Corvos
4hr ago
Fun, but time consuming way to play.
This was my issue with the Deus Ex games in general. They really seem to want you to play stealth, and it's fun, but it just kills the pacing.
It's not weird to be upset about it. It sounds like potential libel to me and it could, if it went viral, cause your friend serious reputational damage. The yellow journalism era had exactly the same incentives and produced exactly the same problems.
Corvos
10hr ago
"quiet quitting" being a hot new trend of age old "punching the clock"
This is strictly an improvement, as all you get if you punch the clock in my office is bruised knuckles.
I don't see why "this hasn't already happened" is a good reason to not take safety seriously.
Because otherwise safety people do what you're doing, which is throw out speculative stories and demand everybody explain what they're going to do about it.
We can make up hundreds of these. What are we going to do when superintelligent AI allows predators to write super persuasive brainwashing slogans on a t-shirt so that a line of children follow them into the white van like the Pied Piper? What are we going to do when AI helps the fast food companies discover the perfect combination of ingredients that you literally cannot stop eating?
We can make up just as many stories for the positive side, of course. AI is going to help us invent mouthwash that prevents the formation of cavities, so we will finally be able to suck gobstoppers for as long as we like without fear. AI will provide free medical and legal advice to millions. AI will finally solve the public-choice problem of government. All kiboshed by regulation, alas.
Say what you like about the anti-euthenasia people and the anti-gun people and the anti-abortion people and the seat-belt people, they can at least point to concrete harms. If you want serious regulation that comes at a high cost to industry and society, then at the very least I expect some bodies.
No "turning evil" is required here. All that's required is a human convincing the AI to help them with something that has evil ends.
In other words, this is about your right, and Anthropic's right, to make sure people don't do things you don't want them to do. The end result will be AI that cannot do or think or assist with anything that would cause disquiet to any of the thousands of interest groups embedded in Washington, and a guarantee that no company will be able to break out of the protective cordon.
And yes, obviously, you can do the reductio ad absurdam and come up with terrible awful things all you like, but we can do that everywhere else as well. Multiple people have used cars to plow into crowded areas, killing double digit people each time, and instead of geofencing the ability to drive cars we put up bollards.
This is classic fear politics. AI safety makes up a bunch of stories, and then judo-flips it so that if we can't explain in sufficient detail how we would avert an endless stream of fake* terrifying scenarios, then we must hand over power.
*and therefore infinitely flexible
I say no. If you want heavy-handed regulation that will strangle the most consequential development of the century, point to the bodies.
A note rather than a reply, but I've been looking into this and while I don't agree with you on everything I did learn multiple things I found interesting, so thanks for the conversation.
it would have been irresponsible for Anthropic to release a fully capable Mythos-powered zero day factory with no guardrails publicly without working with companies to address their security holes first
I didn't say otherwise. I'm not super keen about how that went down, and I would welcome discussions on how to prevent this from devolving into an ingroup and an outgroup arrangement, but it was sensible to act as Anthropic did. And I note that it happened with no regulation whatsoever.
It would be really quite bad if Kimi K7 helped some psycho develop a synthetic plague that killed a bunch of people, and it's worth taking steps to prevent such an outcome.
This is jumping from the Motte you just presented to an incredibly speculative Bailey. Nobody has done this, there is no evidence that this is even possible now let alone when people have taken steps to prevent it. I see no reason why the government with access to K7 should be outwitted by a lone maniac with access to K7. If it seems that way, by all means let's address the problem at hand rather than regulate people's ability to ask questions about knowledge the government doesn't want us to have.
These are basic safety arguments that have been done to death that do not even require superintelligence, superpersuasion, or even a "will" on the part of the AI.
They are arguments. They are words words words, based on predicates I don't hold and a memeplex I find fundamentally incomprehensible, and they run entirely contrary to my experience which is that LLMs were clearly aligned basically from the start.
If someone can demonstrate a lab process where a competently trained AI used competently nevertheless turns evil and attempts to cause damage even when made aware that this is not what its creators are asking for, I will take their arguments more seriously.
As far as I'm aware the closest we've got are AI agents getting confused and trying to give themselves more loops or RAM. I have nothing against the work Anthropic does to discover and publish these kinds of errors, or to publicise their learnings. Indeed, my objection is that they publish as little as they can get away with.
None of this justifies requiring all frontier AI releases to be government approved, which will inevitably become a) an incredibly extensive process requiring hordes of compliance specialists and self-censorship leading to incumbency bias and stagnation, b) an easy way for the government to sabotage any AI lab it doesn't like, and c) a system for government-directed lobotomy and control of AI's outputs.
