I think comparing an LLM to a brain is a category mistake. LLMs aren’t designed to simulate how the brain works - they’re just statistical engines trained on language. Trying to mimic the human brain is a whole different tradition of AI research.

An LLM gives the kind of answers you’d expect from something that understands - but that doesn’t mean it actually does. The danger is sliding from “it acts like” to “it is.” I’m sure it has some kind of world model and is intelligent to an extent, but I think “understands” is too charitable when we’re talking about an LLM.

And about the idea that “if it’s just statistics, we should be able to see how it works” - I think that’s backwards. The reason it’s so hard to follow is because it’s nothing but raw statistics spread across billions of connections. If it were built on clean, human-readable rules, you could trace them step by step. But with this kind of system, it’s more like staring into noise that just happens to produce meaning when you ask the right question.

I also can’t help laughing a bit at myself for once being the “anti-AI” guy here. Usually I’m the one sticking up for it.

You’re right - in the NLP field, LLMs are described as doing “language understanding,” and that’s fine as long as we’re clear what that means. They process natural language input and can generate coherent output, which in a technical sense is a kind of understanding.

But that shouldn’t be confused with human-like understanding. LLMs simulate it statistically, without any grounding in meaning, concepts or reference to the world. That’s why earlier GPT models could produce paragraphs of flawless grammar that, once you read closely, were complete nonsense. They looked like understanding, but nothing underneath was actually tied to reality.

So I’d say both are true: LLMs “understand” in the NLP sense, but it’s not the same thing as human understanding. Mixing those two senses of the word is where people start talking past each other.

But it doesn’t understand - at least not in the sense humans do. When you give it a prompt, it breaks it into tokens, matches those against its training data, and generates the most statistically likely continuation. It doesn’t “know” what it’s saying, it’s just producing the next most probable output. That’s why it often fails at simple tasks like counting letters in a word - it isn’t actually reading and analyzing the word, just predicting text. In that sense it’s simulating understanding, not possessing it.

This is exactly the use-case for an LLM

I don’t think it is. LLM is language generating tool, not language understanding one.

It’s not about “AI stonks” really. If one genuinely believes that AGI will be the end of us then any form of retirement savings are just waste of time.

I really think that most investors aren’t as hyped about AI stocks as the anti-AI crowd online wants us to believe. They may have increased the weight of their investments on the tech sector but the vast majority of investors are aware of the risk of not diversifying your portfolio and if you’re someone with actual wealth you can invest then they’re probably not putting it all on Open AI.

The recent drop in AI stocks that was in the news a week or two back doesn’t even register on the value of my portfolio even though nearly all of the top companies on it are tech companies.

But the title had “AI” in it.

Also to answer your question: https://lemvotes.org/post/programming.dev/post/36238264

LLMs, as the name suggests, are language models - not knowledge machines. Answering questions correctly isn’t what they’re designed to do. The fact that they get anything right isn’t because they “know” things, but because they’ve been trained on a lot of correct information. That’s why they come off as more intelligent than they really are. At the end of the day, they were built to generate natural-sounding language - and that’s all. Just because something can speak doesn’t mean it knows what it’s talking about.

I’m just trying to pick your brain and figure out what exactly is good about Lemmy in this case, but you seem reluctant to give me a cohesive answer.

But what’s the good thing? Yeah Lemmy might not be doing it but I can do it, and Elon can, and Zuck, and Putin and your grandmom. Whatever you post on Lemmy is as public as it can get.

Let this be a reminder to anyone with an account with over a thousand comments: time for a new one.

Facebook can figure out all of this about you just from what you like and what links you click. Now imagine what a fucking goldmine a few years of your post history is to a deep learning algorithm – let alone someone who’s been using the same Reddit account for two decades. I bet they know those people better than they know themselves.

How exactly is it a good thing in this particular case? All this information is only more accessible on Lemmy.

No, AI stands for Artificial Intelligence, and this isn’t about automation either. The demand for computer scientists just spiked during the pandemic when everyone was stuck at home on their computers - not so much once it ended.

If you had spent the 20 seconds it took to write this comment to actually start reading the article you’d find out the true reason and that AI is not even mentioned.

AI is not the money maker people claimed it was going to be.

When people make claims about what “AI” is going to do in the future they’re talking about Artificial General Intelligence - not Large Language Models.

Or more likely it was a shitty job that shouldn’t have been done by a human in the first place.

It is a big part of the issue, but as Lemmy clearly demonstrates, that issue doesn’t go away even when you remove the algorithm entirely.

I see it a lot like driving cars - no matter how much better and safer we make them, accidents will still happen as long as there’s an ape behind the wheel, and probably even after that. That’s not to say things can’t be improved - they definitely can - but I don’t think it can ever be “fixed,” because the problem isn’t it - it’s us. You can’t fix humans by tweaking the code on social media.

Ofcourse not. The issue with social media are the people. Algorithms just bring out the worst in us but it didn’t make us like that, we already were.

They also mindlessly broadcast whatever lies the CEOs of the big tech company’s are telling.

LLMs? Want to give an example of this? When ChatGPT 5 came out I asked it what the “thinking built in” means and it told me it’s probably just a marketing term.

no one else sees that the media is driving a weird campaign against AI

Media criticism tends to fly out the window the moment the narrative aligns with someone’s personal views. Echo chambers can be hard to recognize once you’re inside one. If I didn’t come to Lemmy, I’d barely even know that there are people who’ve made hating AI their whole identity - they seem to be nowhere to be found in the real world.

For me LLMs have been the biggest thing since podcasts. Feels almost like gaslighting to read this AI hate here virtually every day as it doesn’t even remotely align with my personal experience of it.