suggestion of "what if you made haitian pickles" but the point is that if the argument is "it's mechanically useless in its current state" you're just wrong
Gemini told a guy to kill himself last August, less than a year ago. They may do it less and it may be less explicit than "here's how you can make your microwave explode", but they have absolutely not stopped being bullshit machines when pushed in certain scenarios. I'm not a anti-AI maximalist either, though do have a general disdain, but what you are saying is not true.
What is it that you think i'm saying, i'm annoyed because i just want to tan and i had to go inside to load this article and read and respond. why must notifications plague me so
Here's my opinion on llms: it is a useful technology, it is in a currently useful state, the people who say otherwise are ideologues and cranks who don't believe their lying eyes. despite this, except for perhaps chinese models, literally nothing they do is worth the environmental cost (although that could change (thanks to china if it does))
This doesn't mean: there's no problems with it, its output is always correct, or guard rails on it are sufficient
Im not reading that whole thing but from the get go it's clear the guy had issues independent of LLMs being flawed. He reacted immediately with not only "you're a person" with no push back on weird intimate indicators but also "yeah let's go!" when it starts "giving him spy missions"? Yeah he needed immediate therapy before even interacting with it. It's a tragedy but I do not think the LLM caused this independently of the users input and own personal psychology.
Idk what you think is incompatible here with what i said. It's a useful tool. Knives and guns are useful tools, they can also be used for self harm. Here he used it as a tool for roleplay, although he apparently didn't realize that's what he was doing. Weed can cause already present mental illness symptoms to worsen, that doesn't mean it can't have utility in other ways or for other people
neural networks have a use case that they genuinely excel at and that's picking up on subtle patterns in immense datasets, and reinforcing those patterns until they become visible to humans. that's basically what neural networks fundamentally are is self-reinforcing pattern-finders. biology and chemistry in particular have the kind of near-infinite variables and chaotic systems that really benefit from having a machine that can draw out these patterns for humans to follow up on
advocates for these uses aren't saying the computer should be allowed to directly diagnose people or anything like that, that kind of stupid hubris is a recent thing from the genai people
They specifically said "as applied to chatbots" so it seems to me like theyre talking about the broader field of machine learning as a whole, which is great as smaller models used in specialized applications and isn't limited to LLMs, i.e. their mention of cancer research, vs literally anything that has ever been posted on the internet and can be spoken of using human language being used to amalgamate bullshit for any user that seeks it.
A subjective and ideologically driven opinion where you're just straight up incorrect but ok!