Remarkable how most of the major companies have the same AI business model, which is "how can I pay the least amount of money to humans to create an app that does a bad job replacing humans?"
God forbid they pay a bunch of humans to build a good product that augments what other humans are doing.
https://gizmodo.com/github-copilot-ai-microsoft-openai-chatgpt-1850915549
@maxkennerly But that’s not the problem the article identifies. Instead, it’s saying that companies spend too much on IT infrastructure, and can’t make it back with the usage fees they’re charging. The first example the article gives is a product that customers love, but, it says the per-month subscription fee isn’t enough to keep the infrastructure running.
@rlpaulprodn If the AI products were better—such as by using hand-curated training data and using experts for reinforcement learning—the companies would have a better product and thus be able to charge more and/or attract more users.
Instead, they try to be cheap on labor.
@maxkennerly @rlpaulprodn It's Microsoft. They can charge whatever the hell they want. Moreover, they own GitHub, which is a wealth of training data, at their fingertips.
It's not cheap because they're trying to be "cheap on labor." It's cheap because they want to get developers hooked early so they can capture the market. The prices will go up once they have the market share they want.
