As I read your thought-provoking article, I was reminded of an e-newsletter on AI in business that I co-published with a colleague. Sometimes just to get it started, Ed (who was used to writing on deadline), would get me a first, what we called "dirty draft" even if it was very rough and quick. Since I was the better wordsmith, I would then polish it and make my own rewrite/contribution. On some of those articles, I felt like an old-school, deeply involved editor - that if we weren't sharing authorship for the entire newsletter, he would have been first or sole author.
In your example, when the AI creates the first draft, even if all the points are subsequently revised by the human writer, it felt like the AI I should get at least co-authorship credit, or perhaps even first authorship. It's interesting how my opinion of who would be author was influenced by whether the first draft was created by a human or an LLM.
Thanks for a fascinating read that brought me to think about new (and old) questions!
I think I separate the "issue" into two categories, one of process and one of ideation. I'll start with the latter, first.
Ideas are, in fact, quite cheap. Any writer (or creator of any type, for that matter) knows this innately. Everyone 'has ideas'. It is not hard to come up with a book idea, or article idea, or look at a scene and say "this should be painted in a way that is beautiful." I think anyone can—and does—do this with some regularity. The outset of staring at a blank page with only an idea is the epitome of this, an idea devoid of anything else.
Now, for process. If I told you that during some recent travels I "saw an expanse of oceanside that was so beautiful it deserved to be painted in a way that made you feel the beauty of the place," you might hear that and agree with me. But this is boring. Anyone could have that idea, and the idea means much of nothing. But, should I then bring out my pencils and sketch a draft of an idea that I intended to paint, and then using that draft, painted something beautiful, I would have engaged, via process, with the creation of something.
I think that many people (particularly those who don't often create art) tend to overvalue their ideas and undervalue the process of bringing them about. And much of that "value attribution system" stems from the process itself. Circumventing hours spent in front of a blank page removes the very thing that gives art value in the first place: the process.
I don't know how this applies to non-art work. Is "the process" important for a work's presentation deck? Probably not. But, to reference a quote from the piece, "The jumpstart it gave may have been exactly what allowed an anxious or blocked writer to move from inchoate thoughts to articulated prose." How is one to strengthen themselves from anxious or blocked writer when confronting a blank page than to learn how to fill it, delete it, do it again, edit it, delete it, and then do it once more? Perhaps it is simply about meaning, and how we create and assign it. I worry that "inspired by AI" or "ideated to completion using AI" is an encroachment in the field of authorship, even as I myself send my own pieces to it, seeking some strange form of approval despite its programming to give it to me anyway. I don't know, but your piece is wonderful.
Should we imagine Sisyphus happy after we've given him a Ford f150 and tow rope?
So interesting! I thinks this begs many important questions about the origin of ideas. I was just reading back through Pride and Prejudice, and I noticed myself very clearly echoing Austen’s prose later in my own writing. I tend to absorb writers as I read them through osmosis. Their thinking becomes part of my thinking, whether or not I realize it. Is any ship really built from scratch?
Maybe the use of LLMs condenses this whole process temporally and makes the lineage more explicit, but it does seem worth thinking about how it’s that different. 🤔
Thanks for sharing. As an older than normal PhD candidate, I often feel behind the curve as I hear my peers share "their work". That is until I realize they can't defend "their work" because they didn't research or write it.
Excellent article. Thank you for this. As a high school teacher I do know that students I’ve had have prompted LLMs to include seemingly human sourced errors randomly in the work its to generate for them. There are also LLM humanizers which take a LLM document and, based on parameters such as grade level or essay type for example, edit the document to remove signs of AI and insert random errors in. Unless a piece is written in my class in front of me my default setting is to doubt its human origin.
The Ship of Theseus analogy works up to a point. By starting a piece with the aid of LLMs and then changing the text or expanding on the ideas presented in it, the author is creating a different “ship.” If you added two masts, put an outboard engine on it, repainted Theseus’ ship bright orange and christened it “Minotaur Slayer” it would be a different ship. Yes, some planks may remain but the new version would be fundamentally different from the original. If I ask Claude for a prompt to break from a writers block period and then use that to expand into a full story then a morsel of the original prompt remains but that is dwarfed by my own ideas and words. It’s an issue of scale. If I read a newspaper article or an advertisement that inspires a story of mine, is that all that different from using a LLM to create a prompt to inspire a new story or essay? As long as the generated prompt remains only a small cell of the larger body of work then it would stand to reason the authorship of the piece is overwhelmingly human.
'The Ship of Theseus analogy works up to a point. By starting a piece with the aid of LLMs and then changing the text or expanding on the ideas presented in it, the author is creating a different “ship.” If you added two masts, put an outboard engine on it, repainted Theseus’ ship bright orange and christened it “Minotaur Slayer” it would be a different ship. Yes, some planks may remain but the new version would be fundamentally different from the original.'
That's the whole point of the Ship of Theseus: it is a thought experiment about how many things that one can change any the ship until, eventually, it is a different ship. One of the most persistent puzzles in philosophy is how to make sense of the relationship between identity and change. And the same logic of change over time applies to human edited LLMs as you right pointed out. The ancient meets the cutting edge. What a time to be alive!
