Why are… some fraction of the programmers so upset about this if it’s a nothingburger?
Because it’s being used as an excuse to fire people ans overwork whoever’s left. Nobody wants to work in a sweatshop.
I’ve been using the new models ever since the initial release of GPT-3 at work. It’s been my job to find practical uses for it. There exist a few, but the biggest issue with these models (and with LLMs as a whole) is that they cannot be guaranteed to work 100%, or close to 100%, of the time. Naturally, that’s a high bar to hit, but it includes things like safeguards against generating offensive, discriminatory, or even defamatory material.
Your problem space is limited to places where either an informed human must review the output, or you accept the liability for when it fails. Most businesses do not want to accept that liability.
Now onto programming specifically: a human reviews the output and is expected to take ownership of it. Sure, you can use it in that way as a tool if you want, regardless of the quality of the output. However, when someone comes to me on a Friday and asks me to review their 700+ file AI-generated PR (true story), I’m sorry, but that’s going to take at least a week. By the time I’m done reviewing your PR, the demo we were rushing to prepare for will have already passed. Yet now I’m expected to review this slop (which I would not have, and didn’t, approve regardless of the author) in addition to doing the work I was doing before. Keep in mind, of course, that my job before included the rare 100hr work week to rush to meet a deadline. So all I can do with these gigantic PRs of actual garbage is either reject them, which looks bad on me because we’re rushing even harder now, or rubber stamp them because I know nobody cares if the code works anyway.
So yeah, wake me up when it’s actually capable of automating my job and not just an excuse for some idiotic management to fire their programmers who, according to the garbage they’re saying about AI, should be the most productive people in the company by a huge margin and therefore more valuable than any of their other employees, at least according to their claims about AI.
Because it’s being used as an excuse to fire people and overwork whoever’s left. Nobody wants to work in a sweatshop.
If this was all it is, there would be no real reason for anyone to be complaining. If the technology really were a dead end and all the managers were just deluded by marketing, then the situation will self-correct sooner rather than later when companies run into trouble when AI fails to produce results after they fired most of their developers.
We’re currently into year 2 (or more, depending on how you count) of this shift happening and I’m not seeing many signs of such a correction on the horizon.
So yeah, wake me up when it’s actually capable of automating my job
Automation does not need to replace every human to have an impact on the labour economy. It’s enough for it to increase efficiency to make some portion of the workforce redundant. AI doesn’t need to automate your job to put you out of a job. All it has to do is enable a colleague of yours to do both their job and yours.
This is factually happening particularly to entry level programming jobs. Rejecting AI is not going to change this reality. The only sane choice is to accept it on our terms and fight to ensure the change works in our favour and not against us.
We’re currently into year 2 (or more, depending on how you count) of this shift happening and I’m not seeing many signs of such a correction on the horizon.
The entire stock market is in a bubble. You don’t need me to explain why - there are so many articles and videos online covering it, from well-known video journalists like GamersNexus to written articles by Financial Times. This Wikipedia article covers both sides, though you’ll notice that the oppositions on that page come from people with a financial interest in it. The correction to this bubble would cripple the economy (which seems to be a recurring phenomenon these days).
This is factually happening particularly to entry level programming jobs.
Entry level jobs were never about productivity. They were about investment. Weird to say that AI is “replacing” them when it was never about them being super productive.
The rest makes no sense. You’d hire more people if their output is so high because every programmer would be worth way more than what you’re paying them.
Because it’s being used as an excuse to fire people ans overwork whoever’s left. Nobody wants to work in a sweatshop.
I’ve been using the new models ever since the initial release of GPT-3 at work. It’s been my job to find practical uses for it. There exist a few, but the biggest issue with these models (and with LLMs as a whole) is that they cannot be guaranteed to work 100%, or close to 100%, of the time. Naturally, that’s a high bar to hit, but it includes things like safeguards against generating offensive, discriminatory, or even defamatory material.
Your problem space is limited to places where either an informed human must review the output, or you accept the liability for when it fails. Most businesses do not want to accept that liability.
Now onto programming specifically: a human reviews the output and is expected to take ownership of it. Sure, you can use it in that way as a tool if you want, regardless of the quality of the output. However, when someone comes to me on a Friday and asks me to review their 700+ file AI-generated PR (true story), I’m sorry, but that’s going to take at least a week. By the time I’m done reviewing your PR, the demo we were rushing to prepare for will have already passed. Yet now I’m expected to review this slop (which I would not have, and didn’t, approve regardless of the author) in addition to doing the work I was doing before. Keep in mind, of course, that my job before included the rare 100hr work week to rush to meet a deadline. So all I can do with these gigantic PRs of actual garbage is either reject them, which looks bad on me because we’re rushing even harder now, or rubber stamp them because I know nobody cares if the code works anyway.
So yeah, wake me up when it’s actually capable of automating my job and not just an excuse for some idiotic management to fire their programmers who, according to the garbage they’re saying about AI, should be the most productive people in the company by a huge margin and therefore more valuable than any of their other employees, at least according to their claims about AI.
If this was all it is, there would be no real reason for anyone to be complaining. If the technology really were a dead end and all the managers were just deluded by marketing, then the situation will self-correct sooner rather than later when companies run into trouble when AI fails to produce results after they fired most of their developers.
We’re currently into year 2 (or more, depending on how you count) of this shift happening and I’m not seeing many signs of such a correction on the horizon.
Automation does not need to replace every human to have an impact on the labour economy. It’s enough for it to increase efficiency to make some portion of the workforce redundant. AI doesn’t need to automate your job to put you out of a job. All it has to do is enable a colleague of yours to do both their job and yours.
This is factually happening particularly to entry level programming jobs. Rejecting AI is not going to change this reality. The only sane choice is to accept it on our terms and fight to ensure the change works in our favour and not against us.
The entire stock market is in a bubble. You don’t need me to explain why - there are so many articles and videos online covering it, from well-known video journalists like GamersNexus to written articles by Financial Times. This Wikipedia article covers both sides, though you’ll notice that the oppositions on that page come from people with a financial interest in it. The correction to this bubble would cripple the economy (which seems to be a recurring phenomenon these days).
Entry level jobs were never about productivity. They were about investment. Weird to say that AI is “replacing” them when it was never about them being super productive.
The rest makes no sense. You’d hire more people if their output is so high because every programmer would be worth way more than what you’re paying them.
Yeah I’m kinda done arguing with that guy too, probably asking Grok to come up with arguments 😭