That’s how every ecosystem eventually grows. First come the weeds aimed at colonizing new areas as fast as possible. Then come the small plants, then the shrubs, the trees until it’s a self sustaining forest of different species.

to do very little?

…for 12 hours a day…

Our place measures PR counts so a ton of tiny ass PRs is actually the move.

I felt that way until I just stopped using it entirely

Hey thanks for sharing I’m in basically the exact same boat. There are also pressures in the work place to use it because of the time savings. Personally I try to keep a balance and work on things myself most of the time and if the LLM is used it’s more like a buddy coder than a magic code generation tool, but sometimes it’s just too easy to use it that way. I think it’s important to take pride in your work, so keep doing things your way when you want!

I just started using it as well under a push at work. I’ve so far only used it to write out a script that I knew what I needed and how to do it but it ended up saving me a couple hours. I suppose there wasn’t an interesting problem to solve there so it makes sense to use it but I worry about relying on it since it seems just capable enough to start trusting before the rug gets pulled from under you.

I like to give it unnecessarily complex tasks and work on a hobby.

An elegant solution. I wonder if it is more efficient than a 1TW data centre.

Seriously, though, we use AI in the worst way possible. I like writing code, I hate checking someone else’s code. Why aren’t AIs doing the checking and engineers doing the coding?

Forget “human in the loop” - let’s shoot for “reviewed by AI”. Deep down, we know the humans are barely paying attention when they check Claude’s homework.

Nah, fuck AI. I will pivot careers. Anyone that claims to be a developer but generates all of their code is just a modern day script kiddie. Idgaf about efficiencies because the ethical implications of AI are far worse than the extra money I can make for a company.

I will never be an AI code reviewer, because that’s all prompting is. Like I said, I’ll pivot careers out of dev work before I do that. Pick up a trade and work until I’m 80 if I have to, but that’s still better than being one of Claude’s bitches.

Outside of the ethics of AI, the idea of sitting in front of a screen and being a “prompt engineer” takes everything I enjoy out of coding. If it was that or death, I’d choose death.

I have not found AI code to be so helpful. I actually use it a lot less and more strategically than back when it first became viable. It can save a lot of time in the right situation though.

You’re right, and the person you’re replying to is, too.

My take is an analogy (strictly from the cells in my brain lol) -

The “ground” we all stood on, that previously felt ~flat (trying to be brief here) rapidly shifted under all of us, and very suddenly merely standing in one place, meaning anything from keeping a job, retaining enjoyment (even understanding) of the craft and crafted thing, an ideological stance, a hundred other things -

The ground sloped suddenly, and merely remaining in one place became a struggle not to slide and tumble downhill. And so we all are, tumbling downhill, at varying rates of speed and control, and it seems like the slope might be getting steeper, though at least much more slowly, now…probably.

Resulting from the sudden slope, the most experienced became wildly capable, that shit is real (at least at present). The least experienced? Those who need to get the experience, and to pay their bills, experience their lives, and to maybe later become the wise arbiters? Well, fuck them. Fell into the bottom of that new slope. Fuck everyone else, too, actually, hits everyone in lotsa ways, slope don’t care.

It’s all fucked, the ground has become a maw.

The reality is that these tools are not going away. Thinking that will happen is not realistic.

They’re part of the deal now. People who refuse to use them will simply become obsolete. It starts in technology, but it will quickly move around.

Thinking you can just avoid it isn’t practical. It is no different from cars displacing horses and computers displacing typewriters.

Think of how absurd it would sound if you came across someone who refuses to use a computer and only a typewriter.

Sorry, you need a degree and twenty years of experience for this entry-level position.

I guess the AI isn’t iterating over its “ideas” and code yet. I come up with overcomplicated and overcomplex code too. But my natural neural network keeps reevaluating the code while I think about it, write it down, test it, and often even come back to it later after I moved on to another part of the codebase.
It looks to me like AI is using the infamous waterfall method to write code. It does the “thinking” phase and does bugfix iterations when compilation or tests fail. But never does it question the design it came up with after it wrote it down. It doesn’t reevaluate whether the last round of fixes made tests work that test irrelevant implementation details. It doesn’t question whether there really need to be four methods sharing 90% of their body for doing the basically the same thing only subtly different.
I hold the suspicion that AI will be as slow as humans when it becomes as reliable because what makes us reliable is the redundancy in our thinking processes. The constant reevaluation in subtly different states of mind helps to catch design flaws and silly mistakes.

Reevaluation of finalized code is probably not something you can just put into memory. You would need to modify the harness to automatically inject a new prompt triggering reevaluation of the design and implementation after each step. This would likely increase token use and execution time by an order of magnitude. And if you do it by spawning fresh agents, there would be a high chance of having the AI go in circles, switching between a few different versions of the same shitty design or implementation. So you need to keep the context, which grows each iteration. So context rot may become a real problem. The AI might deviate from the initial goal just like some human devs tend to do…

And natural neural networks learn on the job. They adapt to the codebase and industry they are working with/in. claude.md and memory text files help a lot. But they are still sometimes ignored. And I don’t know why. Also, human rules are almost always somewhat fuzzy. Long lines are to be avoided - but unit tests and that one massive dict of dicts really are a lot more readable when ignoring that rule. There might be a lack of good training material on what makes code readable for humans.

Possibly motivated to do more complex problems to spend more tokens?