I am antagonistic to the adoption of LLMs for economic, social, political and environmental reasons.
Despite that, I would say this description of what they/how they work is so ridiculously simplistic that it verges on dangerous. "It's just a probabilistic database" is about as bad as a take on human beings that they are "Just meat, with some chemical energy -> electrical energy transformation", or even traditional computers as "nothing but logic gates".
Which is to say, the description is not wrong, but completely misses the point.
I think the point, with a more charitable reading of the parent comment, is that people forget the fundamental underlying technology and its inherent limitations.
"It's just a four wheeled platform powered by a box of controlled explosions" isn't a terribly useful description of a car until you get into a conversation with someone who thinks their car is depressed because it spends most of its time locked up in a garage.
I like the analogy. And yes, I'm not interested in conversing with a person who thinks that about a car.
On the other hand, the "wheeled platform powered by a box of controlled explosions" doesn't get close to capturing the impact of the car on individual freedom, the environment, the economy, and more. So if someone was saying "there's nothing to talk about here, it's just a box of controlled explosions", we can safely ignore that person too.
And I want to stress that I say all this tentatively. We still don't understand what LLMs are actually capable of or what the best way to understand them is. They may indeed be as limited as their "probabilistic database" critics suggest. Alternatively, by building a deep and incredibly complex probabilistic database of one of our primary cognitive tools, it may turn out that they are capable of much, much more, and that understanding them in "purely" mechanistic ways is not useful.
> On the other hand, the "wheeled platform powered by a box of controlled explosions" doesn't get close to capturing the impact of the car on individual freedom, the environment, the economy, and more. So if someone was saying "there's nothing to talk about here, it's just a box of controlled explosions", we can safely ignore that person too.
This is also a good point, and I agree with you 100% here. I don't think one could responsibly argue that the impact and consequences are not significant, regardless of if they believe it's near AGI or it's just fancy next token prediction.
> is about as bad as a take
What would be a good take here?
"LLMs are incredibly complex systems with emergent properties we do not fully understand, and it seems possible that because they run on top of a hugely complex model of one our most important cognitive tools (written language), they are capable of exploring the things we do with our cognitive skills in ways we would not imagine when considering their purely mechanistic or operational behavior. Or they may not."
Something like that.
Sure, if we're going based on feelings, that'd be an interesting take. Anything grounded in any research or actual science though?
Sure. There are several papers (all the ones I've seen come from the companies themselves, which is unfortunate) which treat the LLMs as semi-biological systems and attempt to understand their behavior at a higher level, which suggests higher order behavior that isn't well predicted by the "probabilistic database" angle.
Here's one that I happen to have the URL handy for: https://transformer-circuits.pub/2025/attribution-graphs/bio...
That’s not unfortunate, that’s conflict of interest by a party which directly profits from advertising their products as such.
But let’s assume this is not the case for a moment, following the article, there are some graphics and information, without providing and internal details for third party validation, no access or details to the training data, etc. basically not following any scientific methods which would reaffirm credibility.
I understand, that working with these models, it feels like they are more than “probabilistic databases” but they don’t have to be anything else in order to provide such interesting outputs.
If that's your interpretation/assessment of that paper, then I don't think we have much more to say to each other. I'm not saying you're wrong, just that we're not operating in the same universe of understanding.
> that paper
You have yet to share a paper.
> just that we're not operating in the same universe of understanding.
I concur.
Then what is the point? You didn’t actually describe where your definition differs and why. What makes it “dangerous” to think that way?
It is dangerous because it completely fails to grasp or convey what these systems might be capable of.
The point is that when you build a system with the level of internal complexity that LLMs have, they tend to have emergent properties that are not well-described by a naive and/or low level description of their operation. You will not understand what contemporary "ordinary" computers do or are used for if you think of them as only a series of silicon semiconductor transistors, even thought that is undoubtedly what they are.
We don't yet really know what the right level is to understand the behavior we see from LLMs, but I'd wager "just a probabilistic database" is not it.
Because we aren’t discussing capability; we are discussing what LLMs “think” of Humanity and its “flaws”. Calling a model a glorified text file of floating point numbers for probabilistic matrix multiplication is technically correct, and helps to strip this veneer that it is a consciousness with its own thoughts, which many articles (including the one linked IMO) want you to think. LLMs are naturally far more capable than what their nature or that description implies, but we see that every day, particularly on HN. Just as everyone knows how to work the basic functions of a computer, and most people treat computers as magic black boxes rather than treated sand formed into logic gates encased in plastic, which helps to bring them back in line with mundanity.
LLMs are plenty capable, and they are dramatically shaking up various fields, but that’s not important to the conversation you responded to. If you were talking from a general view that reductionist statements fail to address capabilities I wouldn’t say anything (outside of a hearty “that’s by design”), but you claimed they were missing the “point”. I think they understood the point of the topic thoroughly, and would say you’re missing the trees for the forest; focusing so heavily on the broader implications, that you miss the finer nuance in conversation.
People have been "taking the intentional stance" towards clearly non-intentional and maybe/maybe-not intentional systems for as long as we've had pets, and maybe longer than that.
Using phrases like "LLMs think ..." is misleading in some ways, but not in others. Taking the language police stance towards any discussion of what LLMs can, and cannot, do is, IMO, not useful.