If the use and inclusion of LLM derived output is truly defensible, why would you need to conceal its provenance or existence?
The debasement and solvency of competency wrought by the unbridled abdication to responsibility to such an incompetent construct is beyond depressing, this Is truly a world no longer worth inhabiting.
The question is why it matters in the first place. As long as the person submitting code takes ownership of the code, and there is a review of the code being done the same as before, nothing actually changes. Your whole line of argument is absurd because it’s premised on the idea that we can trust code produced by a human more than code produced by an LLM. And that’s obviously not true because humans make mistakes all the time. So, the focus on how the code was produced is fundamentally misguided. The focus in any successful project is on validation through things like code reviews, specifications, testing, and other well established practices. That’s what responsible teams have always done.
To me, it matters because the use and reliance on LLMs demonstrably erodes cognitive capacity and ability, and submissions tainted by them should be subjected to an elevated level of scrutiny, which can only be facilitated with a consummate element of transparency. That elevated scrutiny could, and probably should, also be applied to non-LLM infected submissions, but doing so wouldn’t negate the labelling of authorship.
Good faith proponents of LLM assisted tooling should be in favour of overt transparency and labelling, as it’s a requirement to facilitate future historical analysis of their utility, efficacy, and legitimacy; you know, to build a body of evidence.
I assume you’re citing a couple of studies with tiny samples that have been going around here. There’s very little solid evidence to support the claim that LLMs demonstrably erode cognitive capacity. But if you have large scale, peer reviewed studies definitively showing this to be the case feel free to share them.
People lose certain skills working with LLMs, but that has always been the case with automation. Very few people know how to write assembly by hand, and practically nobody knows how to operate punch cards. Yet, we hardly claim that there was some loss of cognitive capacity as a result. People just spend their time focusing on different things now.
And I’ve already explained to you why the focus on the tooling is deeply misguided. You ignored my points entirely, which tells me that you’re not actually trying to have a discussion in good faith here yourself.
To me it matters because if you got code from someone else, you’d probably want to credit it to them, right? So why shouldn’t LLM output be treated the same?
True in some regards, however the word completion in my IDE is only active when I’m typing new code in, it only picks words that already exist in the code (along with language specific keywords, etc). It never types the following word for me, and I have to manually accept the completion I want. It is inactive when I’m typing comments in, and so on.
The LLM you tell to write you a function, and it writes the whole thing, comments and all. It’s not quite the same. The original GitHub Copilot was closest to traditional IDE completion though, and there if it wrote a whole long function for me I’d definitely give credit where it’s due.
It’s a different workflow, but all LLM does is just more automation. The intention and the ideas still come from the developer, just because the tool automates the work of typing out code doesn’t really change anything. Nobody who’s done any serious software development thinks that typing code is the important part of coding.
The LLM can come up with all the implementation details and such by itself, while the developer only provides wishes for new features. Figuring out where to draw the line between it just being fully “vibecoded” or being code the developer would have written manually is super hard, so why not just always say if LLM generated the code? What’s the harm in that?
Anyone who believes, and confidently espouses, that the autocomplete from the dictionary included in an IDE is functionally indistinguishable from the function and output of a LLM, is betraying a foundational lack of understanding of either. The former is constrained and deterministic, whereas the latter is explosively imprecise and intrinsically unreliable.
You keep invoking the notion that LLMs are merely tools, and should be treated as mindless subservient instruments deployed by a programmer in command. The thing, however, with tools is they can often trivially become weapons if wielded with incendiary intent, or indiscriminate instruments of destruction under the direction of incompetence.
Skill is acquired through use over time. LLMs are far too new for anyone to have legitimately accrued enough experience which could be mistaken for competence. With this in mind, anyone claiming to know what they, or the model they’re poking with a prompt, are doing, is a charlatan.
Anyone who makes lazy straw man arguments is betraying utter lack of intellectual integrity. Nowhere did I say their output is indistinguishable, what I actually said is that they work on the same principles. These are both stochastic systems.
The LLMs are very obviously mere tools that are instruments programmers use. Every tool can be wielded as a weapon, there’s nothing unique about LLMs in this regard. I also have no idea what this has to do with AI use disclosure which is the context for this thread.
How LLMs work is well understood, and that’s a completely different subject from how to use them effectively as a skill. With this in mind, the person who can’t even make a coherent argument, and keeps conflating unrelated things, is the actual charlatan here.
Who said anything about concealing it? Not making adisclaimer mandatory and concealing something are two different things.
From the point of view of a reviewer, the code of the PR is all that matters. You shouldn’t look at who wrote the code or what tools they used.
If the code is slop, it does not matter that a human or an AI wrote it, it is slop. Same if the code does what it’s supposed to do and has a high enough quality to pass review.
If the use and inclusion of LLM derived output is truly defensible, why would you need to conceal its provenance or existence?
