[bende511]

The things LLMs can actually be put to good use for either are bad for society (spam, scams, cheating on homework) or do not justify the massive investment. Better emails is great, but its just not worth billions of dollars.

I want to be clear, from an academic perspective these things are fascinating and very very cool! You can teach a computer to talk like a person might simply by having it read all the books! But there is not a business reason to have this. Not at the current costs. Maybe LLMs are a step we need to get actually useful AI tools, but as currently constituted its just not there.

The killer app of LLMs is how its gonna kill the industry when the investments fail to make the expected return. A lot of people are gonna lose their jobs when managers try to replace them with AI, and even more are gonna lose their jobs when that doesn't work and all those companies go under, and there's another tech bubble burst

[vonneumannstan]

>Better emails is great, but its just not worth billions of dollars.

I mean lets just say your premise is right, how much time is spent worldwide writing emails that could be automated? Just on a pure man hour basis it's surely worth billions if it did nothing else.

>But there is not a business reason to have this. Not at the current costs.

ChatGPT Pro tier is $200/mo, if you can't find ways to get that value I'm sorry but it's a skill issue and you're ngmi. The API tier is like $/million tokens. It's a joke to complain about the cost at this point...