claude MAX 20 fuckin X motherfuckerrrrr
-
as a reminder, i have a zero-tolerance policy for this garbage. you do not get to do shit work with an AI assistant, then sign off on it, and then blame the AI for getting it wrong.
I don't care.
you WILL be banned from my repos if you send me AI slop. end of.
to be honest, i hate the AI debate as a topic. like with all things, there are uses where the technology can be a performance gain.
*my* issue is, and has always been, with tool-assisted charlatanism.
i hated the fuzzers last decade because of this.
-
to be honest, i hate the AI debate as a topic. like with all things, there are uses where the technology can be a performance gain.
*my* issue is, and has always been, with tool-assisted charlatanism.
i hated the fuzzers last decade because of this.
@ariadne i think the way the tool is marketed and presented, as a sort of glorified slot machine for code, actively encourages and rewards people for becoming worse engineers. this is in addition to your reasoning.
-
to be honest, i hate the AI debate as a topic. like with all things, there are uses where the technology can be a performance gain.
*my* issue is, and has always been, with tool-assisted charlatanism.
i hated the fuzzers last decade because of this.
with the fuzzers, we used to get lots of bug reports that were not exploitable in any practical fashion
these reports would almost always include "can you tell me more about your bug bounty program?"
-
@ariadne i think the way the tool is marketed and presented, as a sort of glorified slot machine for code, actively encourages and rewards people for becoming worse engineers. this is in addition to your reasoning.
@whitequark my issue is simple: competent engineers can vibe-code and get acceptable results, but this comes at the expense that there will no longer be any future generations of competent engineers.
because to become a senior engineer, you must first be a junior. and AI is already wiping out those jobs.
-
@whitequark my issue is simple: competent engineers can vibe-code and get acceptable results, but this comes at the expense that there will no longer be any future generations of competent engineers.
because to become a senior engineer, you must first be a junior. and AI is already wiping out those jobs.
@ariadne @whitequark Only "acceptable" if you disregard that the code they give you has no provenance. Their copyright claims on it and claims to be able to license it suitably for you aren't valid, and you have no idea who else might have copyright claims on it.
-
@whitequark my issue is simple: competent engineers can vibe-code and get acceptable results, but this comes at the expense that there will no longer be any future generations of competent engineers.
because to become a senior engineer, you must first be a junior. and AI is already wiping out those jobs.
@ariadne I've watched competent engineers get less competent after spending a bunch of time vibe-coding, so I am not actually convinced that what you're saying is generally true except on the shortest timescales. But, again, this is in addition to what you're saying.
-
with the fuzzers, we used to get lots of bug reports that were not exploitable in any practical fashion
these reports would almost always include "can you tell me more about your bug bounty program?"
@ariadne There's a lot of labor that goes into fixing a bug, and also a lot that goes into a full-chain attack. The problem I see now, which I don't have any real ideas for, is that finding bugs is getting much easier, writing exploits is now getting somewhat easier, but the whole process of determining and landing the right fix is not at all easier. We enter a time when there is an ever growing pile of known-but-unfixed bugs. It's a politics and economics problem, really.
-
@ariadne i find it really sus how they're offering FOSS devs free trials, it's like they're trying to get everyone hooked on them
@starchturrets @ariadne "like"??? They are trying to get everyone hooked on this bullshit. I'm not convinced that the whole "AI psychosis" thing is actually a mistake.
-
with the fuzzers, we used to get lots of bug reports that were not exploitable in any practical fashion
these reports would almost always include "can you tell me more about your bug bounty program?"
@ariadne IMHO fuzzers are capable of finding real bugs in a way that AI isn't. That said, once someone produces a tool that can be pointed at code to find "bugs", people will try to extract money from it.
I also haven't been on the receiving end of this.
-
with the fuzzers, we used to get lots of bug reports that were not exploitable in any practical fashion
these reports would almost always include "can you tell me more about your bug bounty program?"
@ariadne `these reports would almost always include "can you tell me more about your bug bounty program?"`
Most (if not all) random folks doing fuzzing on GCC/LLVM have resulted in academic paper rather than asking about a bug bounty program.Maybe that is because the first fuzzing done on GCC/LLVM was done by an university student/professor rather than say some company/random person.
