LLMs and the Nature of Honest Authorship

Editing is not authoring.

I've been having a fascinating debate recently with someone claiming to be an Objectivist who sees large language models (like ChatGPT or Claude) as mediums of self-expression.

He wants to be considered honest, even though he's undeniably been using LLMs to compose responses to my fully human-generated comments.

He doesn't like that I've been calling him dishonest for claiming to be the author of his posts.

He claims I'm biased, prejudiced, and non-objective for not treating the LLM's output as his own.

He isn’t a native English speaker, and this complicates the issue. LLMs are quite nice for translating from one language to another. I generally would not fault someone who wrote some text in their native language and then used an LLM to translate it to English, especially if they tagged the text as "translated to English by LLM".

But when someone, out of a desire to appear rational and intelligent, takes my arguments and feeds them into an LLM for the sake of generating a response, and then (even after careful review) provides that response to me as if he'd written the response himself, it offends me as a genuine thinking person.

During this episode, I have been quite averse to entertaining his pretense, and my anger at it has made it hard for me to take him seriously. Does such a hack really deserve my careful attention?

I've alternated between righteous hostility (with its concomitant missteps) and extremely generous consideration toward him, because he claims to be (and wants to be considered) an Objectivist intellectual.

Here’s what I've been asking myself: Is this person an ally, or is he betraying a crucial value by presenting his machine-generated responses as his own?

As a moralist, I think, "The guy is lying about having done original thinking. He's been lazy and dishonest, and he's fooling himself into believing he's an intellectual."

As a tech enthusiast, I think, "Wow, this LLM writes some nice sentences, even if the overall effect is voluminous, condescending and arbitrary sophistry."

In short, I can’t enjoy reading this person's output. It claims to be rational, but it isn’t. Instead, it does a lot of “rationality signaling” by quoting many things that Ayn Rand wrote. It’s tiresome.

I hold that there is an important moral issue here. Yes, it's a matter of honesty, but about what? Combing through the mostly machine-babble of his latest response to me, I did find a pretty good paragraph that gets to the heart of the issue:

"If someone uses AI to produce a draft and then critically evaluates it—correcting what doesn’t fit, validating what does, and assuming full authorship of the final result—then they haven’t evaded their mind. On the contrary, they’ve exercised judgment on a technical input. Isn’t that exactly what we do with any rough draft? What difference does it make if the draft was generated by AI or written during an initial burst of inspiration?"

The question asks, effectively: "What is the essential difference between real writing and using an LLM as a substitute for one's original thinking? Both only generate a rough draft that must be conscientiously edited and rationally approved by one's mind. As a good editor, don't I deserve an author's credit for the text?"

It's a deep metaphysical issue: Are bursts of inspiration from one's subconscious something one should be credited for? Shouldn't we celebrate excellent prompters and editors of LLM output the way we celebrate real authors?

No. We shouldn't. Becoming the kind of person who gets great, genuine inspirations (what we often call "brilliant ideas") requires extensive prior study and thinking. My fundamental premise is that minds generate ideas, and LLMs are only clever artifices that manipulate language to imitate past minds.

In other words, using an LLM to generate arguments against those communicated by other people is a failure of intellectual independence, even if you carefully vet the output. Vetting output isn't the same as writing, even if you do it conscientiously.

Being a great editor is not the same thing as being an author. There's something special, human, and even sacred about originating conceptual communication directly from one's mind.

Aspiring young writers: Don't let that go. Don’t settle for a fake mind, and don’t impose one on others. Don't delegate your original thinking to an LLM. Do not sacrifice your intellectual integrity for the seductive pleasure of generating many pages of plausible rhetoric in five seconds—rhetoric you can then dump, with self-satisfaction, into the lap of anybody you disagree with.

Instead, be a real thinker and author. Go beyond editing. Do something daring and innovative: Create new ideas and write them down.