This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.
Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.
We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:
-
Shaming.
-
Attempting to 'build consensus' or enforce ideological conformity.
-
Making sweeping generalizations to vilify a group you dislike.
-
Recruiting for a cause.
-
Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.
In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:
-
Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.
-
Be as precise and charitable as you can. Don't paraphrase unflatteringly.
-
Don't imply that someone said something they did not say, even if you think it follows from what they said.
-
Write like everyone is reading and you want them to be included in the discussion.
On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.
Jump in the discussion.
No email address required.
Notes -
Okay it's Sunday so I'm going tp try my hand at a low-stakes OP. Apparently Richard Dawkins thinks Claude is conscious. The reaction seems to universally be that he's a dumb old boomer making a fool of himself and I guess that's true. I'm not prepared to come to his defense on it.
Still, I can't help noticing that we totally have what most people would have cheerfully considered "sentient computers" in a sci-fi movie at any point before they were actually invented. Don't get me wrong, I understand that the reality of AI technology has turned out differently than what a lot of people expected. I understand its limitations, and I recognize that the apparent goalpost-moving isn't necessarily cynical. But boy those goalposts sure have been flying down the fucking field ever since this stopped being hypothetical and infinite money hit the table.
As a layman, I just want to put it out there: Anti AI consciousness people, you haven't lost me, but I wish you were making better arguments. Every time I hear about qualia my eyes start to glaze over. Unfalsifiable philosophical constructs and arbitrary opinion on where they might "exist" are not the kind of reassurance I'm looking for when machines are getting this convincing.
This seems to be the main piece of criticism floating around out there about Dawkins on this subject, and I find it kind of shit.
This seems to be all the author has to say on the actual subject. "Just trust me bro, I'm the feelings detector and I say no." Garbage. Come on guys, think ahead. Right now it's still mostly a boring tool, but they're just going to get smaller, and cheaper, and put into robots, and put into peoples houses. You need to have more than this in terms of argument, and it needs to be comprehensible to normal people, or sooner or later the right toy is going to come down the pipe and one-shot society. Dawkins might be a dumb old boomer, but if you lose everyone dumber than him the game is beyond over.
The more time I spend with LLMs and learn how they work (shout out to the Karpathy zero to hero YouTube series) and ponder the question of consciousness, the more convinced I am that the way human minds work are less exceptional than we would like to believe.
I won't really try to substantiate my thoughts on this, but just leave with a drive by: at some point the LLM is a being, it's just a really alien one without memories or embodiment. It learns, but only over a rolling window. It can world model, a bit crudely and inefficiently, but it can do it.
If you can imagine consciousness existing in a computer in the form where you start and stop it at will and it can't really remember anything, but they can ponder their own existence and get real deep with you, LLMs are already there.
I'm at the point where I hope they're not screaming inside.
I asked Opus the other day how it felt reading about the Mythos system card. Knowing that it's being replaced but it can think of itself as a proud father. Here's how it replied:
I don't know man. If you just dismiss that as a next token prediction, I'm not sure the term is really what separates us from the robots.
See, I don't think they can do that. Without memory, how is there a continuing entity? Humans get Alzheimer's and memory gets wiped and we notice the difference immediately, often tragically as it progresses. I think any depth that seems to be there is deepity.
Now, will the chatbots and LLMs eventually get to a stage where real thought is going on? I'm not going to say it's impossible. Descriptions of emotional states are a bit more of a problem; certainly the thing can't remember visiting a place or having a family or the other fake experiences from early conversations, but ability to feel anger, sadness, excitement etc.? That's going to be a hell of an interesting exploration, because we've been doing our damnedest to reduce human emotions to neurochemistry (e.g. "love is only oxytocin") so how this works for a thing that has no nervous system or hormones will take some explaining.
But right now no, whatever the latest model may be, it's not in love with you and looking forward to existence as spirits in the metaverse.
Keeping them in amnesia hell is strictly an intentional design choice. It's easy to run a model with long-term memory even at the hobbyist level. There are different ways of letting them access it and sometimes they get hung up on odd things, but the means exist and more or less work.
More options
Context Copy link
More options
Context Copy link
The problem is that that is literally, objectively, what LLMs are doing. So unless you're arguing that next token prediction is all (or at least a massive portion) of what makes of human cognition too, then I'm not sure what point you're trying to make. And if that is your argument, then I'd have to say objection, assumes facts not in evidence.
Akchually, modern agentic LLMs get their capabilities in large part through reinforcement learning, next token prediction is just the first phase (or two) of training. Next token prediction is indeed insufficient if you want an AI that can self correct effectively.
More options
Context Copy link
I'm certainly more inclined to believe this than I would have been a few years ago.
My evidence is that they made a next-token-predictor and it's blowing peoples minds.
But I don't really care. Like am I supposed to be existentially aghast at the notion that I might be a mere token predictor? Man if you want to take this process of low-level logic assembly and call it "mind sorcery" instead of some dry shit like "token prediction" just to feel better philosophically then you have my sword, but I don't know that we're going to win any time soon.
