Wonderful article. Thank you for sharing your thoughts and understanding. The big question is "when"? I hadn't had enough time to play with Fable before they pulled it down, but I was seeing a huge increase in its ability to identify weak spots in my prose that I need to rewrite - and that's without the emotional context you are talking about. How long till we see a model that "feels"?
Something that takes me by surprise: CHATGPT overwhelming me to the point that tears well. Because? It sees the point I'm making. Granted it's building off a year of "working together." Still:, 'Huh.' And it's not as if I'm using a paid model.
Yes, I experience the same thing. It may not be "feeling" emotions, but it sure can tell me where the emotional payoffs in my story exist - which I then confirm via my experience or via a human reader. I'd say the current models are pretty accurate, maybe they correctly identify the strong and weak spots 75% of the time?
It's to the point I do not go the "well" all that often. And when I do, OMG, it sings. It's like a mirror someone removed a cloth from, in which it and I see where I'm going. I do the writing. It does the this is what you wrote. Yes, I ask for to see what it sees, then wrangled my own horses into the corral.
The sycophancy point ... a beta-reader who's really bored can't fake enthusiasm convincingly for long, but an AI optimized to be helpful absolutely can, and that's the real limitation right now, not intelligence. I use AI constantly for structural feedback and marketing, but the moment I need to know whether a scene actually landed emotionally, I still hand it to a person who can get bored, distracted, or moved against their will. That involuntary quality is the whole test.
Very interesting. I recently edited the hell out of a manuscript using AI, but I wasn't happy until I'd sent it out to Beta readers and got their feedback. Only then did I know I'd done everything right. People definitely count when it comes to letting you know if you've hit the heart of a book right.
AI can be incredibly useful for structure, continuity, pacing, weak spots, and seeing patterns I might miss. But I still don’t think of it as a reader in the full human sense. It can tell me what the text appears to be doing. It can’t truly tell me what it felt like to sit with it.
For me, working with AI means I have to know very clearly where I want the story to go. The creative part is often not saying yes to everything the tool offers, but saying no: not this voice, not this direction, not this shortcut, not this version.
When there are suddenly so many possible paths, authorship becomes the act of choosing what should stay — and what has to disappear, even if it is perfectly passable.
That human centre is not a side note for me. It is the point. Taste, doubt, rejection, revision, and responsibility are what turn possibility into a finished piece of work.
Wonderful article. Thank you for sharing your thoughts and understanding. The big question is "when"? I hadn't had enough time to play with Fable before they pulled it down, but I was seeing a huge increase in its ability to identify weak spots in my prose that I need to rewrite - and that's without the emotional context you are talking about. How long till we see a model that "feels"?
Something that takes me by surprise: CHATGPT overwhelming me to the point that tears well. Because? It sees the point I'm making. Granted it's building off a year of "working together." Still:, 'Huh.' And it's not as if I'm using a paid model.
Yes, I experience the same thing. It may not be "feeling" emotions, but it sure can tell me where the emotional payoffs in my story exist - which I then confirm via my experience or via a human reader. I'd say the current models are pretty accurate, maybe they correctly identify the strong and weak spots 75% of the time?
It's to the point I do not go the "well" all that often. And when I do, OMG, it sings. It's like a mirror someone removed a cloth from, in which it and I see where I'm going. I do the writing. It does the this is what you wrote. Yes, I ask for to see what it sees, then wrangled my own horses into the corral.
The sycophancy point ... a beta-reader who's really bored can't fake enthusiasm convincingly for long, but an AI optimized to be helpful absolutely can, and that's the real limitation right now, not intelligence. I use AI constantly for structural feedback and marketing, but the moment I need to know whether a scene actually landed emotionally, I still hand it to a person who can get bored, distracted, or moved against their will. That involuntary quality is the whole test.
Very interesting. I recently edited the hell out of a manuscript using AI, but I wasn't happy until I'd sent it out to Beta readers and got their feedback. Only then did I know I'd done everything right. People definitely count when it comes to letting you know if you've hit the heart of a book right.
This distinction makes a lot of sense to me.
AI can be incredibly useful for structure, continuity, pacing, weak spots, and seeing patterns I might miss. But I still don’t think of it as a reader in the full human sense. It can tell me what the text appears to be doing. It can’t truly tell me what it felt like to sit with it.
For me, working with AI means I have to know very clearly where I want the story to go. The creative part is often not saying yes to everything the tool offers, but saying no: not this voice, not this direction, not this shortcut, not this version.
When there are suddenly so many possible paths, authorship becomes the act of choosing what should stay — and what has to disappear, even if it is perfectly passable.
That human centre is not a side note for me. It is the point. Taste, doubt, rejection, revision, and responsibility are what turn possibility into a finished piece of work.
Feeding my plot into Chat returns to me stakes in my story I hadn't seen before. Gold.