So the motivation is different and most of the time both GCC/LLVM welcomes the fuzzers (even with duplicates filed). Though an ICE after an error is just pushed back to almost never want to fix line. And there was 2 weeks where these kind of fuzzing reports were being filed at rate of 10 a day. So it overloaded bug triagers (mostly myself). -
@ariadne `these reports would almost always include "can you tell me more about your bug bounty program?"`
Most (if not all) random folks doing fuzzing on GCC/LLVM have resulted in academic paper rather than asking about a bug bounty program.Maybe that is because the first fuzzing done on GCC/LLVM was done by an university student/professor rather than say some company/random person.
So the motivation is different and most of the time both GCC/LLVM welcomes the fuzzers (even with duplicates filed). Though an ICE after an error is just pushed back to almost never want to fix line. And there was 2 weeks where these kind of fuzzing reports were being filed at rate of 10 a day. So it overloaded bug triagers (mostly myself).@pinskia I do welcome fuzzer reports, but I reserve the right to close them if they are nonsense.
-
@ariadne @whitequark Only "acceptable" if you disregard that the code they give you has no provenance. Their copyright claims on it and claims to be able to license it suitably for you aren't valid, and you have no idea who else might have copyright claims on it.
@dalias @whitequark in our world, this is applicable.
in the commercial world, it isn't, because they have lawyers who are OK with this, and that is ultimately what matters.
-
@ariadne I've watched competent engineers get less competent after spending a bunch of time vibe-coding, so I am not actually convinced that what you're saying is generally true except on the shortest timescales. But, again, this is in addition to what you're saying.
@whitequark oh, definitely. it is like any other skill, continuous practice is required to maintain it.
-
@dalias @whitequark in our world, this is applicable.
in the commercial world, it isn't, because they have lawyers who are OK with this, and that is ultimately what matters.
@ariadne @whitequark 10 years ago, Google's lawyers were so scared of clearly-non-copyrightable header files myself and the other authors declared public domain that they badgered me into reaching out to get explicit license grant from everyone.
Now they're fine with code of completely bogus provenance.
The world has been turned utterly upside down and I hate it.
-
@ariadne @whitequark 10 years ago, Google's lawyers were so scared of clearly-non-copyrightable header files myself and the other authors declared public domain that they badgered me into reaching out to get explicit license grant from everyone.
Now they're fine with code of completely bogus provenance.
The world has been turned utterly upside down and I hate it.
@dalias @ariadne @whitequark yep so much for their anti AGPL license stance.
-
as a reminder, i have a zero-tolerance policy for this garbage. you do not get to do shit work with an AI assistant, then sign off on it, and then blame the AI for getting it wrong.
I don't care.
you WILL be banned from my repos if you send me AI slop. end of.
@ariadne perfectly sane statement.
-
@starchturrets @ariadne "like"??? They are trying to get everyone hooked on this bullshit. I'm not convinced that the whole "AI psychosis" thing is actually a mistake.
@juliancalaby @starchturrets @ariadne
My company is forcing all of us to use AI or else.
They also wrotr scripts to track if people are using it or not, scripts built with AI... That obviously fail miserably at tracking the real usage metric.Freaking brilliant... Had to send an email today with my premium requests % in GH.
What was the saying when a metric becomes the target it seases to be a valuable metric?
The world we live in lately is... Brilliant.
-
@juliancalaby @starchturrets @ariadne
My company is forcing all of us to use AI or else.
They also wrotr scripts to track if people are using it or not, scripts built with AI... That obviously fail miserably at tracking the real usage metric.Freaking brilliant... Had to send an email today with my premium requests % in GH.
What was the saying when a metric becomes the target it seases to be a valuable metric?
The world we live in lately is... Brilliant.
@YvanDaSilva @starchturrets @ariadne I'm high enough up in the company that I work for that I'm able to tell the next level up that if they pull that sort of shit, I'm out, and that has enough weight that they haven't.
I am counting my lucking fucking stars.
-
undefined oblomov@sociale.network shared this topic