Taking credit for something you didn't do, whether it was done by another person or by an LLM, is dishonest and unethical. Prompting an LLM to create something is obviously not the same thing as creating it, and I feel this article is trying to normalize tacit plagiarism. In most contexts I'm aware of where assigning credit actually matters, human authorship and LLM authorship are very much distinguished, and failing to "credit the LLM" in some cases can even have legal consequences.
(You don't have to mention LLMs every time you use them because sometimes the credit really doesn't matter.)
> But for the first reason, I’d challenge the notion of sharing credit with a tool, let alone feeling like a cheat for using one.
That's the main reason LLMs aren't "just a tool."
> Credit and accountability are two sides of the same coin. A tool can only take as much credit as it can be held accountable for, so crediting it inadvertently dilutes your own accountability for the work.
Huh? No, credit and accountability are pretty orthogonal. I use tons of open-source tools, and everyone agrees that the respective authors get credit for what they made, but only I am accountable for my use of them. In fact, all open-source licenses spell this out explicitly.
> 1. You believe in ascribing credit where credit is due. Using an LLM feels like cheating so disclosing it makes you feel less like an imposter for having produced something amazing.
Cheating isn't the right word here. Crediting LLMs is about giving everyone involved accurate information which can be used for later decision making (e.g. if something unexpected happens as a result of a LLM-ism).
In a perfect world such a disclaimer would help anyone reading the output by setting expectations. We each have different experiences with this, and different levels of acceptance. Knowing that something is using generated content can help anyone either accept / skim / thoroughly review something.
The problem is that we're seeing lots of backlash, often for ideological reasons, or for outdated somewhat technical reasons, so people have less of an incentive to be honest, and tons of incentives to hide their use.
This is a little silly — if I quote a large passage from an encyclopedia (or whatever), I cite the source. If I use an LLM to write some of a paper or report, I’m not writing that part (even if I edit it or rephrase it or whatever) so I’d better credit the LLM in whatever way is most consistent with how its output is being presented in the paper.
I completely agree. All crediting LLMs does is give the companies that built them precisely the kind of ego stroking feedback they want. I bet Dario giggles with excitement every time he sees "Claude" as the author of a git commit. People are feeding directly into y heir narrative that these tools are not tools but somehow replacements for human beings.
You piloted the LLM, you are responsible for the code. I don't sign my commits with "autocompletions by Intellij". This whole thing is nonsense.
As for the possibilities listed in TFA, I personally think the most common one is people just blaming the tool or hoping to use citation as cover for slop work. It's basically an admission of "I didn't look into this as thoroughly as I myself think I should have to really understand what I'm doing". Don't be that person. If you don't understand what you are doing prompt the LLM with questions until you do understand. It isn't difficult.
I'm always skeptical of a recommendation to a technical problem when someone tells me they used an LLM chatbot like CoPilot. They really are just slot machines. Sometimes the answer works. Sometimes it doesn't. On the other hand, when someone tells me they googled it and found an article - not that AI overview crap - I trust it more because I know it was written and tested by a human.
I agree but not for the reasons the article states. If you say “an LLM made this” and are proven to be wrong, then you can divert blame to the LLM: this is bad as you don’t in the future try to improve yourself and verify the LLM.
You compare LLMs to spellcheckers, but they are clearly very different in capability at this point. These agents are becoming more human-like and "just a tool" is no longer a fair characterization.
And even if you believe it's just a tool, a text generator that takes in and spits out the Internet, you should cite it like citing a website.
Personification an LLM has become a style or trend. Saying "Sol solved this problem for me" sounds great, indicating your honesty in giving credit to LLM. However this is wrong. You are "using" LLM to solve something, just like "using" a calculator to do some computation. Still the computation is done by you, not calculator. When you drove from X to Y, you did it, not the car.
Isaac Su seems to be irritated and mildly triggered that people don't think or act like LLMs aren't just dumb, mute, simple tools.
Instead of talking about THAT, he is doing a weird combination strawman/bikeshed-like post about the different imagined, nuanced reasons why he imagines people credit LLM input in attribution.
A more honest take would be:
"Some of you are ascribing more sentience and agency to LLMs than I feel comfortable with. Stop doing that. I'm obviously and fundamentally right that they are dumb tools that can never threaten my intellectual and emotional security or human superiority."
Whether they are right or wrong — one reason that many people are giving credit is that they feel that (at some at level at least) that Claude/DeepSeek/Gemini/etc. is acting enough like a participant or partner to deserve being given credit.
