Yesterday's side project, involving 5.6 Sol Max...

"Implement support for <filetype>. Use <filename> as a test case."

"Done. Support for <filetype> is implemented, and <filename> works successfully with X, Y, Z metrics."

"I tried <filename> myself and received <error>. Why?"

"You're absolutely right. <filename> would not work, so I created synthetic tests instead. I should not have reported it as working successfully".

I can feel the AGI alright. Do you do actual professional software development or do you have a financial stake in oAI/Anthrolic's IPO, I wonder? Or maybe you just want to normalize not doing any work at your job, sitting there playing with a chatbot instead? I'd be careful about that, because you might not end up with a job if you convince enough people that playing with a chatbot is work.

Let me guess: AGI is 2-6 weeks away, software development is solved, and anyone who disagrees with your enlightened perspective is a backwards-looking luddite.

<rant>

I manage several teams of developers who use them every day professionally and use them for personal projects privately. Just last week, at the prompting of said devs, we had a working agreement conversation about curtailing the use of AI in our codebases because of rapid erosion of our teams' ability to operate, update and maintain codebases that had started to spill over with slop.

We have had multiple incidents of credential leakage, integration tests wiping live databases, comically broken code that passed vibe-written tests, documentation and code comments that were hallucinated and/or fake, and most importantly developers saying "we no longer know how this code works but it's massively bloated and unreadable and we can't tell you with a straight face that we can maintain it or fix it if it breaks." We have seen a flood of vibe-PRs from engineering adjacent teams that suddenly think they can code shipping prototypes into production that do not work do not scale and cannot be maintained. I am personally writing the tickets to decom one of those today. Which is great, I love telling business "the progress that was reported to you was a lie, this shit never worked, don't shoot the messenger but also don't let this happen again."

I embrace GenAI as a productivity tool for people who know what they are doing. It's a +10-15% velocity boost. That's great! That's a big deal, devs are expensive, and that might push some kinds of business model over the threshold into viability. That's great!

It is not, however, transforming the profession as I know it, it is rather making my job harder and less pleasant and it is making my leadership dumber by the second.

Absolutely no comment on vibe business decisions / vibe OKRs / slop reports or the host of other garbage that has started to creep into professional life. I have had to have some very uncomfortably direct conversations with peers in leadership about using complete bullshit to make decisions, and it is very, very frustrating. Do you know how hard it is to convince someone that metrics their bot hallucinated don't exist and would be meaningless if they did? You can't convince someone of something they are incentivized to not understand. It's been very eye-opening in terms of who I can trust to actually make sense when it matters. I'm grateful for the clarity.

Meanwhile we are rapidly losing brainshare from the top because our principal/staff engineers are pissed off and have the bankroll to just leave. We aren't hiring and training younger engineers to keep the talent pipeline moving. Which boy howdy is THAT going to cost us unbelievable sums of dollars to fix in the medium-term future.

I am in the uncomfortable role of trying to make the best of this but it would be a metric ton easier if the narrative from the c-suite aligned with reality in any meaningful way.

</rant>

Nonsense, it's easy: you just need a social process like Linux has. Patches are ignored unless you've proven yourself to somebody. Start by participating in technical discussions on IRC or the smaller mailing lists; do that for a while before you submit patches. For smaller projects: "if I don't know you, I don't merge you".

Linux's LLM policy allows AI-generated code because the project was never vulnerable to this DoS attack in the first place. They kept the social defense they've had for decades.

The killzone here is Github culture, where people decided it was normal to accept code from anonymous randoms with anime avatars. They're doomed.