As I read your thought-provoking article, I was reminded of an e-newsletter on AI in business that I co-published with a colleague. Sometimes just to get it started, Ed (who was used to writing on deadline), would get me a first, what we called "dirty draft" even if it was very rough and quick. Since I was the better wordsmith, I would then polish it and make my own rewrite/contribution. On some of those articles, I felt like an old-school, deeply involved editor - that if we weren't sharing authorship for the entire newsletter, he would have been first or sole author.
In your example, when the AI creates the first draft, even if all the points are subsequently revised by the human writer, it felt like the AI I should get at least co-authorship credit, or perhaps even first authorship. It's interesting how my opinion of who would be author was influenced by whether the first draft was created by a human or an LLM.
Thanks for a fascinating read that brought me to think about new (and old) questions!
Cool story! Thank for sharing.
Incredible essay. Cheers for sharing!
Loved the piece.
I think I separate the "issue" into two categories, one of process and one of ideation. I'll start with the latter, first.
Ideas are, in fact, quite cheap. Any writer (or creator of any type, for that matter) knows this innately. Everyone 'has ideas'. It is not hard to come up with a book idea, or article idea, or look at a scene and say "this should be painted in a way that is beautiful." I think anyone can—and does—do this with some regularity. The outset of staring at a blank page with only an idea is the epitome of this, an idea devoid of anything else.
Now, for process. If I told you that during some recent travels I "saw an expanse of oceanside that was so beautiful it deserved to be painted in a way that made you feel the beauty of the place," you might hear that and agree with me. But this is boring. Anyone could have that idea, and the idea means much of nothing. But, should I then bring out my pencils and sketch a draft of an idea that I intended to paint, and then using that draft, painted something beautiful, I would have engaged, via process, with the creation of something.
I think that many people (particularly those who don't often create art) tend to overvalue their ideas and undervalue the process of bringing them about. And much of that "value attribution system" stems from the process itself. Circumventing hours spent in front of a blank page removes the very thing that gives art value in the first place: the process.
I don't know how this applies to non-art work. Is "the process" important for a work's presentation deck? Probably not. But, to reference a quote from the piece, "The jumpstart it gave may have been exactly what allowed an anxious or blocked writer to move from inchoate thoughts to articulated prose." How is one to strengthen themselves from anxious or blocked writer when confronting a blank page than to learn how to fill it, delete it, do it again, edit it, delete it, and then do it once more? Perhaps it is simply about meaning, and how we create and assign it. I worry that "inspired by AI" or "ideated to completion using AI" is an encroachment in the field of authorship, even as I myself send my own pieces to it, seeking some strange form of approval despite its programming to give it to me anyway. I don't know, but your piece is wonderful.
Should we imagine Sisyphus happy after we've given him a Ford f150 and tow rope?
So interesting! I thinks this begs many important questions about the origin of ideas. I was just reading back through Pride and Prejudice, and I noticed myself very clearly echoing Austen’s prose later in my own writing. I tend to absorb writers as I read them through osmosis. Their thinking becomes part of my thinking, whether or not I realize it. Is any ship really built from scratch?
Maybe the use of LLMs condenses this whole process temporally and makes the lineage more explicit, but it does seem worth thinking about how it’s that different. 🤔
How many questions in philosophy bottom out in how we account for change? 😂
I agree. I think change, identity, and cause make up like half of philosophy puzzles, with the other being the normative stuff.
Well if the radical empiricists are right, then there are no novel ideas. Only recycled ones.
Reverse engineering works for a strech...may do the trick...for a while.
Thanks for sharing. As an older than normal PhD candidate, I often feel behind the curve as I hear my peers share "their work". That is until I realize they can't defend "their work" because they didn't research or write it.
Excellent article. Thank you for this. As a high school teacher I do know that students I’ve had have prompted LLMs to include seemingly human sourced errors randomly in the work its to generate for them. There are also LLM humanizers which take a LLM document and, based on parameters such as grade level or essay type for example, edit the document to remove signs of AI and insert random errors in. Unless a piece is written in my class in front of me my default setting is to doubt its human origin.
The Ship of Theseus analogy works up to a point. By starting a piece with the aid of LLMs and then changing the text or expanding on the ideas presented in it, the author is creating a different “ship.” If you added two masts, put an outboard engine on it, repainted Theseus’ ship bright orange and christened it “Minotaur Slayer” it would be a different ship. Yes, some planks may remain but the new version would be fundamentally different from the original. If I ask Claude for a prompt to break from a writers block period and then use that to expand into a full story then a morsel of the original prompt remains but that is dwarfed by my own ideas and words. It’s an issue of scale. If I read a newspaper article or an advertisement that inspires a story of mine, is that all that different from using a LLM to create a prompt to inspire a new story or essay? As long as the generated prompt remains only a small cell of the larger body of work then it would stand to reason the authorship of the piece is overwhelmingly human.
'The Ship of Theseus analogy works up to a point. By starting a piece with the aid of LLMs and then changing the text or expanding on the ideas presented in it, the author is creating a different “ship.” If you added two masts, put an outboard engine on it, repainted Theseus’ ship bright orange and christened it “Minotaur Slayer” it would be a different ship. Yes, some planks may remain but the new version would be fundamentally different from the original.'
That's the whole point of the Ship of Theseus: it is a thought experiment about how many things that one can change any the ship until, eventually, it is a different ship. One of the most persistent puzzles in philosophy is how to make sense of the relationship between identity and change. And the same logic of change over time applies to human edited LLMs as you right pointed out. The ancient meets the cutting edge. What a time to be alive!