@SebAaltonen No one's asking for a "woke" label, because the central complaint there is that the product is bad. Generative AI is ethically controversial and legally unsettled, because it's running other people's IP through a certain class of computer program and saying the result is your IP.
@NOTimothyLottes So much of this bubble is round-tripping. They have to keep increasing the amount of money going in a circle, and they're running out of plausible hardware transactions. One more sign it's on the verge of popping.
@NOTimothyLottes@noop_dev AI technology is interesting and useful, but the bigger effect we're seeing isn't from the technology itself, but the way it's being treated as an excuse to disregard copyright, and that's a genie that can be put right back in the bottle at any moment.
@NOTimothyLottes@noop_dev I think the AI stock bubble pops soon, and suspect that with it will go the political support that has prevented both generative models and their outputs being treated legally as derivative works of the training data and other inputs.
@NOTimothyLottes@mikeev@tramstalker It's not just AI, you see it with do-it-all engines, frameworks, and scripting languages that let you slop a lot of libraries together. They don't save nearly as much time as people think.
@NOTimothyLottes@mikeev@tramstalker I think people are fooled by systems that make unmarketable prototypes or demos easy. If you get to that point starting from basic APIs and utility libraries, you'll have learned a lot about how you'll do a polished product. With enough shortcuts, you haven't really started.
@NOTimothyLottes Cost was like:
- $200,000+ if API tokens
- $1000+ if subscriptions
- guy actually didn't have to pay for it himself and doesn't know what it cost
@NOTimothyLottes I was looking into the cost of this, and it's wild that the AI companies are cultivating developer dependence with subscription prices ~0.5% the per-token prices. There's no way those subscriptions are here to stay.
@itemgatestudios@notch A large majority of released games simply aren't fun, and people don't want them even for free. Overlooked gems exist, but most devs that got pre-release wishlists but not sales after release didn't deliver what they advertised.
@itemgatestudios@notch Nobody wants to believe they wasted their time and money making a bad game, but that's what most of the industry is doing most of the time.
@itemgatestudios@notch I don't think it's a matter of forgetting. Most games released just aren't good enough to be worth playing, and this is easier to notice when they're finished. It's far easier to get people's hopes up during development than to satisfy them on release.
@iquilezles Yeah, it's a time when working out a painstaking manual process can be equivalent to developing an algorithm. I'd be inclined to credit you as the main originator of this lossy compression method.
@NOTimothyLottes I don't have any answers for that. The medical science is far above the typical medical practice, and with their regulatory capture, there's not much incentive to improve.
@NOTimothyLottes I felt a need to try to correct my own ignorant statement here, in case anyone reads it: potassium and magnesium deficiencies can have a self-perpetuating tendency, and often can't be corrected by safe dietary intake, only aggressively enough to need medical supervision.
@NOTimothyLottes I think if you're taking care of it yourself, you pretty much have to stick to an intake that'd be safe even if your levels were high, and be patient. People have been recovering from deficiencies with ordinary food for a million years.
@SebAaltonen Same reason they've been doing it so far: to maintain brand loyalty in the consumer market so they don't go out of business when the AI bubble pops and the server market's glutted with used hardware for years.
@NOTimothyLottes I still wonder if after the AI bubble pops, courts will be suddenly all notice that both an AI model and all its outputs are derivative works of all its training data and anything it got fed or searched up as an agent.
@SebAaltonen@KuittinenPetri Using AI isn't a deep skillset, you can start from zero and catch up in a few months, and then it's all outdated again shortly. But if you never learn to code without AI, you'll never be competent to call the AI on its bullshit.