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 -
I'm thinking about the culture war around AI, specifically the whole UBI debate. If AI truly does take over a lot of human work, there's a lot of people who are savagely agitating for a UBI on one side, saying we'll be post work. The other side of course says no that's not how it works, besides we aren't even close to being able to afford that. The left (generally) takes the former, while the right generally takes the latter.
What I'm surprised by is why nobody has so far mentioned what, to me, seems the obvious compromise - we just shorten the work week! As our forefathers did forcing a 5 day, 8 hour work week, why don't we continue there? Go down to a 4 day work week, and/or shorten standard working hours to 6 per day?
If AI truly will obviate the need for a lot of work, how is this not the more rational solution than trying to magically create a UBI out of money we don't have? How come this idea has barely even entered the discourse? I have been talking and thinking about AI unemployment for years and never once have heard someone argue for this compromise.
To be honest the existence and shape of much of this discourse continues to baffle me. There's a discourse around AI causing unemployment, even though AI thus far has not caused any unemployment, and there isn't an obvious mechanism for it doing so. Isn't the evidence so far that incorporating AI into a workplace increases workload, rather than decreases it? It's always possible that this changes, but I'd at least like to see the argument that it will, rather than it just being assumed.
The pattern seems to play out time and time again - Scott's last post about China made me want to scream something. Where is the reason to think that AI is so militarily and economically significant at all? What if this is all nonsense? Isn't this all based on a vision of AI technology that has no justification in reality?
Maybe there's an AI 101 argument out there somewhere that everybody else has read and which passed me by entirely, but right now I continue to be incredibly confused by this discourse. We made systems that can generate text and images, but which are consistently pretty crap at both. Given time I can imagine them becoming somewhat less crap, but where do they pivot or transform into the sorts of devices that could cause massive technological unemployment, or change a war between great powers?
This just isn't true. Big companies are sacking people because of AI. Chegg, Salesforce, IBM, BT Group, Morgan Stanley... More are freezing hiring for juniors. Why are so many artists complaining about AI if it's not costing them anything?
Modern warfare runs on software. The logistics chains, communications, intelligence-gathering and analysis, sensors communicating with eachother to guide missiles over 1000s of kilometres, electronic warfare... all of it relies on an extremely complex base of computer code that nobody really understands that well.
AI improves that. If your drones can't be jammed because they're autonomous and can find targets on their own, that's a critical military advantage. If your radar software gets optimized by some black-box AI to counter whatever arcane modification the enemy made to their jamming software, that's a major military advantage. Optimization of complex systems in unintuitive domains is a strongsuit of AI. See AI-designed computer chips, Google has been doing that for a while. Modern AI systems are also useful for controlling high energy plasma in fusion reactor chambers, predicting the weather (obvious military and economic significance) and countless other complex domains. Cyberwarfare is another obvious domain where AI is relevant: spear-phishing, reconnaissance, actual infiltrations...
If you can quickly process huge amounts of satellite, infrared, aerial, sensor data to provide firing coordinates to your forces, that's a major military advantage. Not to mention fast translation of signals intelligence... There just aren't enough analysts to cope with all the data that militaries can scrape up.
Facebook is making billions and billions from its AI-optimized advertising, as are other big tech companies. Consumer-end text and images are just the tip of the iceberg.
It's not just 'producing crap text'. The text is valuable and useful. Domain-specific programs are valuable and useful. General text-generation (which is capable of doing advanced cyber tasks like writing kernels or performing cyberattacks) is valuable and useful. I can tell it's valuable and useful because people are paying billions for it!
Nvidia products are killing people at the front in Ukraine right now. Hell, an AI found me these links.
https://www.longwarjournal.org/archives/2025/06/ukrainian-intelligence-details-russias-new-v2u-autonomous-loitering-munition.php
https://isis-online.org/isis-reports/russian-lancet-3-kamikaze-drone-filled-with-foreign-parts
In conclusion, it's obvious and straightforward that AI is hugely important. That's why the great powers are racing to develop it, why the US is anxious about China getting AI chips, why the megacorps are investing hundreds of billions in it. The worldview of the AI-believer is simple and makes sense 'powerful technology - big investment - widespread use' whereas the AI-doubter is mired in weirdness 'mostly useless technology - big tech just throwing money down the drain for some inexplicable reason - no widespread use once you ignore most of the use'.
