bubble-predictions.md

AI is very definitely a bubble. to make the most basic economic argument for that position, no one who's in the AI game is making money, and on average they are down millions or billions per year. free and free-ish AI services available now, such as ChatGPT, Copilot, and Gemini are being operated at catastrophic loss, and while they might be able to capture users, no one has actually figured out how to convert those users into paying customers in large enough volumes to recuperate the immense costs of running millions of GPUs to generate glue pizza recipes.

here's my prediction:

1. over the next couple of months (possibly years, but betting months -- no one has that much cash), major AI services are going to stop offering the bait and severely choke up on how much they're willing to offer for free.
2. as a result, a lot of "innovators" in the space turn out to just be wrappers around ChatGPT et al and basically shutter entirely.
3. in terms of actual humans, the vast majority of people won't (or can't) pay the new and exorbitant costs -- they'll simply leave. the bubble might pop suddenly, or deflate slowly, but the economics just don't allow for anyone but the largest of whales to subscribe.

i can take that and make a persuasive argument against using AI services -- you can't count on them to remain free or cheap, they are just too expensive to run. if you're making images, you are going to have your metaphorical paintbrush yanked from your paws and have to pay a fat subscription fee to get it back; if you're trying to compose books, your spellcheck-on-steroids is going to be locked behind a paywall, and so on and so forth.