Oh I think it is shit, especially reading that 250k and the result was still 20% off of target in a scenario that should, in theory, be right in GenAI’s wheelhouse.

Btw the decomp I was referencing was of the original modern warfare 2 from 2009.

But it’s it really the first step. Feel like they have been making plenty of steps already.

What they have so far argued is that it isn’t infringement until an operator asks for output that infringes, and then it is the operators fault. I hate that argument, but so far they sidestep the dicey issue by putting the burden on the operator.

The rogue agent thing is bullshit, but we have shitty criminal enforcement and complicit would-be plaintiffs rather than a lack of law.

They did for example rule against that one company ripping off anothers legal material despite the “but AI” defense. The court didn’t care one way or the other about the process, they just cared that the output was obviously infringement. AI or no AI.

Based on the high schoolers my kids associate with and how they have broadly ditched the idea of software development as a viable career to try for.

Also our internship program dropped like 90% in applicants this year compared to last. Most years we do interviews to pick the ones that get slots, this time we just took them all and many groups just didn’t get their interns.

I think to the extent AI works, it strongly steers people away from foss collaboration. Why collaborate if the AI can just generate a bespoke solution? If the AI output doesn’t work, then other people aren’t going to be excited to try to help you make it work among a sea of other broken code.

Maybe post bubble pop. If AI falls to work as well as hoped, a move back to the former status quo. If it does work out well enough, maybe sofware developer goes from six figure to a boring middle class job that doesn’t attract so many comers.

Missing is his assessment that Vibe coding can be a nice on ramp to learning how to really code.

I’m super skeptical because I have never seen it happen in a newcomer, but it is pretty good at helping me find good candidates for libraries and how to use them consistent with my broader project design. So on ramp to a solid candidate from not knowing anything is gentler. Seeing the required arguments vaguely mapped to my existing data helps understand, even if I should throw out the generated code as being otherwise more trying than it’s worth, I still learned something useful.

So the lettuce wins again…

To be fair, the slop users are relentless. Trying to even decide when to say yes to becomes a maddening chore if your community is popular and attract tons of random people excited about the novelty of being able to “contribute”.

As in other cases, AI is at its worst when you have to deal with other people’s output. So closed source can have a reasonably happy time because everyone that isn’t on the same page is within slapping distance. But open communities and projects… Oof…

Note this theory of “if AI can clone/decompile, then it is fair game” is a pretty baseless “legal theory”. People are stitching together incomplete understandings of cases. It is kind of like all those “sovereign citizen” legal theories where people think they have the magical legal recipe only to get slapped hard when they have to try to reconcile that with a real court.

I will say that a whole upcoming generation is pretty much noping away from entering this industry. Can’t blame them, as at least for a large chunk of their early earning years, companies will not be hiring them.

I’ve seen lots of deeply AI fanatic behavior where they are entirely oblivious to how obnoxious other people are finding their output.

This case seems like they did some reflection and internalized the situation, which is nice to see. Most decent folks step back when they realize they are actively shitting on the work of people they respect. Hence AI slop being relatively more popular among sociopathic folks.

Note there is a reasonable difference between AI use and slop, but the “lazy didn’t give a shit” nature of AI slopists means they can produce way more volume than the more quality users.

It’s actually a lot tougher to iron out bugs that GenAI can’t sort out for itself than I usually see with human code. Trying to fix a bug the vibe coder couldn’t get the AI to squash is always a maddening nightmare…

Well the real problem is the signal to noise, the noise looking almost just right, and people over excited over the novelty of “I can help now!”, even if they really can’t.

If the thing runs right, then it would be less of a problem, but it almost runs right.

Problem is just the sheer volume of shit to vet. Vibe coders get a bit pissy when folks seemingly can’t “keep up” with their volume of shit and just keep shitting more and making an unmanageable backlog. They figure if the AI in their unskilled hands can get it over 80% there, then surely a skilled person can somehow use AI to knock out that last bit fast.

Issue being that however far the vibe coder got, the skilled people they look to finish it off could have just gotten it that far themselves just as quickly. It’s like someone sending a search to Google on someone’s behalf and pastes the result and thinking that was somehow way easier than the person just searching Google themselves.

Sight amendment, they prompt until it gives them something that looks like it works, based on their often incomplete understanding of the target use case and likely laziness to explore all the nuance of the result.

Like people who were heavy users of Photoshop that tried the vibed knockoff lamented that it failed at all but the most basic image editing tasks. The vibe coder likely never used Photoshop for anything more than casual use, so they couldn’t have vetted the result properly even if they were inclined to.

Yes, working with some “Enterprise” software development teams in my career…

Their codebases were nightmares of people trying shit without actually understanding until the tests seemingly passed.

So no surprise that same crowd is having a pretty decent experience for them with AI: it codes almost as shitty as they did, in the same way.

Yeah, a big issue more broadly. Congratulations, it technically achieved a very narrow requirement, but unlike what we are used to, the result is difficult to take forward, neither human nor GenAI are able to reasonably modify the result in precise ways.

I would be surprised if there is any case where direct use of decompiled code was ruled as “clean room” and transformative. People keep using that phrase without really understanding it.

Think the people assuming that clean room principles apply to this scenario are way too optimistic about the legality.

The most prominent clean room example was making a knockoff of the IBM BIOS. The thing is it was cloning a programming interface surface, and doing so with zero visibility into the object code, but instead an api reference written by people who did study the object code. Wine is even more clean room, guessing at a viable implementation based on black box observation of Windows behavior with zero understanding of the object code.

Translating object code into source code is not “clean room”. But even if it were, we are talking about a deep clone down to the implementation and up through the human presentation. Not merely some programming interface details with a completely novel backend and with other code mostly owning the presentation to a human.

People keep thinking they find a technical loophole but consistently get shot down the millisecond a rights holder takes note.

See the myriad of fan projects that don’t even try to clone a game engine but implement a game a new way, they get shut down hard.

Saw someone about a decomp that after $250k was only 80% accurate.

That said, it is a use case that AI should be viable for: a known source of truth that can be used to verify a guess and drive retries. It needs a lot of tries, hence token costs, but it should, given enough retries, be able to let much “get it”.

I am quite willing to believe in activist and advocacy. However when they explicitly call to not vote as happens suspiciously often, then I have to reconsider their objective, since refusing to vote does absolutely nothing to advance their stance.