4 ms·

I’m watching some of this happening first and second hand, and have seen a lot of evidence of companies spending a ton of money on these, spinning up department

I’m watching some of this happening first and second hand, and have seen a lot of evidence of companies spending a ton of money on these, spinning up departments, buying companies, pivoting their entire company’s strategy to AI, et c, and zero of its meaningfully replacing employees. It takes very skilled people to use LLMs well, and the companies trying to turn 5 positions into 2 aren’t paying enough to reliably get and keep two people who are good at it. I’ve seen it be a minor productivity boost, and not much more.

hnthrowaway6543 2y ago
> and the companies trying to turn 5 positions into 2 aren’t paying enough to reliably get and keep two people who are good at it. it's turning 5 positions into 7: 5 people to do what currently needs to get done, 2 to try to replace those 5 with AI and failing for several years.
- vundercind 2y ago
  I mean, yes, that is in practice what I’m seeing so far. A lot of spending, and if they’re lucky productivity doesn’t drop. Best case I’ve seen so far is that it’s a useful tool that gives a small boost, but even for that a lot of folks are so bad at using them that it’s not helping. The situation now is kinda like back when it was possible to be “good at Google” and lots of people, including in tech, weren’t. It’s possible to be good at LLMs, and not a lot of people are.
Vegenoid 2y ago
Yes. The people who can use these tools to dramatically increase their capabilities and output without a significant drop in quality were already great engineers for which there was more demand than supply. That isn't going to change soon.
- vundercind 2y ago
  Ditto for other use cases, like writer and editor. There are a ton of people doing that work whom I don’t think are ever going to figure out how to use LLMs well. Like, 90% of them. And LLMs are nowhere near making the rest so much better that they can make up for that. They’re ok for Tom the Section Manager to hack together a department newsletter nobody reads, though, even if Tom is bad at using LLMs. They’re decent at things that don’t need to be any good because they didn’t need to exist in the first place, lol.
  - TeMPOraL 2y ago
    I disagree. By far, most of the code is created by perpetually replaced fresh juniors churning out garbage. Similarly, most of the writing is low-quality marketing copy churned out by low-paid people who may or may not have "marketing" in their job title. Nah, if the last 10-20 years demonstrated something, it's that nothing needs to be any good, because a shitty simulacrum achieves almost the same effect but costs much less time and money to produce. (Ironically, SOTA LLMs are already way better at writing than typical person writing stuff for money.)
    - vundercind 2y ago
      > (Ironically, SOTA LLMs are already way better at writing than typical person writing stuff for money.) I’m aware of multiple companies that would love to know about these, because they’re currently flailing around trying to replace writers with editors + LLMs and it’s not going great. The closest to success are the ones that are only aiming to turn out stuff one step better than outright book-spam, and even they aren’t quite where they want to be, hardly a productivity bump at all from the LLM use and increased demand on their few talented humans.

hnthrowaway6543 2y ago

> and the companies trying to turn 5 positions into 2 aren’t paying enough to reliably get and keep two people who are good at it. it's turning 5 positions into 7: 5 people to do what currently needs to get done, 2 to try to replace those 5 with AI and failing for several years.

vundercind 2y ago
I mean, yes, that is in practice what I’m seeing so far. A lot of spending, and if they’re lucky productivity doesn’t drop. Best case I’ve seen so far is that it’s a useful tool that gives a small boost, but even for that a lot of folks are so bad at using them that it’s not helping. The situation now is kinda like back when it was possible to be “good at Google” and lots of people, including in tech, weren’t. It’s possible to be good at LLMs, and not a lot of people are.

Vegenoid 2y ago

Yes. The people who can use these tools to dramatically increase their capabilities and output without a significant drop in quality were already great engineers for which there was more demand than supply. That isn't going to change soon.

vundercind 2y ago
Ditto for other use cases, like writer and editor. There are a ton of people doing that work whom I don’t think are ever going to figure out how to use LLMs well. Like, 90% of them. And LLMs are nowhere near making the rest so much better that they can make up for that. They’re ok for Tom the Section Manager to hack together a department newsletter nobody reads, though, even if Tom is bad at using LLMs. They’re decent at things that don’t need to be any good because they didn’t need to exist in the first place, lol.
- TeMPOraL 2y ago
  I disagree. By far, most of the code is created by perpetually replaced fresh juniors churning out garbage. Similarly, most of the writing is low-quality marketing copy churned out by low-paid people who may or may not have "marketing" in their job title. Nah, if the last 10-20 years demonstrated something, it's that nothing needs to be any good, because a shitty simulacrum achieves almost the same effect but costs much less time and money to produce. (Ironically, SOTA LLMs are already way better at writing than typical person writing stuff for money.)
  - vundercind 2y ago
    > (Ironically, SOTA LLMs are already way better at writing than typical person writing stuff for money.) I’m aware of multiple companies that would love to know about these, because they’re currently flailing around trying to replace writers with editors + LLMs and it’s not going great. The closest to success are the ones that are only aiming to turn out stuff one step better than outright book-spam, and even they aren’t quite where they want to be, hardly a productivity bump at all from the LLM use and increased demand on their few talented humans.