Whether the hardware to run it is on-prem at a large company, on local machines or in data centres, it is still a big increase in the demand for that hardware, and that costs money. Your argument doesn’t work, because it relates two unrelated things:

1. how energy intensive AI is compared to other activities
2. how hardware intensive AI is compared to current hardware usage

It should be clear that one can be high while the other is low.

The rest of your comment became a highly passionate rant instead of a reply to what I said about energy usage, which I will chalk up as yet more evidence that

people hate it with an irrational passion. So they criticise all aspects of it, whether reasonably or not.

I mean both vets suggested it for a reason, and I agreed - but notably the cat did not agree to it either time.

One of my cats hates the vet because twice at the vets the vet went and stuck a finger up her arse, and you know, I think that’s fair enough.

Oh, I don’t think it should be opposed, I just think it should be viewed with deep suspicion.

Only if they have parts.

I believe Apple had the rainbow beachball of death ages ago

How can you live with yourself now?!

I didn’t need to read that >:(

I think at this point it’s speculation. Your speculation is a reasonable one. But with Trump, I think there’s another very reasonable one: he’s a natural fascist so they’re just trying to buy a future unspecified favour. Maybe it’ll be to avoid having to give up a bigger stake in the future, or to avoid a big tax, but they probably don’t know exactly. I’d compare it to all the techbros donating to Trump’s campaign and turning up to his inauguration - they don’t know exactly what they’re getting out of that deal, but they know they’d rather pay homage to the king than be on his shit list.

So to me this is evidence of corruption, even without a known deal having taken place. They know that Trump plays favourites with the enormous power he wields, so they want to be the favourite.

I literally described a situation in which AI has a practical use. If you’re going to disagree, fine - but at least acknowledge where the disagreement is instead of replying without (apparently) having read my comment.

I believe I read somewhere that AI capacity and hence potential usage is going up about 12% per year. That could turn out to be a lot but it would have to go on for a long time to end up significant compared to the actually large polluters.

AI energy usage for a single person is (very roughly) comparable to having the TV on or playing a video game on a high spec PC. Even if the only benefit it gives anyone is the same mild pleasure you get from watching TV, we accept that using energy for these purposes is worthwhile. So, despite this topic recurring constantly, no-one has given a sensible reason why we should particularly call out AI for its energy usage.

Of course, there’s a readily available explanation for why people do it: people hate it with an irrational passion. So they criticise all aspects of it, whether reasonably or not.

The contribution of AI to greenhouse emissions is tiny compared to the actually significant contributors - like transportation and rearing animals.

And with that 90% of people have stopped reading and thing I’m an AI booster so congrats for still being here.

I also think the utility isn’t going to be for techbros primarily. Software developers can already do the things that LLMs are good at, and are better than LLMs in important areas. But think about writing a 500-line script to extract data from a database and do some simple data analysis on it, then displaying it. Software developers, again, can do that easily (it’d probably take a few hours - more if it’s more complex or if they polish it).

But most people cannot. Yet there are a lot of situations in business or in government which could benefit from being able to say, “show me the average of (X) over the last 10 years, stratified by variables A, B, C.” And AI will do write a script to do it (so the actual computation is not done with unreliable bullshit) with ease. The script will have way too many comments, may well be laid out in a nonsensical way, and may reinvent the wheel stupidly, but in my limited experience of asking the LLM we have at work to do things like this for giggles and because I’m not paying for tokens, it will do it correctly.

I’m not saying to buy the hype, but if you think it’s useless for everything but “making it easier for some techbros” you’re just as immune to evidence as the hypesters are.

This is true but you also have to ask what OpenAI is getting out of this deal. They’re not just going to give away stock for nothing, and the US government doesn’t have a policy of helping itself to portions of companies.

So OpenAI thinks it’s getting something for this. Whatever that is, it’s bribing the government to get.

Any question that starts “which language is easier” depends on what languages you already know well. You don’t specify, other than to write your question in English.

Yeah. Whenever we try to sleep in past 5AM, one of our cats freaks out 😭

Uhhh possibly? I think it’s clearer with Shen because the open eye is just a circle like a letter “o”. Here’s an example of Hobbes squinting

you can see it’s similar, and presumably not inspired by the symbols, so there’s a precedent :)

Oh my god, hours? How will you live.

Anyway, I have now.

A new press regulation that required equal-prominence corrections would be welcome. I can imagine some issues with it (you’d need a lot of corrections for live unscripted programs!) but let’s assume they’re manageable. Then, what need is there for this trustworthiness categorisation? Any outlet that doesn’t abide by those rules is getting fined into oblivion.

Anyway, the law handles false commercial speech all the time. So why is this any different?

Because false commercial speech is typically only handled when it causes damages, which can then be estimated and sued for. How do you estimate the damages attributable to one paper printing a false rumour, which it said was printed in good faith and corrected at the first opportunity, but which may have led to riot? Maybe the riot would have happened anyway. Maybe it wouldn’t, but it also wouldn’t have happened if the rumour hadn’t been circulating on Facebook.

How do you estimate the damages attributable to the BBC for allowing Boris Johnson a platform to lie about the benefits of Brexit? To what extent did that contribute to Brexit happening? How much has Brexit cost the country? (There are estimates, but with a huge degree of uncertainty)

Commercial speech leads to identifiable decisions by definite parties through more or less identifiable processes. That is completely different when the decisions are taken by the millions of inhabitants of the country. You could attempt to apply the same process, but it would be an absolute quagmire and would definitely lead to wrong decisions due to the massive level of uncertainty inherent in doing that.

Canary is a shit rag that mainly prints polemic, not fact (just like Novara) but that doesn’t mean they should be cut off from services like banking.

So you know, this first surfaced years ago

The symbol is used in emoticons like this: >_o to mimic the wrinkles that show up when you’re scrunching your eye closed. Nerd artists use the same symbology in comics.