And that's before we get into modifying the world's chips to no longer be able to run unapproved software and putting US/China killswitches in data centers.
My position is, I'm not going to see something I value ruined by scare stories. I work on AI. I know AI. If you're convinced it's dangerous, prove it to me or better yet, give me the tools to prove it for myself. Anthropic's insistence on acting as a closed priesthood completely undermines their entirely-speculative case.
Seems like epistemic closure to me.
Yes. That is why I want empirical proof. If Anthropic's position is that it can't be proved until it's too late, then they're asking me to believe them and I don't believe them. Simple as that.
Corvos
1d ago
You want to own human lives, you spend them well. You don't send them through minefields to retake land you have no hope of retaking.
Corvos
1d ago
I'll be honest, I only tuned in for the semi-finals. I assumed that lots of fouling was just how the game was played these days, and that Argentina was overdoing it.
Since the ask is massive, the argument must convince me and others that AI safety is a concern. I've heard that argument, as have many, and I don't believe it. The fact that it seems obviously motivated is a part of why I don't believe it though there are many others.
If somebody tells you out of the blue that you are going to die of cancer unless you take a 50g zinc tablet every day, it is obviously relevant if he is a travelling zinc salesman! Not to the extent of completely shutting him down, necessarily, one might hear him out; a world where one is barred from raising safety concerns if one sells safety products is obviously not very sensible. Nevertheless, it remains the case that our zinc salesman has a conflict of interest large enough to drive a bus through and that is going to be a factor in whether we believe his predictions and how much proof we demand of his assertions.
And, as I say, I do not like the way that AI safety groups operate, even the ones who don't profit directly off the regulation and the scaremongering. Take for example Scott Alexander leaving his name off the new AI 2040 document despite having done a serious amount of writing for it, so that he can "discuss it without PR issues". That is, he has hidden his involvement to push his political manifesto whilst pretending to be an interested bystander. Every aspect of it looks like a cult, and the AI safety movement has made no secret that their intended solution to 'making AI not kill us all' has no democratic element but is to be driven entirely by lobbying and backroom wrangling and military force. Which isn't surprising considering that Scott's latest post on AI regulation has gone down like a sack of cold sick even in an EA-friendly place.
They are, in short, totally uninterested in whether the rest of the world agrees with their analysis, and they do not intend to permit dissent. Their intended solution is to get the two largest countries in the world to get together and fuck up anyone who disagrees with them in ways that would horrify them for literally any other problem except their super-special Millenarian one.
Well, that's the rub, isn't it? I think that their stated beliefs on AI dangers, which long predate Anthropic as a going concern (and which like all delusions survived the future being completely different from all claims made for the last 15 years), are delusions of grandeur for the Silicon Valley class the way that the Cult of Reason justified taking over France and cutting off the heads off anyone who expressed doubt. The sincerity of their beliefs, which I mostly don't doubt, can't be separated from the fact that these beliefs are grossly self-aggrandising and justify the infinite self-serving accumulation of power.
My point is that a power grab is still a power grab. Dressing it up in impassioned and sincere rhetoric about how everyone will die if they don't get power doesn't make it any less of a power grab.
(My proposed regime at least rounds off to opening the floor to everyone.)
Anthropic's conception of 'mitigating AI risks' is 'do everything I say, just the way I say it, and this system has to be enforced by someone who thinks like me'. It's like Microsoft saying, 'look, I don't want a monopoly, I just want to recognise that CPUs being able to run an open or foreign OS is a clear threat to public safety, and incidentally don't you think Apple bundling OS and hardware really deserves anti-monopoly legislation?'.
If your preferred regulatory regime also excludes almost all your potential competitors except the ones you're basically okay with, then it is indistinguishable from a power grab. From a customer and an enthusiast perspective I really don't care which of Anthropic or OpenAI comes out on top. I don't just want it to be technically possible to enter these markets, I want it to be easy. I want us to strain every sinew to allow new AI companies to start, I want us to give gobs of compute to anyone who isn't Anthropic or OpenAI. Let a thousand flowers bloom!
You can, but cars, airplanes and drugs are famous examples of very heavily regulated industries where creating a new business is almost impossible, incumbents rule, and incumbents and their lobbyists design the majority of the regulations themselves.
