First, 'climate' and 'weather' are two very different things. Second, we can't control climate, we can only affect it in very specific ways. Pumping more CO2 into the atmosphere is easy, getting it out is hard (likely impossible).

And we had a very very clear target all along - limit the global temperture increase to no more than 1.5 C above pre-industrial temperatures. We missed that target.

I don't think it's just a question of agencies and regulations. The reason why capitalism works, or maybe used to work, was that it rewarded producing more/better goods and services with profit. So people who started businesses became rich but in turn the wider society got more/better goods and services, so at least in purely material sense the society benefitted.

However now this very basic relationship seems to be breaking down, and I think in part that's due to there being too many MBAs and them becoming too powerful. When you're trained to see all goods and services as interchangeable you're simply not going to have the necessary expertise in your company's specific domain to improve the products your company is producing or find better/more efficient ways to produce those products. There will only be a few things you'll be able to do to increase profits.

One, cut corners which will necessarily make the products / services your company makes worse. Two, market and hype the product to people who aren't experts in that area (because the experts will see through the bullshit). That doesn't necessarily make the product worse by itself but it encourages a mentality where it doesn't matter if a change actually improves the product so long as it makes it more marketable, which obviously has great potential to make the product worse. This is especially bad because, again, the MBAs in charge may have a hard time understanding a highly technical improvement to the product but they'll easily understand a change that makes it more marketable. Three, come up with more elaborate and possibly manipulative ways to sell the same products for more money. That again is to the detriment of the wider society - people are getting the same stuff but for more money or just more hassle.

You can see it especially well in tech and some other industries where often when a generic rent-a-CEO replaces someone who rose through the ranks, company stops producing anything innovative, except stuff started under the predecessor and starts agressively enshittifying.

What's making it much worse, and I don't understand the reasons behind this very well, is that businesses now seem to be more concerned with getting investment money rather than actually selling their products to the customers. This is obviously to the detriment of the wider society - when a product is designed to impress people who probably won't be using it, likely other MBAs, it's likely to become worse for people who will be using it. That's assuming there will ever be any sort of product once the investment money is secured.

Now I'm not advocating for drowning all MBAs at the deepest point of the ocean or anything like that. Marketing and financial expertise is important to a business, and of course you have companies offering actual financial services, which is where MBAs do have domain expertise. But they shouldn't always have the most powerful roles in the company, they should have advisory roles rather than be in charge. Then we could maybe get back to the model where making better products or services is what gets you the most profit.

This is not a defence of capitalism, just of US. China is capitalist, in some ways more so than US - they don't have many worker and environmental protections that US does (for now). So is Russia. You haven't mentioned Japan or South Korea - they're also capitalist of course. I believe the only country in the world that doesn't practice some form of capitalism is North Korea.

As things stand 'reliable' is not a word you associate with AI. This being the case, you either need to get people to check the AI output, negating most if not all of the benefit from using it in the first place or accept that what AI produces may simply be not that good - far from ideal in either case. Given the insane amount of money invested, adoption by businesses has to be overwhelming. My impression is that this is very far from being the case. Also some of the organisations that are currently using AI on a test basis may decide that it hasn't proven that useful and stop using it.

That's what I was talking about. You need smaller models specialised for the specific scientific/engineering problem they'll be used for. The users would most likely need to be able to train the models themselves on their own data. Used in that way AI would be actually useful. The current direction though is for giant models that are supposed to be able to solve any problem in any domain because they're 'superintelligent' or some such nonsense, including domains where AI should absolutely not be used at all.

The distinction between a commodity or tech bubble is interesting but in both cases you can have a tech or commodity which is definitely useful but whose usefulness is being overestimated, and tech or commodity which has very limited utility and whose current value is mostly based on hype. Bitcoin and especially NFT are good examples of the latter type.

How much computing you get out of a token is surely a purely business decision by the company selling the tokens. Theoretically there's nothing stopping them from making a single token give you unlimited lifetime access to their AI - probably wouldn't make much business sense but I don't know how much sense any of this is making atm. Tokens giving you more bang for the buck may mean the tech is improving or there's more infrastructure or it may just be due to a business strategy this particular company is pursuing.