Let me ask a practical question. That's a lot of if statements you made there.
Has AI actually done any of those things? The specific examples you give of things that already exist are mostly speculative - all I can find about AI-designed computer chips, for instance, are hype stories in pop science magazines, rather than anything credible, and even they include the note that most of the AI designs did not work.
In general I am skeptical of the argument that goes, "I can tell it's valuable and useful because people are paying billions for it!" In a sense that proves that it's 'valuable', insofar as you can define value in terms of what people are willing to pay for, but none of that proves that it's useful. People are willing to pay vast amounts of money for obviously worthless things on a regular basis - NFTs are one infamous example.
I can concede a handful of highly technical niche applications - protein folding, plasma confinement, etc. - though even there I'm a little cautious. (I don't understand those technical fields, but in fields that I do understand, where AI is being hailed as a major breakthrough, the breakthroughs once analysed turn out to be, at best, heavily overrated.) But the AI-believer position, in cases like this, are that AI is literally going to make labour obsolete, or that AI is going to become superintelligent, achieve god-like power, and either usher us all to utopia or to utter destruction. And that's a position that is so far in excess of any reasonable estimation of what this technology does that I have to raise my eyebrows. Or yell at a blog post on the internet, I suppose.
The Nvidia AI chips in Russian missiles, performing autonomous targeting to bypass jamming, per my links.
Also, per the chip article, some do work and that's the key part? It's easier to simulate a chip design and check if it works than to design a chip with superior performance.
I think the trend is pretty clear. Right now AI is causing some unemployment, producing some economic gains (mostly concentrated in big tech), adding some military gains. I expect this trend to continue and accelerate as the tech gets better and adoption improves.
Where is the evidence that incorporating AI into a workplace increases workload, rather than decreases it? Reminds me of the Yes Minister quote about thousands of new staff being hired to deal with the chaos caused by the labour-saving computers... but we don't seem to see increases in employment amongst AI adopters.
NFTs aren't useful but people certainly did value them, it's just a novel subgenre of art/signalling good. I personally don't want to buy a bored ape or ugly abstract paintings, a CS GO knife skin or an extremely expensive watch that's functionally inferior to my Casio but I accept they have some kind of value. Anyway, people aren't buying AI because it's classy to have (indeed, its gotten pretty low-status), they buy it because of its utility, convenience, cost-efficiency.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
It's obvious if you assume the models will improve up to, and then past human level intelligence.
At that point every job that can be done from behind a computer becomes trivial to automate. The remaining jobs become trivial once AI control of robots improve as well.
Now we're not there yet, and maybe we won't ever get there, but it's pretty hard to be confident one way or the other.
It continues to frustrate me that nobody seems to have found (or be seriously looking into, as far as I can tell) theoretical bounds on "intelligence", and some philosophers in these parts seem to assume that something "smart enough" can derive a complete physics, the universe, and divine the state of everything in it given nothing more than the text of the ten hundred most relevant books, which feels very ontologically lazy.
Although I'd be interested in reading anyone looking at this mathematically, presumably needing a very heavy dose of information and complexity theory. Links are appreciated.
I didn't mean to imply that "beyond human level" meant "machine god". But even going just slightly past average human level has potential for massive societal upheaval as it would very quickly devalue much or most human labour (even if some number of high performing humans can still outperform AI, many of us aren't fully using our cognitive capabilities in our day to day).
More options
Context Copy link
Does a set of all sets contain itself?
Yeah, but now you're into the territory of religions, specifically those that suggests a deity actively maintains the (finite?) state of the universe in this way.