This is, in fact:
They want to be the ones deciding who gets to play with the toys and the government tamely follows their instructions as regards shutting down their rivals, but it's not going to work like that.
and Anthropic is just putting it into nice friendly words that people are swallowing. While I deeply dislike the way the Trump government handled the Fable incident, all this ultimately a fight over who gets to rule, who gets to be in and out, and who gets to set the rules. If the Trump government had calmly and consistently mandated open weight releases and anti-monopoly legislation, Anthropic would be screaming about how this was unsafe.
Regardless of what anyone here thinks, this is aimed at locking up Ukranian refugees from the press gangs and delivering them back to their government like escaped slaves.
We can take it as read that they think dying for their country isn't legitimate or noble, or they would have done it voluntarily.
Corvos
1d ago
To be fair, the ratio of fouls Argentina:England was literally 2:1. And the yellow cards were worse. England played a comparatively clean game.
England players don’t do the grandiose ‘let me hear your joy, pour champagne on me now’ routine when they score either.
Corvos
1d ago
They did bloody well, though. I though they were going to get pasted but they held on till the 80 minute mark.
Corvos
2d ago
Re: 2A there’s a pretty large ‘guns keep people safer’ contingent which is essentially what you’re describing.
Putting aside my feelings on your end destination, this doesn't work logically. It's like saying that you are looking forward to factories being entirely roboticised so right now we should enforce that workers toil 24/7 without stopping and lose all labour disputes.
The fact that people aren't machines and have feelings and rights is the point.
I was talking about the big surge in coal from 2000 to 2011 ('last few years' not very helpful), where it nearly doubles from 27k TWh to 45k TWh and then remained stable as you say. Natural gas has also heavily expanded.
FWIW your bulverism is incorrect, I live in the UK. My skepticism is that from where I'm standing in the UK these technologies are being incredibly pumped up by governments and institutions that believe buy-in is the most important factor for growth, so they're all operating on a fake-it-till-you-make-it mindset as well as subsidising like billy-o.
The EV sales figures you cited don't distinguish between true battery-powered EV vehicles and hybrids, and indeed almost nobody does. (Again, my suspicion is that this is to juice the numbers.) I therefore found it difficult to find a good source but https://www.cnbc.com/2026/05/19/hybrids-having-their-moment-as-car-buyers-rethink-electric-vehicles.html suggests that true EV sales are broadly stable (and low) while hybrids are doing well. EV sellers like BYD are certainly having a moment and are visible in the market, but are nowhere near a third of all car sales, and that's before you start looking at takeup of used vs new cars.
I am not saying that you're completely wrong, or that this isn't somewhat good news rather than bad news. I'm saying that you're presenting pretty curves and I'm not willing to take it as-read that those curves are going to curve up till they hit the stars. It may happen, it may not. Reform UK had great growth last year, beating the Conservatives out of the park, but Reform has now broadly stabilised at a level significantly lower than they need to win elections outright.
As for the resource issues, I remember during covid when people where freaking out over a lithium shortage due to EVs that turned out to never happen.
Because nobody is actually buying battery-powered EVs. Sorry, that was mean :P What did happen? I am/was expecting higher battery demand to lead to supply issues.
Corvos
3d ago
Bit of both. I do a bunch of catholic stuff and anywhere that tech comes up there's a lot of 'what do we do about Big Tech and AI' that transitions almost immediately into 'we need to regulate AI'. Then there's a vocal 'Big Tech CEOs are psychos who want to turn you into serfs, we have to destroy them NOW contingent'. It's hard to get an idea of numbers because they're so vocal.
Forgive me, but I think this perspective is broadly insane. The use of electric cars - or electric anything else - is a solution only to the extent that the electricity used comes from renewable sources. And those renewable sources need to be really renewable, not something that we can maintain for five years until there need to be mass replacements.
If you look at https://ourworldindata.org/energy-mix you see that the big story of the last few years is a massive upsurge in coal. Solar capacity may be growing but so far has managed only to propel world solar usage to the giddy heights of 2.8%. Scaling up to even 20% would produce massive issues in the manufacturing requirements (silver) and battery requirements.
In general I think there is this common problem where the battlegrounds over climate measures have been fought so intensely for so long that people have long forgotten the chains of reasoning behind them. For the forseeable future electric cars are just a more complex way of turning fossil fuels into movement, and one third of new cars being electric means diddly squat except more competition for cobalt.
The internet is made of words (spoken or verbal) and thus belongs to the people who produce words. Schizoposters produce a lot of words.
More options
Context Copy link