Assuming it cost Microsoft $0 dollars to provide their AI services (this is up there with "Assuming all of physics stops working), and every dollar they make from Copilot was pure profit, it would take Microsoft 384 years to recoup one year of investment in AI.

And thats without even getting into the fact that in reality these services are so expensive to run that every time a customer uses them its a net loss to the provider.

When Amazon started out, no one had heard of them. Everyone has heard of Microsoft. Everyone already uses Microsoft’s products. Everyone has heard about AI. It’s the only thing in tech that anyone is talking about. It’s hard to see how they could be doing more to market this. Same story with OpenAI, Facebook, Google, basically every player in this space.

Even if they can solve the efficiency problems to the point where they can actually make a profit off of these things, there just isn’t enough interest. AI does plenty of things that are useful, but nothing that’s truly vital, and it needs to be vital to have any hope of making back the money that’s gone into it.

At present, there simply is not a path to profitability that doesn’t rely on unicorn farts and pixie dust.

Well, yeah, that’s fair

Yes, but the basic problem doesn’t change; you’re spending billions to make millions. And Deepseek’s approach only works because they’re able to essentially distill the output of less efficient models like Llama and GPT. So they haven’t actually solved the underlying technical issues, they’ve just found a way to break into the industry as a smaller player.

At the end of the day, the problem is not that you can’t ever make something useful with transformer models; it’s that you cannot make that useful thing in a way that is cost effective. That’s especially a problem if you expect big companies like Microsoft or OpenAI to continue to offer these services at an affordable price. Yes, Copilot can help you code, but that’s worth Jack shit if the only way for Microsoft to recoup their investment is by charging $200 a month for it.

Two year old article. Might be good to note that in the title.

Testing shows that current models hallucinate more than previous ones. OpenAI rebeadged ChatGPT 5 to 4.5 because the gains were so meagre that they couldn’t get away with pretending it was a serious leap forward. “Reasoning” sucks; the model just leaps to a conclusion as usual then makes up steps that sound like they lead to that conclusion; in many cases the steps and the conclusion don’t match, and because the effect is achieved by running the model multiple times the cost is astronomical. So far just about every negative prediction in this article has come true, and every “hope for the future” has fizzled utterly.

Are there minor improvements in some areas? Yeah, sure. But you have to keep in mind the big picture that this article is painting; the economics of LLMs do not work if you’re getting incremental improvements at exponential costs. It was supposed to be the exact opposite; LLMs were pitched to investors as a “hyperscaling” technology that was going to rapidly accelerate in utility and capability until it hit escape velocity and became true AGI. Everything was supposed to get more, not less, efficient.

The current state of AI is not cost effective. Microsoft (just to pick on one example) is making somewhere in the region of a few tens of millions a year off of copilot (revenue, not profit), on an investment of tens of billions a year. That simply does not work. The only way for that to work is not only for the rate of progress to be accelerating, but for the rate of accelleration to be accelerating. We’re nowhere near close to that.

The crash is coming, not because LLMs cannot ever be improved, but because it’s becoming increasingly clear that there is no avenue for LLMs to be efficiently improved.

Port que no los dos?

Given how good a job they did with 40K, I’m confident.

You guys get YouTube ads?

Smart Tube Next

Newpipe / Pipepipe

Firefox + Ublock Origin

Credit for actually using rhyming slang correctly.

I can’t count the number of times I see characters saying shit like “'Aving a bath” or “Gotta get back to the ol’ trouble and strife.” The whole fucking point is that you say the part that doesn’t rhyme, while using the whole phrase as a mnemonic for the thing that it rhymes with. It’s a code language designed for discussing crimes. It’s not meant to be easily intelligible. If you’re saying the rhyming part it becomes much too easy to work out.

The universal translator is powered entirely by plot and absurdity. Hence why everything Quark says comes out in English, but anything said in Klingon is left untranslated. As for slang, again a definite “maybe” based on what’s established in Darmok.

Basically, don’t think about it too hard, none of it makes sense.

I agree that that would be a real danger, yes.

About an hour before dinnertime our dog gets super excitable. She starts to stare at us constantly, and if we get up to to do anything at all she goes apeshit with excitement because OH MY GOD IS IT FINALLY TIME?!?!

They’re also doing a store cosmetic with some very clear Judge Dredd stylings, and totally-not-a-lawgiver, and honestly, I’m going to have a hard time saying no to that.

I really enjoy how they’ve managed to do a cop themed pack that leans hard into the “We are not the good guys” premise of the game.

The key detail is that, like with rear brake lights, they extinguish when the foot is removed from the brake pedal. So it’s not so much the presence of the brake light, but the presence of an inactive brake light that would, serve as a warning that a car is about to start moving. This would be very helpful to drivers on a road when other drivers are pulling out too early from a side road or driveway. That little bit of extra warning is, in many situations, enough for you to pump the brakes, hit the horn, or both.

And they’re cousins, so this is gonna be really awkward at the next family dinner.

I mean, the workers weren’t exactly getting paid well. I guess a paycheck is better than no paycheck, but they weren’t building a socialist utopia here.

Airsonic or Navidrome server, plus one of the various Subsonic apps. I like Substreamer or Dsub but there are plenty of other options.

Bro, you started this fight. You don’t get to complain when they hit back.

What I find darkly fascinating about this “eat the cost” strategy is that if you were to dare suggest taxing Walmart more, Trump’s fans would call you a commie librul socialist who just wants handouts. But when Trump demands that they subsidize his agenda suddenly they’re all for it.

“But doctor, he says, I am Dorian!”