'Relativating' is not a word, 'relativizing' is. You can't just use words you've made up, it's like raping the English language (teehee).

Equating corporate espionage to treason is a bit riduculous. OpenAI is a private company. Many have suggested that US is an olygarchy and corporatocracy but surely it's not so bad that stealing company IP and other TOS violations is punished by execution.

I think the major difference here is that all those things had and still have undeniable utility. Of course, even the most useful things have limits to their usefulness - there's places where building railroad doesn't make much business sense, there's businesses which are never going to be viable even when going online, or businesses which aren't going to benefit that much by going online etc and that's why the related bubbles became bubbles . Not sure there can be such a thing as an 80 year bubble. But anyway, railways, automobiles and the Internet are all unquestionably massively useful.

AI otoh, not so sure. AI in general is a useful technology in certain specialised areas, but the AI bubble has been centered around LLMs. And I don't see LLMs as being actually that useful for much at all. Business case for railways was transporting huge amount of goods over long distances - from the start railways delivered this reliably and continue doing so. Business case for the Internet was access to information and doing business transactions anywhere in the world regardless of physical distance - the Internet has delivered that from the start, and much more. Business case for cars was personal transportation over long distances - even the earliest automobiles delivered that reliably. Business case for LLMs was replacing a substantial percentage of white collar workers with AI - LLMs have failed to deliver this, except for a very limited number of most junior workers who were paid peanuts anyway. You don't invest literal trillions into something that maybe sort of helps existing workers do their job - and people here who work with AI will attest that it may not even do that at all.

Maybe a more better analogy is the Tulip Mania in 17th century Netherlands. Tulips are a pretty flower but at the end of the day they aren't anything more than that. Likewise, LLMs can be a fun toy but at the end of the day they aren't much more than that. Maybe if the direction of the AI hype could be diverted from going all in on LLMs to focusing on more specialised AIs it could yet prove to be a useful technology but I see very little signs of that happening.

The amount of lies in news is relative. Major players could take things out of context and distort the nature of the story, manipulate statistics, report what happened in a very selective fashion, rely on dubious sources etc but they're not likely to just straight up make stuff up from thin air, at least not too often. The only alternative to that is getting your news from Facebook or some other social media and there all bets are completely off.

To start with this is a rather ridiculous strawman. 'You don't like people complaining about comedians, you want to bring the minstrel shows next?'. No, those are different things. But even with minstrel shows, I know today's wokes see them as some kind of equivalent of murder and cannibalism on live TV, but they were simply light entertainment that expressed views that, at the time, a large part of the US population held. Times changed of course, a few decades later most people would find those views extremely stupid (not to mention they'd find the humour really cringy), so no network would air such a show.

Any expression of an opinion potentially perpetuates the views that the person expressing the opinion holds. If you say that perpetuating those views leads to some unspecified "harm" and thus should be censored, then potentially any opinion can be censored. The only difference is, which group has sufficient power to enforce censorship against those who disagree with them.

This isn't new at all. For most of world's history you would be subject to some kind of punishment (up to and including execution) for insulting the dignity of the king or some powerful nobleman, saying things that are contrary to common public morality, blasphemy, stating that Earth revolves around the Sun etc etc. Even today in many countries you have such laws. Then, as the principles of liberalism became popular, we've moved to a different model, where if you wanted others to believe that what you're saying is true you'd have to justify it and persuade others that you're right, rather than silencing those who disagreed with you. Hopefully I don't have to explain how liberalism benefitted us.

So I find it very alarming that now it seems those principles are being undermined from all directions. People are trying to tear down what we've been building up for centuries. Most countries with a history of liberalism still have their constitutions and many of the laws on the books based on those principles, but socially liberalism seems to be mostly dead - very few people seem to believe in or even understand its most basic principles. There's only right-wing and left-wing authoritarians now.

It seems like the approach they used in training this AI could be ported to other games, possibly even to real-world problems. This I believe is one of the situations where AI technology could actually be useful, a specialised domain with a limited number of parameters. Not an AI trained to put words together with zero understanding of what any of them mean.