The debasement and solvency of competency wrought by the unbridled abdication to responsibility to such an incompetent construct is beyond depressing, this Is truly a world no longer worth inhabiting.
The question is why it matters in the first place. As long as the person submitting code takes ownership of the code, and there is a review of the code being done the same as before, nothing actually changes. Your whole line of argument is absurd because it’s premised on the idea that we can trust code produced by a human more than code produced by an LLM. And that’s obviously not true because humans make mistakes all the time. So, the focus on how the code was produced is fundamentally misguided. The focus in any successful project is on validation through things like code reviews, specifications, testing, and other well established practices. That’s what responsible teams have always done.
To me, it matters because the use and reliance on LLMs demonstrably erodes cognitive capacity and ability, and submissions tainted by them should be subjected to an elevated level of scrutiny, which can only be facilitated with a consummate element of transparency. That elevated scrutiny could, and probably should, also be applied to non-LLM infected submissions, but doing so wouldn’t negate the labelling of authorship.
Good faith proponents of LLM assisted tooling should be in favour of overt transparency and labelling, as it’s a requirement to facilitate future historical analysis of their utility, efficacy, and legitimacy; you know, to build a body of evidence.
I assume you’re citing a couple of studies with tiny samples that have been going around here. There’s very little solid evidence to support the claim that LLMs demonstrably erode cognitive capacity. But if you have large scale, peer reviewed studies definitively showing this to be the case feel free to share them.
People lose certain skills working with LLMs, but that has always been the case with automation. Very few people know how to write assembly by hand, and practically nobody knows how to operate punch cards. Yet, we hardly claim that there was some loss of cognitive capacity as a result. People just spend their time focusing on different things now.
And I’ve already explained to you why the focus on the tooling is deeply misguided. You ignored my points entirely, which tells me that you’re not actually trying to have a discussion in good faith here yourself.
To me it matters because if you got code from someone else, you’d probably want to credit it to them, right? So why shouldn’t LLM output be treated the same?
LLM is a tool that a human uses, do you also credit your IDE when you write code?
My IDE does not write the code for me, at most it autocompletes words.
The LLM works on the exact same principle as autocomplete does. You’re just drawing an arbitrary line on technology you’re used to.
True in some regards, however the word completion in my IDE is only active when I’m typing new code in, it only picks words that already exist in the code (along with language specific keywords, etc). It never types the following word for me, and I have to manually accept the completion I want. It is inactive when I’m typing comments in, and so on.
The LLM you tell to write you a function, and it writes the whole thing, comments and all. It’s not quite the same. The original GitHub Copilot was closest to traditional IDE completion though, and there if it wrote a whole long function for me I’d definitely give credit where it’s due.
It’s a different workflow, but all LLM does is just more automation. The intention and the ideas still come from the developer, just because the tool automates the work of typing out code doesn’t really change anything. Nobody who’s done any serious software development thinks that typing code is the important part of coding.
The LLM can come up with all the implementation details and such by itself, while the developer only provides wishes for new features. Figuring out where to draw the line between it just being fully “vibecoded” or being code the developer would have written manually is super hard, so why not just always say if LLM generated the code? What’s the harm in that?
Anyone who believes, and confidently espouses, that the autocomplete from the dictionary included in an IDE is functionally indistinguishable from the function and output of a LLM, is betraying a foundational lack of understanding of either. The former is constrained and deterministic, whereas the latter is explosively imprecise and intrinsically unreliable.
You keep invoking the notion that LLMs are merely tools, and should be treated as mindless subservient instruments deployed by a programmer in command. The thing, however, with tools is they can often trivially become weapons if wielded with incendiary intent, or indiscriminate instruments of destruction under the direction of incompetence.
Skill is acquired through use over time. LLMs are far too new for anyone to have legitimately accrued enough experience which could be mistaken for competence. With this in mind, anyone claiming to know what they, or the model they’re poking with a prompt, are doing, is a charlatan.
Anyone who makes lazy straw man arguments is betraying utter lack of intellectual integrity. Nowhere did I say their output is indistinguishable, what I actually said is that they work on the same principles. These are both stochastic systems.
The LLMs are very obviously mere tools that are instruments programmers use. Every tool can be wielded as a weapon, there’s nothing unique about LLMs in this regard. I also have no idea what this has to do with AI use disclosure which is the context for this thread.
How LLMs work is well understood, and that’s a completely different subject from how to use them effectively as a skill. With this in mind, the person who can’t even make a coherent argument, and keeps conflating unrelated things, is the actual charlatan here.
Who said anything about concealing it? Not making adisclaimer mandatory and concealing something are two different things.
From the point of view of a reviewer, the code of the PR is all that matters. You shouldn’t look at who wrote the code or what tools they used.
If the code is slop, it does not matter that a human or an AI wrote it, it is slop. Same if the code does what it’s supposed to do and has a high enough quality to pass review.