Seems like the upside is that it makes the job way easier? What am I missing?
Now of course, you may think you are such a good engineer that companies will kill for you… perhaps that’s true now, but its not true for 90% of the engineers out there. And as the pool of engineers gets reduced, the chances of you being not as good as you thought go up. So the real question is: can you (we all) still make a good living by not using llms. You know support each other and fuck the higher ups? No, we cant. Wwe are full of ourselves, full of elitism (this is HN). We are rational folks, we believe in numbers, in data; we know what we deserve. fuck the rest. The ones who win are the higher ups, of course, not us.
To me, it's pretty simple. I have things to do. This makes it easier for me to do those things. Sometimes that means I can do more things, and sometimes it means I can spend less time on my work, and often both.
I have no idea what the future will hold. But to me, it would be very odd to avoid using extremely useful tools for my current work, because of that uncertainty about the future.
If you're proposing an organized boycott, I would certainly entertain that proposal. But for me, the bar would be high for both certainty that the hypothetical consequences are likely and bad and that the boycott would have a chance of being effective.
At this particular moment, I'm pretty skeptical on both counts. And I'm flatly against the kind of vibes and guilt tripping driven "boycotts" that you're attempting here.
(And I'm way more bullish on the normal legislative and regulatory processes. I think organized boycotts are something to think about if those processes fail.)
Is this a thing? Are there companies out there that don't want to go faster?
There's still an opportunity for engineers to eat their bosses lunch and just start their own company. It's never been easier to start a lower cost competitor.
Employment isn't a social law of nature: it's a transaction of money for "units of work", just like the business might have with other vendors. Governments should be making it easier to become a vendor.
The juniors are eliminated and the seniors indulge in cognitive surrender because it feels good.
Here’s a thought: consider the potentially analogous case of performance-enhancing drugs for athletes. The drugs unambiguously make them better at their jobs, but the drugs have severe long-term health costs and wreak havoc on the fairness of the playing field. It’s easy to see why an athlete might choose not to use them, even when others are.
Of course, those negative factors alone are not enough to dissuade people en masse who want to get a leg up on their competitors, so the use of performance-enhancing drugs must be further restricted by institutional bans.
Business is not like this, because the value of what a business does is in its actual output, not in its entertainment value for spectators.
Of course there are other rules (legislative and regulatory) that apply, for other good reasons. But their goal is not to create an entertaining competitive environment, but rather to control externalities of what companies do.
I favor AI regulation, but I also don't think treating it like a performance enhancing drug would be a smart way to regulate it. Higher business productivity is useful to society in a way that breaking home run records is not.
It's interesting though, for a long time I said that if I were going to do a personal programming project I was excited about, I would write all the code by hand, because I do really miss doing that, and I also worry about forgetting how to. But now I'm not so sure. I find my daydreaming about personal projects to be a lot more focused on the outcome than the process, lately. More like "wow, I could do so much in an hour or two a day now! think of the possibilities!" than an excitement about writing code and creating pleasant abstractions.
The revealed preference is very far in the opposite direction at the moment.
Serious question. I think the reason that there's such a disconnect among AI-for-work users about whether it's a panacea or bullshit accelerator is that different software developers have massively different duties and conceptions about what their job is or should be.
2. Figure out what components already exist and what new things we need to build and how things should be integrated.
3. Actually build the things according to what was figured out in the previous step.
4. Review my own work and other people's work.
5. Release things and make sure they work.
6. Respond to emergent issues in things that have been released.
I find the current generation of AI tooling to be very useful for all of these tasks. Less so for task #1 than the others.
What are other people doing that is different?