This doesn't contradict the OP's point in any way. The fact that AI came up with something that doesn't correspont to reality is not the point, it is well known that AI does that. That's not the concerning thing here. The fact that the US military was prepared to take a hostile military action because it believed an AI hallucination is the concerning thing here. When you're making decisions that can literally cost a lot of people their lives, you cannot include a so-called 'intelligence' that doesn't understand anything it's saying in the decision making process. And maybe there was no chance of an actual war but boarding other nation's ships for absolutely no reason can damage diplomatic relations, not just with China but with any other nation that has ships in the region that could potentially be boarded on such an unreasonable basis. Assuming this is something the current administration wants to avoid.

Given that it was certainly an excuse for rather than a reason for that war, no, it wouldn't, but that has nothing to do with AI.

That's one (and the most charitable) interpretation of what happened. It may have all been a part of the process, or someone may have intervened at the last minute to stop this.

As in, we're not sure if this building really has terrorists/enemy combatants in it, but let's bomb it a bit anyway and then later we'll figure out if the intelligence was good? That would explain some past incidents. Even in this case, what if they started the interception/boarding before getting the confirmation? What if they weren't even waiting for confirmation and were preparing to board before the abort order came through? And you might say, they're the US military, they wouldn't fuck up like that. Given that they're using ChatGPT in making military decisions, the kind of thing that gets lawyers held in contempt of court, I'm not so sure. If you're making decisions based on something as untrustworthy as AI, which you know may well have made the whole thing up, then one, you absolutely shouldn't do that, but two, you should at the very least always double-check before taking any action.

It's extremely naive to think Anthropic wants the slowdown of AI research out of some sort of social conscience. To begin with, AI companies have shown time and again that they don't have any. This is a desperate attempt to do two things. First, to try and freeze the AI landscape while their models still have some advantage over the competitors, mainly to stop the Chinese from doing the Chinese thing of reverse-engineering Western technology and then trying to improve on it or at least selling cheaper versions. Second, and maybe more important, is to sell the idea that AI continues to significantly improve and will continue to improve at the same high rate, to the point where it eventually becomes dangerous. In reality any actual recent technological improvements have been very small and incremental and most AI improvements came either from people figuring out how to use the existing technology better or from the same technology being deployed at greater scale.

If the goal is to stop AI from doing evil stuff, then theoretically you would need more AI research into how to put better guardrails into AI. Except by now it's become extremely obvious that people will always find ways round any guardrails. Also if the supposed goal is to stop AI from hacking websites, then stopping or slowing down research would be shutting the stable door long after the horse has bolted. We know quite well that most website code on the internet is chock full of vulnerabilities and does not require any sort of superhuman intelligence to hack. AI with its current capabilities is already more than enough for the task. Also, I understand that AI agents doing similar things to what's described in this story are increasingly being used as white hat hackers, in order to find vulnerabilities and fix them.

Yes, they know that in the minds of the people they're trying to sell it to, i.e. upper management all the warnings about the AI being dangerous will translate to 'it's powerful and can make us a lot of money'. At the same time it may make them sound caring and responsible to the larger populace, assuming anyone believes them, but particularly gullible people still might. Also, they can say that the progress of their AI stalled because they're trying to avoid it's dangers and totally not because they don't know how to improve it any more.

I'm not sure how much trust in the tech industry there was to begin with. After continued enshittification, numerous user data mishandling scandals, outright psychological manipulation of people in order to increase engagement, pushing ads into everything etc etc it was already pretty obvious that the main goal of a lot of tech industry is not to make better tech. Plus they've already tried to create similar hype based bubbles with cryptocurrency and NFTs (which are also technologies that have some uses in very limited number of situations but which they claimed were going to change everything). But certainly what trust still remained has been damaged.

Any additional jobs gained in construction of datacenters have to be temporary. Once you're done constructing the datacenter, your job is finished. Unless you assume that the AI bubble keeps growing indefinitely and new datacenters will continuously keep getting built, but even in that case at some point you'll run out of land.