←back to thread

165 points atombender | 9 comments | | HN request time: 0.658s | source | bottom
1. andoando ◴[] No.45151023[source]
Jesus fuck were spending hundreds of billions in a few years on AI? Slow the fuck down. Lets establish some proof of value first
replies(4): >>45151098 #>>45151116 #>>45151143 #>>45152091 #
2. aprilthird2021 ◴[] No.45151098[source]
Pretty sure most of the Mag 7 have on paper already north of $100B in AI infra spend this upcoming year.
3. jeswin ◴[] No.45151116[source]
> Lets establish some proof of value first

Recently spoke to someone who said that most startups in their (very large) portfolio are using AI to write most of their code. I personally know this to be true in several startups I've seen; and I've seen slow but steady adoption in larger companies.

We can debate about spending hundreds of billions per year, but the value is beyond question at this point at least in software.

replies(2): >>45151151 #>>45151190 #
4. apparent ◴[] No.45151143[source]
I think he was talking about domestic spend in general, which would include AI and everything else Meta does. Also, he probably just picked a number that would be in the proper order of magnitude, so that when it turns out to be incorrect, it is not wildly so.
replies(1): >>45151177 #
5. Eldt ◴[] No.45151151[source]
I find this hard to believe given the quality of code that AI models currently produce. Does the startup have a successful product? Or does it have a spaghetti ball of technical debt waiting to explode?
6. cmiles74 ◴[] No.45151177[source]
He repeated Tim Cook’s number, probably he was nervous and it was top-of-mind.
7. dustrider ◴[] No.45151190[source]
Just to make the point. I agree there’s adoption but value is still being figured out.

See the MIT Nanda study, and the other one from a few weeks back on the perceived vs actual productivity increases.

There is value, but so far nowhere near as much as the people pushing AI would like you to believe has actually been delivered

replies(1): >>45151242 #
8. CamperBob2 ◴[] No.45151242{3}[source]
The Nanda study is junk. Yes, I'm slower as a developer when I use AI. That's because I'm doing things I simply couldn't do before.
9. Draiken ◴[] No.45152091[source]
AI will win because mediocrity is enough to generate profits and all that's all we care about.

I hate it but I don't see how we can stop it.