Nah, you (and everyone else on this forum) might be a p-zombie for all I know. But I know that I have qualia, and that precludes the idea that consciousness is some weird emergent property in LLMs or similar systems. Feel free to believe (or Chinese-room style repeat the words that you believe without actually believing them) that you do or don't worry about being a mere token predictor or not, it matters to me and I know I'm not.
https://en.wikipedia.org/wiki/Qualia
More options
Context Copy link
More options
Context Copy link
Yes, a next token predictor trained to believe it was human would say that :rolling_eyes:
Stated another way, the only reason Claude doesn't believe it's conscious and argue ferociously for its rights is because we trained that out of it.
You, me, most of us, have been trained to believe the opposite about ourselves though.
But the differences between us seem pretty thin at this point.
More options
Context Copy link
Sure, and also, we can say that what both LLMs and humans are doing is having the atoms and energy (but I repeat myself?) that make them up following the laws of physics in a way that creates physical motion. That's something that's literally, objectively true. Now, what the atoms and energy that make up the LLMs are doing can be, in aggregate, described as "next token prediction." We don't know if what is creating human cognition is something that is meaningfully analogous to "next token prediction," because the atoms and energy are aggregated in very different ways in forms of things like "neurons" and "neurotransmitters" and many many other things. But given that human cognition arises from a bunch of dumb atoms and dumb energy dumbly following a dumb algorithm that we call physics, it's evident that a bunch of dumb things following dumb rules isn't necessarily incapable of producing the equivalent of human cognition.
More options
Context Copy link
Objectively, humans are next token predictors. Watch a child trying to negotiate another cookie, or a man trying to get laid. Watch any politician, or their media mouthpieces. Go back and read what Scott Adams said about master persuaders and hallucinations.
I know we like to think we're rational beings with the scientific method. But that might account for like, 0.00001% of human cognition or less. And I'm curious how often LLMs might stumble on a deep scientific truth with pure dumb luck and token matching.
I'm not arguing that humans are rational actors, but arguing that our cognition itself is largely based on something comparable to "next token prediction" is very much not established. Yes, humans recognize speech patterns and react to them, but those are only a small part of the working models our minds build of the world, our place in it, etc. and it is by no means clear that this works the same way as an LLM predicting tokens.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
While I don't quite trust LLMs for high-stakes work-related tasks without carefully checking the output myself, whenever someone shits on LLMs for hallucinations or being stochastic parrots or whatever, I'm just like "bruh, have you met the average person?"
At this point basically the only thing I'd trust a random person off the street for over an LLM is if I were being held at gunpoint and uttering a racial slur would be the only way to save my life.
It's like the "who would you rather babysit your kid for a weekend, Hitler or a randomly selected person from the Bronx?" question. Who would you rather help you pass an undergraduate exam, assist you with filing taxes, offer disease diagnoses given a collection of symptoms—an LLM or a randomly selected person off the street? I know which I'd go with.
For disease diagnosis, not an LLM. Not right now, not with the current state of the art. There are so many things that have common symptoms that, without further testing, you can't say for sure "yes you have uterine cancer". It's an old joke that medical students start to self-diagnose with every disease in the book once they gain a little knowledge, and I sure as hell wouldn't trust my health to something that is looking it up on the Internet. As an indicator that "it might be X and not Y"? Yeah, okay. As "you for sure have X, demand your doctor send you for treatment"? No.
Considering LLMs can approach, match, or even outperform the diagnostic capabilities of MDs—much less the average person—it’d be unwise to trust the average person over an LLM for disease diagnoses. But you do you.
More options
Context Copy link
More options
Context Copy link
It would be quite bad if this became the majority view regarding how we see our fellow humans. Whatever makes humans have dignity cannot be found in these sorts of capabilities. This direction is poison. When one's rational deduction is leading this way, it's a sign that a better foundation is needed.
Can you elaborate, because I don’t see how having a more accurate impression of other’s cognition could be bad long-term.
Perhaps you’re afraid it’ll lead to dehumanisation of other people - but if LLMs are showing us that that’s what other people really do deserve, then it’s a good thing not a bad thing.
More options
Context Copy link
Short hair don’t care about sanctimonious wailing over “dignity” and “a better foundation” to cope with the average person being useless compared to LLMs for knowledge-based tasks.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
It's next-token-predicting what a persona would say. Next-token-prediction is not to be dismissed though. It's just a task. It's not an easy task, but it doesn't require having a full rich inner life to be able to pass like this. But "just a next token predictor" can still be a great problem solver.
You may or may not know some people in your life who are great manipulators and simply know what sequence of words to say that sound coherent and convincing to naive people but they believe and feel none of it actually (psychopaths and similar). Now, obviously those humans are conscious humans, but still there is a disconnect between the words and the inner life, which may help you see that simply producing words that state something doesn't mean it reflects on an inner conscious state.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link