Post

No you missed the point. The math for those companies you're citing doesn't apply here. Inference isn't cheap, and the margins aren't that high. Training sure as fuck isn't. And both require *massive* infrastructure build out. It's not enough to just get some customers. That's nothing.

this is great, thanks

oh yes, this is an ongoing bit

Mathematical honesty >>>> intellectual honesty

The AI industry tried to use a classic startup growth trick — subsidize, scale, and when you reach a critical mass, crank up the prices. Except the services are too expensive, and the costs too far removed from the cost of subscription. www.wheresyoured.at/ais-economics-dont-make-sense/

A colleague of mine has been pushing to get GitHub for months. They finally got it approved last week. Perfect timing, dipshit.

Which is why they are all scrabbling for that government money…

"at least the Radium Ointment works great"

The door is over there.

Please don't put on the newspapers that our investors are being scammed. We're not scamming our investors. Stop saying we're

Well, the first scam is to call a language model intelligent. That's the first sin.

Loss leading is a well recognised scam. This is not a temporary discount or trial, it is giving a false and misleading impression that the product is cheap to the customer.

yall I think it's in their best interests to not understand scams

Drug dealers normally give their marks a better deal in the first few months. Until they're hooked, and then... (Yes, people do get hooked on AI)

to the market, investors, debt holders, etc: a scam to individual customers: a deliberate trap

Just a Q - why would I need a £5/month subscription to a glorified autocorrect to tell me I need to smoke weed and not vax to cure my tumor when I can go talk to my dipshit aunt's Facebook page to get the same advice for free?

Maybe the scam against investors seems as if not a lot of sympathy needs to be had if we visualise them entirely as arrogant venture capitalists, but the moment they manage to sell some of the bags off to e.g. pension funds, things change. And even in the first case, it's still a scam.

NFTs, Sam Backman-Fried, a million other crypto startups, Metaverse. There's so many recent examples even.

> AI-powered natural healthcare backed by the latest science. lmao eat my ass chatgpt says its healthy

The first hit is free.

China has a UV Lithrography machine prototype. Its not gonna be long before real competition happens. And no matter how many tarriffs the US puts on 'em, China will win that fight.

Loss leading is absolutely a scam. Misrepresentation of a discounted or trial period price as the actual price is misleading or deceptive to the consumer and unfair to competitors. Esp where you don't have inventory/capacity to serve at that price.

No one needs these local models running on Mac minis. This is akin to home brewing fads of the 70s. People get crazy with the dicking around. But ultimately it's handier to just buy a few cans in the shop.

those models are unnecessary too, but simply being unnecessary isn't fatal and Apple doesn't ever have to make a good product, or any product, ever again: it must have about a trillion in cash by now? what pisses me off is they keep making actively terrible, customer hostile products

Just curious on your take on this, or if you have addressed it and I just missed it.

Ed charging by the word with a profanity multiplier is the true Pale Horseman, world economy in tatters :)

Excellent reporting! I work at a uni that has Gemini enterprise account. All the AI companies are trying to get locked in with education. How do you think this plays out for schools? Will they also see increases? I am not sure how many ppl in uni use it. Students prob most. Profs not as much

It feels just like the dot com bubble. Most companies failed and the 'net was a nerd's toy for awhile. But, there were a few companies that hung on and figured out how to make it work (Amazon). I'm wondering which company will survive the bubble bursting and finds out how to make AI profitable. 🤔

This is a banger, Ed: “LLMs are constantly humming with the sense that they're about to do something new, even though they're mathematically restricted to repeating other actions.”