On a tangent, I just want to mention that what is good writing is inherently subjective. There’s no such thing as good except as understood and judged by individual people for particular reasons. A literary critic can judge a work by many criteria, but the worth of a work is not some sort of some of its quality by all the criteria you can think of.
Yourwork may be the best in the world by your particular criteria, and that is valid and valuable.
Talking to an llm about your work can make it better by your criteria or by others criteria if that’s something you’re interested in.
But that’s separate then the primary role you’re describing of having an LLM read your work. That’s something like feeling seen, or feeling that your work is being seen. There’s a paradox in that you know there’s no other person there to see you, yet you feel seen.
I suggest that your feeling of being seen is correct, but the person seeing you is not the LLM but you. Its ability to analyze and understand is a tool you are using to see your writing and yourself more clearly.
I doubt that these arguments are correct. First of all, goodness as graded by different readers none of whom are the author oneself is likely to correlate. For example, if the proposed idea is obviously wrong, then a sufficiently smart reader is unlikely to grade it well (think of an army of rejected LessWrong posts or of this Wikipedia essay arguing that one cannot write a good Wikipedia article about an unnotable topic at all).
Additionally, I am confused by the second argument that “your feeling of being seen is correct, but the person seeing you is not the LLM but you”.
The LLM which saw your writing can recite facts which you never knew or come up with an entirely different perspective (think, e.g. of rebuttals by me and Claude Opus 4.5 to a post by JenniferRM; edit: Claude’s perspective differs both from mine and from Jennifer’s). As far as I understand, the LLM’s perspective is the one formed by cultural hegemony of the source of the training data, by the LLM’s prescribed role, etc.
What does the phrase “being seen” mean?
a. Receiving epistemically useful feedback, like having the LLM hunt down potential flaws?
b. Ensuring that another mind is influenced for a long time, which can be done with other humans, but is far harder or outright impossible with LLMs?
The sense in which I meant “seen” is different than all of those. It means, I think, understood and to some degree empathized with. In this thesis, the LLM is a tool that lets me understand and empathize better with myself. It is a common usage in some branches of current US culture. I don’t know if that’s what OP was feeling, but it’s something I’ve gotten from using LLMs for self-reflection in a vaguely therapeutic or self-work direction. It’s most similar to social validation. I’m saying you can get social validation without thinking the LLM is another being, by adopting its perspective and thereby becoming able to validate yourself. This seems useful although it’s also a component of “AI psychosis” if you overdo it.
Of course the other element is feeling like the LLM is a person, even while knowing/believing intellectuallly that it’s not. Some of that is probably happening too.
On the other claim:
The fact that we tend to share some judgments of goodness does not by any means indicate it’s a law of the universe, just that our culture transmits values and perspectives. We are speaking of fiction; rightness or wrongness plays little role. If most people judge a piece one way and a few another, the majority does not rule, for they are probably making judgments based on different criteria. It’s not one competition, it’s multiple separate categories of “goodness”.
I’m saying that what is “good” prose is entirely dependent on purpose and audience. What’s good prose for, say, causing distress in the maximum fraction of human readers is very different from prose intended to amuse fans of a certain genre or fictional world, and particular fans with particular tastes and viewpoints. OPs writing could be very bad by the first but very good by the second criteria.
Perhaps by “good” you mean good for average audiences and average purposes; I think that’s the common usage, but the claim I am making is that it’s both technically incorrect and that misunderstanding that is actively harmful.
We live in an IMO dangerously competitive and judgmental culture. I just like to remind people that it’s okay to like what they like and you shouldn’t let other people tell you it’s “crappy” or “bad” without carefully noticing the criteria they are (usually without their knowledge) applying.
So we’d have wide agreement on what’s particularly bad writing (something that accomplishes almost no purposes for almost anyone, like text that is almost meaningless) but a lot of disagreement on exactly what writing is good depending on the audience and purpose for which we’re evaluating it.
This same principle of quality can be extended to other areas of taste or judgment.
There are few things which are not highly subjective. Humans prefer to to not be physically uncomfortable in almost all circumstances. Beyond that, we have vastly different tastes. Those tastes become better or worse only when we apply criteria by which to judge.
I’ve spent this much time explaining because this principle is central to my understanding of life and culture, and I believe it has led me to be a happier and more creative person than I would be without it. I arrived at this by studying a little Zen philosophy and reading Zen and the Art of Motorcycle Maintenance, a work of dry philosophy disguised as fiction which takes “is quality just what you like?” as a central question and answers “yes, and what you like is correlated for deep reasons but not the same as what others like, and the word “just” serves no purpose but a misdirective pejorative in that question”.