More options
Context Copy link
More options
Context Copy link
I think this includes a number of questionable assumptions built into the idea of 'human level intelligence'. The models we have now are very good at doing some things that humans struggle with, but are also completely incapable of some things that are trivial for humans. There isn't a unified 'intelligence' where we are at a specific level, and machines are approaching. Rather, human intelligence is a highly-correlated cluster of aptitudes; aptitudes which do not necessarily correlate in machines. It seems at least plausible to me that existing AI models continue to get better at the sorts of things they are currently good at without ever becoming the kind of thing we would recognise as intelligent.
Now on one level that doesn't matter - I'm just suggesting that AI might keep improving without ever becoming AGI. But AI doesn't need to become AGI to cause technological unemployment, or to give some nation or other a major military advantage, or whatever else it is we're worried about. But I'd still like to know what the mechanism we're predicting for that unemployment, or military advantage, or whatever else might be, because it is not immediately obvious how a language model produces any of those things.
More options
Context Copy link
More options
Context Copy link
AI and techno-futurism in general are the dominant religion of our times. They aren't seen as a religion because the worshipers ignore metaphysics and fundamental axioms. So yes, it's based on a religious vision a lot of the time.
Metaphysics and fundamental axioms preclude human level AI? Can you elaborate?
Depends on what you mean by 'human level AI'. I believe they preclude machine consciousness, but I don't know it well enough to explain it. I'd recommend the book The Experience of God: Being, Consciousness, Bliss if you want an intro to classical theistic metaphysics.
By "human level" I mean an AI that can perform tasks at the same quality as a human, such that you could replace a human employee with an AI employee.
I actually think consciousness might not be required for this level of AI. It's like how chess AIs are superhuman at chess without being conscious (probably).
I'm also not a theist. Any arguments about AI that really on theism are not going to be convincing to me.
More options
Context Copy link
More options
Context Copy link
I think what he's saying is that techno-futurism is not perceived as a religion because techno-futurists do not make metaphysical or fundamental claims.
Personally I think this is mainly a semantic difference. It's not clear to me that there's a difference between "X is not perceived as a religion because X does not do these things typical of religion" and "X is not a religion". Isn't religion defined, at least extensionally, by the things typical of religion?
I don't think the concept of religion helps very much here. Better to just say that AI hype is a form of collective irrationality or delusive behaviour, if that's what he means.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
Rational or not, companies are radically reducing full time employees (FTEs) in their long term plans (LTPs). This occurred over the past 2 years, but I’m actually seeing the hiring budgets impacted now. This is everywhere and part of the bad job market right now.
This is a direct response to AI. We can debate on whether or not that’s a smart reaction, but it’s most certainly happening.
Got any specific examples? Would this be something they announce in their annual earnings reports or something else?
More options
Context Copy link
More options
Context Copy link
For example, if you / Trump / Xi take ChatGPT5000 and type in
You can turn a text generation chatbot into a do-things AI by just asking it what should be done next and then following its advice… in theory. In practice that seems not to work well, and it’s not clear why.
Because it's just picking statistically likely responses based on its training data, so it can't really suggest anything radically different (or more insightful or creative) than the human-generated information it was trained on.
You’re correct but they can’t do bog standard everyday things like running a store either.
I think there are more fundamental issues related to
a: chaining multiple stochastic processes causes randomness to build up in the system producing wacky results (even with a supervisor agent since that is also a stochastic process)
b: a lot of the things that we do are ‘learn with your body’ tasks that aren’t adequately expressed with words.
More options
Context Copy link
Yes, but the average human being can't do that either.
Doesn't that just further underscore ChickenOverlord's argument? The position he(?) is arguing against is that AI will somehow get better than any human at this, and CO is pointing out that as currently implemented, AI isn't really analyzing anything except language and so is unlikely to outperform the human-generated data it's trained on. Seems to me you're just giving a further reason to think the bar of what it can do is rather low.
ThisIsSin's argument is that the bar of what AI needs to do is low - not what it can do.
The idea being that even if the AI can't surpass the best humans, it can learn from them in order to be better than the rest.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link