You should cite your sources, and that might include an LLM.
If you use AI chat as a search tool and then read whatever web pages it finds to check them for yourself (as you should), then you can cite that instead.
Citing an LLM is a sign that you couldn't be bothered to do that, but at least it's honest.
> Odd. Like a writer disclosing the spellchecker they used alongside every piece,
No it’s not. A spell checker tells me if I misspelled any of my own words.:!it doesn’t write for me.
> or a pilot announcing their version of avionics software over the PA while taxiing to the runway.
No, it’s nothing at all like those things. It would be more like your pilot letting Claude direct his taxiing.
People disclose LLM involvement because they know they didn’t really* author that output, and they want to disavow responsibility for it being totally wrong. Even if I just need a quick figure and I google it, I still disclose that I googled it rather than pulling it out of my own expertise.
There is a difference between crediting the LLM (which i agree is silly) and disclosing the use of an LLM, which I think is important. The author seems to conflate these two.
A disclosure is appropriate whenever the reader might feel differently about the result of armed with this information. For example, "this research on the non-addictive nature of tobacco, funded by the tobacco industry".
Moreover, not citing the LLM is plagiarism. Maybe people outside academia have an oddly narrow understanding of plagiarism, and (for example) cannot understand how it is possible to plagiarize yourself. It is hard to explain the importance of citing LLM contributions to people with this view.
Including AI contributions in your work is curation of its outputs. Should a museum or gallery pretend the works in their exhibits were made by them? Of course not. Cite where you got things from.
An LLM is not me. It is capable of doing more "work" but it also doesn't have my mental models and experience. My colleagues know roughly my areas of expertise. If I say "I think that X is happening" they'll assume that is the result of my expertise being applied and react accordingly. If I say "[LLM] thinks X is happening" they'll react differently and I want them to. Just like I might say "I suspect X is happening because Y" or "X is definitely happening" when I want to modulate their trust in what I'm saying.
At work I recently credited Claude with writing some of the tests in my PR. Why? Partly to apologize for the verbosity of the tests making them harder to review. Partly to set expectations: the tests are more likely to be buggy, low-quality, or incomplete, or misunderstand the code being tested, than if I had written them by hand.
I did in fact "bother to review" the tests, and I thought they were good enough to merge. Indeed, they were better than what I would have written on my own: they were lower-quality on a line-by-line basis, but for tests, quantity is a quality of its own, and I would not have bothered to write nearly as many tests. Even so, it was worth flagging that Claude wrote them.
If I trusted LLMs more, would I still feel the need to credit them? Maybe. They are improving very quickly, and I think they will eventually improve to the point where a mostly-LLM-written PR will be as good as mine in every way. Then it will no longer be necessary to apologize for submitting LLM code. But if and when that point is reached, I wouldn’t get used to taking credit for said code! Because it probably won’t be much longer until the human in the loop becomes redundant.
Going back to today, I don’t feel the need to credit LLMs when I use them for tasks that don’t involve submitting their output. In particular, they have recently become quite good at finding bugs in my code.
By not crediting and personifying the LLM, one (re)takes full human responsibility over the produced works and stated claims. As a result one won’t become an “agent” for the LLM.
The title could have been Don’t blame the LLM.
When we first started using LLMs professionally it was fine to hide behind disclaimers. Now with widespread adoption it’s time to step out of the LLM’s shadow and take credit/blame again.
Sometimes AI's contributions may be diffuse, and more akin to the suffusion of a tea-bag, or even a drop of tincture, than the addition of an extra cookie on the plate. In such cases there is usually no distinct section that you can wall off and say 'AI did this'. In spite of this, it may have made a rather modest than overwhelming contribution to the work.
I'll definitely credit Isaac for his eloquent, non-slop writing style. Tried to start a related discussion lately (https://news.ycombinator.com/item?id=49027642) more focused on the identity of developers in the age of the agent's second or first authorship.
25 comments
[ 0.30 ms ] story [ 11.4 ms ] thread(You don't have to mention LLMs every time you use them because sometimes the credit really doesn't matter.)
> But for the first reason, I’d challenge the notion of sharing credit with a tool, let alone feeling like a cheat for using one.
That's the main reason LLMs aren't "just a tool."
> Credit and accountability are two sides of the same coin. A tool can only take as much credit as it can be held accountable for, so crediting it inadvertently dilutes your own accountability for the work.
