Yeah, the glorious future where every half-as-good-as-expert developer is now only 25% as good as an expert (a level of performance also known as being "completely shit at it"), but he's writing 10x the amount of unusable shitcode.
I think more low tier output would be a disaster.
Even pre AI I had to deal with a project where they shoved testing and compliance at juniors for a long time. What a fucking mess it was. I had to go through every commit mentioning Coverity because they had a junior fixing coverity flagged "issues". I spent at least 2 days debugging a memory corruption crash caused by such "fix", and then I had to spend who knows how long reviewing every such "fix".
And don't get me started on tests. 200+ tests, of them none caught several regressions in handling of parameters that are shown early in the frigging how-to. Not some obscure corner case, the stuff you immediately run into if you just follow the documentation.
With AI all the numbers would be much larger - more commits "fixing coverity issues" (and worse yet fixing "issues" that LLM sees in code), more so called "tests" that don't actually flag any real regressions, etc.
I suspect that the kind of people who would "know how to use it" don't use it right now since it has not yet reached "useful if you know how to use it" status.
Software work is dominated by the fat tail distribution of time it takes to figure out and fix a bug. Not by typing code. LLMs, much like any other form of cutting and pasting code without having any clue what it does, gives that distribution a longer, fatter tail, hence its detrimental effect on productivity.
And the other "nuanced" take, common on my linkedin feed, is that people who learn how to use (useless) AI are gonna replace everyone with their much increased productive output.
Even if AI becomes not so useless, the only people whose productivity will actually improve are the people who aren't using it now (because they correctly notice that its a waste of time).
That philosophy always ends in stepping into dogshit to try to boost stock prices.
When they tested on bugs not in SWE-Bench, the success rate dropped to 57‑71% on random items, and 50‑68% on fresh issues created after the benchmark snapshot. I’m surprised they did that well.
After the benchmark snapshot. Could still be before LLM training data cut off, or available via RAG.
edit: For a fair test you have to use git issues that had not been resolved yet by a human.
This is how these fuckers talk, all of the time. Also see Sam Altman's not-quite-denials of training on Scarlett Johansson's voice: they just asserted that they had hired a voice actor, but didn't deny training on actual Scarlett Johansson's voice. edit: because anyone with half a brain knows that not only did they train on her actual voice, they probably gave it and their other pirated movie soundtracks massively higher weighting, just as they did for books and NYT articles.
Anyhow, I fully expect that by now they just use everything they can to cheat benchmarks, up to and including RAG from solutions past the training dataset cut off date. With two of the paper authors being from Microsoft itself, expect that their "fresh issues" are gamed too.
Yeah I'm thinking that people who think their brains work like LLM may be somewhat correct. Still wrong in some ways as even their brains learn from several orders of magnitude less data than LLMs do, but close enough.
You can film with an actual camera then use video to video to make it look very AI. If you're just grifting, that would be the way to go I think.
They're also very gleeful about finally having one upped the experts with one weird trick.
Up until AI they were the people who were inept and late at adopting new technology, and now they get to feel that they're ahead (because this time the new half-assed technology was pushed onto them and they didn't figure out they needed to opt out).
I was writing some math code, and not being an idiot I'm using an open source math library for doing something called "QR decomposition", and its efficient, and it supports sparse matrices (matrices where many numbers are 0), etc.
Just out of curiosity I checked where some idiot vibecoder would end up. AI simply plagiarizes from some shit sample snippets which exist purely to teach people what QR decomposition is. It's actually unusable, due to being numerically unstable.
Who in the fuck even needs this shit to be plagiarized, anyway?
It can't plagiarize a production quality implementation, because you can count those on the fingers of one hand, they're complex as fuck and you can't just blend a few together to try to pretend you didn't plagiarize.
The answer is, people who are peddling the AI. They are the ones who ordered plagiarism with extra plagiarism on top. These are not coding tools, these are demos to convince the investors to buy the actual product, which is company's stock. There's a little bit of tool functionality (you can ask them to refactor the code), but it's just you misusing a demo to try to get some value out of it.
And to that end, the demos take every opportunity to plagiarize something, and to talk about how the "AI" wrote the code from scratch based on its supposed understanding of fairly advanced math.
And in coding, it is counter productive to plagiarize. Many of the open source libraries can be used in commercial projects. You get upstream fixes for free. You don't end up with some bugs or worse yet security exploits that may have been fixed since the training cut-off date.
No fucking one in the right mind would willingly want their product to contain copy pasted snippets from stale open source libraries, passed through some sort of variable-renaming copyright laundering machine.
Except of course the business idiots who are in charge of software at major companies, who don't understand software. Who just failed upwards.
They look at plagiarized lines and count them as improved productivity.
Indistinguishable from a business idiot.
I love to show that kind of shit to AI boosters. (In case you're wondering, the numbers were chosen randomly and the answer is incorrect).
They go waaa waaa its not a calculator, and then I can point out that it got the leading 6 digits and the last digit correct, which is a lot better than it did on the "softer" parts of the test.
In 3 months they'll think they're 40% faster while being 38% slower. And sometime in 2026 they will be exactly 100% slower - the moment referred to as "technological singularity".