I feel like the AI is unduly helped by the format here. I don’t actually consume a whole book one quote at a time, and that is also not the reading experience the author wrote for.
For example, with the literary fiction one, the story about the boy and the church feels sort of complete, like a micro-story, and I think that might be why I prefer it in isolation. You also don’t need to understand anything about these characters before reading, they are just the boy and the grandfather, and that is all they are and all you need to know. But with the quote from the judge, I have this feeling that there is more to this character, even without ever having read the book the quote is from, and that makes me feel a bit disoriented while reading. Am I supposed to agree with this guy? To root for him? Does the wider narrative disprove him? I don’t know, and that makes the quote hard to read.
But I don’t read novels to read fifty cool quotes in a row, I want cohesion, a narrative, character development, some sort of central thesis. Can AIs currently provide that? I doubt it, and this article doesn’t ask.
And I suspect that part of this is also about whether you read the book the human quotes are based on. I don’t know who that judge is, but maybe my disorientation would just disappear if I did, and then the quote would land better. So maybe the survey results are just dominated by people who feel confused about the actual book that sits around the human quotes, and the whole thing is more about how well you know the literary canon they draw from.
I feel like the AI is unduly helped by the format here. I don’t actually consume a whole book one quote at a time, and that is also not the reading experience the author wrote for.
For example, with the literary fiction one, the story about the boy and the church feels sort of complete, like a micro-story, and I think that might be why I prefer it in isolation. You also don’t need to understand anything about these characters before reading, they are just the boy and the grandfather, and that is all they are and all you need to know. But with the quote from the judge, I have this feeling that there is more to this character, even without ever having read the book the quote is from, and that makes me feel a bit disoriented while reading. Am I supposed to agree with this guy? To root for him? Does the wider narrative disprove him? I don’t know, and that makes the quote hard to read.
But I don’t read novels to read fifty cool quotes in a row, I want cohesion, a narrative, character development, some sort of central thesis. Can AIs currently provide that? I doubt it, and this article doesn’t ask.
And I suspect that part of this is also about whether you read the book the human quotes are based on. I don’t know who that judge is, but maybe my disorientation would just disappear if I did, and then the quote would land better. So maybe the survey results are just dominated by people who feel confused about the actual book that sits around the human quotes, and the whole thing is more about how well you know the literary canon they draw from.
(I had 3⁄2 AI preference, 1,1 2,1 3,2 4,1 5,1)