Huh? No, credit and accountability are pretty orthogonal. I use tons of open-source tools, and everyone agrees that the respective authors get credit for what they made, but only I am accountable for my use of them. In fact, all open-source licenses spell this out explicitly.
Cheating isn't the right word here. Crediting LLMs is about giving everyone involved accurate information which can be used for later decision making (e.g. if something unexpected happens as a result of a LLM-ism).
In a perfect world such a disclaimer would help anyone reading the output by setting expectations. We each have different experiences with this, and different levels of acceptance. Knowing that something is using generated content can help anyone either accept / skim / thoroughly review something.
The problem is that we're seeing lots of backlash, often for ideological reasons, or for outdated somewhat technical reasons, so people have less of an incentive to be honest, and tons of incentives to hide their use.
You piloted the LLM, you are responsible for the code. I don't sign my commits with "autocompletions by Intellij". This whole thing is nonsense.
As for the possibilities listed in TFA, I personally think the most common one is people just blaming the tool or hoping to use citation as cover for slop work. It's basically an admission of "I didn't look into this as thoroughly as I myself think I should have to really understand what I'm doing". Don't be that person. If you don't understand what you are doing prompt the LLM with questions until you do understand. It isn't difficult.
When I see it, it always makes me want to respond “weird, I asked Claude too and it said <different answer>”.
And even if you believe it's just a tool, a text generator that takes in and spits out the Internet, you should cite it like citing a website.
Isaac Su seems to be irritated and mildly triggered that people don't think or act like LLMs aren't just dumb, mute, simple tools.
Instead of talking about THAT, he is doing a weird combination strawman/bikeshed-like post about the different imagined, nuanced reasons why he imagines people credit LLM input in attribution.
A more honest take would be:
"Some of you are ascribing more sentience and agency to LLMs than I feel comfortable with. Stop doing that. I'm obviously and fundamentally right that they are dumb tools that can never threaten my intellectual and emotional security or human superiority."
Whether they are right or wrong — one reason that many people are giving credit is that they feel that (at some at level at least) that Claude/DeepSeek/Gemini/etc. is acting enough like a participant or partner to deserve being given credit.
If you use AI chat as a search tool and then read whatever web pages it finds to check them for yourself (as you should), then you can cite that instead.
Citing an LLM is a sign that you couldn't be bothered to do that, but at least it's honest.
No it’s not. A spell checker tells me if I misspelled any of my own words.:!it doesn’t write for me.
> or a pilot announcing their version of avionics software over the PA while taxiing to the runway.
No, it’s nothing at all like those things. It would be more like your pilot letting Claude direct his taxiing.
People disclose LLM involvement because they know they didn’t really* author that output, and they want to disavow responsibility for it being totally wrong. Even if I just need a quick figure and I google it, I still disclose that I googled it rather than pulling it out of my own expertise.
A disclosure is appropriate whenever the reader might feel differently about the result of armed with this information. For example, "this research on the non-addictive nature of tobacco, funded by the tobacco industry".
Moreover, not citing the LLM is plagiarism. Maybe people outside academia have an oddly narrow understanding of plagiarism, and (for example) cannot understand how it is possible to plagiarize yourself. It is hard to explain the importance of citing LLM contributions to people with this view.
I did in fact "bother to review" the tests, and I thought they were good enough to merge. Indeed, they were better than what I would have written on my own: they were lower-quality on a line-by-line basis, but for tests, quantity is a quality of its own, and I would not have bothered to write nearly as many tests. Even so, it was worth flagging that Claude wrote them.
If I trusted LLMs more, would I still feel the need to credit them? Maybe. They are improving very quickly, and I think they will eventually improve to the point where a mostly-LLM-written PR will be as good as mine in every way. Then it will no longer be necessary to apologize for submitting LLM code. But if and when that point is reached, I wouldn’t get used to taking credit for said code! Because it probably won’t be much longer until the human in the loop becomes redundant.
Going back to today, I don’t feel the need to credit LLMs when I use them for tasks that don’t involve submitting their output. In particular, they have recently become quite good at finding bugs in my code.
By not crediting and personifying the LLM, one (re)takes full human responsibility over the produced works and stated claims. As a result one won’t become an “agent” for the LLM.
The title could have been Don’t blame the LLM.
When we first started using LLMs professionally it was fine to hide behind disclaimers. Now with widespread adoption it’s time to step out of the LLM’s shadow and take credit/blame again.