On a tangent, I just want to mention that what is good writing is inherently subjective. There’s no such thing as good except as understood and judged by individual people for particular reasons. A literary critic can judge a work by many criteria, but the worth of a work is not some sort of some of its quality by all the criteria you can think of.
Yourwork may be the best in the world by your particular criteria, and that is valid and valuable.
Talking to an llm about your work can make it better by your criteria or by others criteria if that’s something you’re interested in.
But that’s separate then the primary role you’re describing of having an LLM read your work. That’s something like feeling seen, or feeling that your work is being seen. There’s a paradox in that you know there’s no other person there to see you, yet you feel seen.
I suggest that your feeling of being seen is correct, but the person seeing you is not the LLM but you. Its ability to analyze and understand is a tool you are using to see your writing and yourself more clearly.
I doubt that these arguments are correct. First of all, goodness as graded by different readers none of whom are the author oneself is likely to correlate. For example, if the proposed idea is obviously wrong, then a sufficiently smart reader is unlikely to grade it well (think of an army of rejected LessWrong posts or of this Wikipedia essay arguing that one cannot write a good Wikipedia article about an unnotable topic at all).
Additionally, I am confused by the second argument that “your feeling of being seen is correct, but the person seeing you is not the LLM but you”.
The LLM which saw your writing can recite facts which you never knew or come up with an entirely different perspective (think, e.g. of rebuttals by me and Claude Opus 4.5 to a post by JenniferRM; edit: Claude’s perspective differs both from mine and from Jennifer’s). As far as I understand, the LLM’s perspective is the one formed by cultural hegemony of the source of the training data, by the LLM’s prescribed role, etc.
What does the phrase “being seen” mean?
a. Receiving epistemically useful feedback, like having the LLM hunt down potential flaws?
b. Ensuring that another mind is influenced for a long time, which can be done with other humans, but is far harder or outright impossible with LLMs?
c. Social validation?
The sense in which I meant “seen” is different than all of those. It means, I think, understood and to some degree empathized with. In this thesis, the LLM is a tool that lets me understand and empathize better with myself. It is a common usage in some branches of current US culture. I don’t know if that’s what OP was feeling, but it’s something I’ve gotten from using LLMs for self-reflection in a vaguely therapeutic or self-work direction. It’s most similar to social validation. I’m saying you can get social validation without thinking the LLM is another being, by adopting its perspective and thereby becoming able to validate yourself. This seems useful although it’s also a component of “AI psychosis” if you overdo it.
Of course the other element is feeling like the LLM is a person, even while knowing/believing intellectuallly that it’s not. Some of that is probably happening too.
On the other claim:
The fact that we tend to share some judgments of goodness does not by any means indicate it’s a law of the universe, just that our culture transmits values and perspectives. We are speaking of fiction; rightness or wrongness plays little role. If most people judge a piece one way and a few another, the majority does not rule, for they are probably making judgments based on different criteria. It’s not one competition, it’s multiple separate categories of “goodness”.
I’m saying that what is “good” prose is entirely dependent on purpose and audience. What’s good prose for, say, causing distress in the maximum fraction of human readers is very different from prose intended to amuse fans of a certain genre or fictional world, and particular fans with particular tastes and viewpoints. OPs writing could be very bad by the first but very good by the second criteria.
Perhaps by “good” you mean good for average audiences and average purposes; I think that’s the common usage, but the claim I am making is that it’s both technically incorrect and that misunderstanding that is actively harmful.
We live in an IMO dangerously competitive and judgmental culture. I just like to remind people that it’s okay to like what they like and you shouldn’t let other people tell you it’s “crappy” or “bad” without carefully noticing the criteria they are (usually without their knowledge) applying.
So we’d have wide agreement on what’s particularly bad writing (something that accomplishes almost no purposes for almost anyone, like text that is almost meaningless) but a lot of disagreement on exactly what writing is good depending on the audience and purpose for which we’re evaluating it.
This same principle of quality can be extended to other areas of taste or judgment.
There are few things which are not highly subjective. Humans prefer to to not be physically uncomfortable in almost all circumstances. Beyond that, we have vastly different tastes. Those tastes become better or worse only when we apply criteria by which to judge.
I’ve spent this much time explaining because this principle is central to my understanding of life and culture, and I believe it has led me to be a happier and more creative person than I would be without it. I arrived at this by studying a little Zen philosophy and reading Zen and the Art of Motorcycle Maintenance, a work of dry philosophy disguised as fiction which takes “is quality just what you like?” as a central question and answers “yes, and what you like is correlated for deep reasons but not the same as what others like, and the word “just” serves no purpose but a misdirective pejorative in that question”.