The full automation of the economy won’t be for a century or more, and as manual labor dries up there will be new work in robotics technician fields and AI development and training.
I’m not saying we don’t need to prepare with taxation and UBI, but the jobs aren’t just going to disappear overnight, that’s ridiculous.
You didn’t even bother reading it. This isn’t counting investment, it’s straight up losing money right now, not counting investment. Microsoft’s customer model charges $10 a month/user to use it and it’s turned out to cost $30 a month/user. The other big firms are seeing similar costs.
These LLM require huge amounts of processing, then when your users are spending resources to do very simple tasks, which is basically all the models are useful for right now, it costs a stupid amount of money to do stupid things.
This is not to say that it cant be useful in the future, or smaller purpose built models can’t be useful. But these vast generic models literally hemorrhage money as it stands.
Some features are meant to drive engagement, not revenue. Some products are sold at a loss to drive engagement. This article is a very simplistic view of how technology has always worked for a product in it’s infancy.
Ah yes, the trend of every new technological development ever. Apparently investment is “hemorrhaged money” until it’s profitable.
And then it will only be profitable for a time until the people have no money to spend cuz no jobs.
The full automation of the economy won’t be for a century or more, and as manual labor dries up there will be new work in robotics technician fields and AI development and training.
I’m not saying we don’t need to prepare with taxation and UBI, but the jobs aren’t just going to disappear overnight, that’s ridiculous.
You didn’t even bother reading it. This isn’t counting investment, it’s straight up losing money right now, not counting investment. Microsoft’s customer model charges $10 a month/user to use it and it’s turned out to cost $30 a month/user. The other big firms are seeing similar costs.
These LLM require huge amounts of processing, then when your users are spending resources to do very simple tasks, which is basically all the models are useful for right now, it costs a stupid amount of money to do stupid things.
This is not to say that it cant be useful in the future, or smaller purpose built models can’t be useful. But these vast generic models literally hemorrhage money as it stands.
Some features are meant to drive engagement, not revenue. Some products are sold at a loss to drive engagement. This article is a very simplistic view of how technology has always worked for a product in it’s infancy.