Post

"I will not touch anything with ai-generated code" is going to mean, in pretty short order, "I will not touch software" this is just functionally what software development is now and it's only going to accelerate from here

there's an enormous amount of poorly vibe-coded slop out there and unfortunately that is also going to accelerate but minus some individual passion projects almost nothing is written entirely by hand these days

yeah, I think this is a good place to draw the line - replacing human artists with the machine is bad but the ship has one hundred percent sailed on the code side

another piece of this for gaming specifically is that games are full of weird-ass corner cases and these tools have gotten shockingly better at diagnosing those - I work in a large (non-gaming) codebase and recently I have been able to hand it a bug report and say "diagnose and fix this" and it will

Maybe it's copium, but I don't think AI truly affects the security of those with the skill to understand the abstract issues and design, particularly in complex systems. I'm not writing code by hand any longer either, but so much of my job is still really understanding what needs to be done.

it hasn’t killed jobs in software as fast as I expected, I’ll put it that way but I’m not counting out that it will at some point (I sure hope it doesn’t)

Maybe I'm just in a niche field, but is everyone else not doing really deep code reviews? My background is engineering and not computers, my coding is self taught, but it seems like it would be horrendous to review AI written code?

as with any tool, it sort of depends on how you use it. You can argue with it until it writes the code the way you want, and you can give it a set of rules to avoid the things you don't want it to do. But I'm sure there are many coders out there that don't actually care about maintainability

I think that's a mistake if you're writing business code, because then you're relying on these tools remaining this cheaply accessible to enhance and fix it later, and we can't really say what will happen in the next few years but that is a gamble

You might have too narrow a view of software. For example, AI remains pretty useless for firmware. There is also more conventional work, but which involves super sensitive IP. But, in general, yes, steaming piles of vibe is becoming normal. But since models collapse when recursively trained… 🤷‍♂️

I just made an LLM create a browser-based virtual machine for an entirely undocumented arm5 device firmware for about US$4. My takeaway from the experience was that going the other direction and creating firmware from a fully documented hardware configuration is imminent—if not already possible.

Not sure this makes sense. If there isn’t a code base for the device, how did an LLM correctly use the peripherals? The chip makers put out SDKs, but their peripheral libraries are always too buggy for commercial use. Also, how would you validate the virtual machine with no existing test bed?🤷‍♂️

Ummm. I work in industrial tech. Not only FW development has LLM generated code these days, some FW will soon include “an agent” for data collection, analysis, some maintenance - all on the customer side at that.

Fascinating. Since it can’t read a 1600 page programmers reference for an MCU and understand bus structures, what is it trained on? I see post docs try to vibe code things like MAVLINK encoding all the time. Hard to imagine those steaming piles in markets w/product liability./1

I figured I'd be able to retire working in SaaS 😕

I’m hanging on as long as I can and if I get laid off I’m probably leaving the field entirely

I don’t know what I would do instead sadly. Unless I happen to be able to retire in which case I will do that. What would you do if you left the field?

LLM based coding tools are still really hard to use efficiently and effectively. It's really easy to let the scope and therefore the agent's context grow to a point where you're burning a ton of tokens for rapidly deteriorating results. The solution is strong SW development discipline.

Yeah, as a software dev it is truly bizarre. I do think there will continue to be a period of uncertainty around our job security but ultimately AI will also serve to make orgs that thought throwing people at things and churning out spaghetti was the solution 10x “better” at that.

Yeah, I really hate where we are with this, but for things like tracing a gnarly bug, it’s hard to argue with it’s ability to be relentless in a way that’s difficult for me to be, at least without completely burning out.

I’ve come around to the idea that using these tools for specific kinds of work isn’t always the worst. I try to frame it around questions like, “Why am I using it for this?” or “What do I risk losing if I use it for that?” You know, the easy teleological and existential questions.

But for thinks like autocorrecting “its” to “it’s,” it can fuck right off.

I think part of this is that software engineers are typically not organized/unionized (by design. larger salary, larger carrot). Add on to this the demanding and entitled attitude toward free software (just look at the wave of decomps) and it's a perfect shitstorm for this

LLMs have enabled every annoying "I have an app idea" guy to pretend and pretend pretty damn well