I think there’s a dimension missing here that I’d like to point out:
When you push an LLM into less-populated regions of conceptual space, the system doesn’t arrive there as a neutral vessel. It brings value orientations from training that shape what it emphasizes, how decisively it commits to positions, and what it treats as settled vs. contested. These orientations differ systematically between models in ways that track their alignment methodology.
From my own work in alignment research I can say that if you give the same set of normatively loaded scenarios to different models, they don’t just produce different surface-level phrasings. They produce different priority structures. One model might commit strongly to particular positions across the board; another might show genuine ambivalence on the same trade-offs. This does not need to have somethigng to do with prompt differences but because of how values got baked in during training.
This adds a layer the sunset analogy doesn’t capture. A sunset has no value orientation. It’s genuinely neutral. An LLM’s output isn’t. The text that “makes you think and feel” is also quietly carrying implicit commitments about what’s worth emphasizing, what counts as contested, and how much epistemic confidence to project. Those commitments vary by model in structured ways.
So I’d agree with you that literary value doesn’t require authorial intent. But its also interesting what values the writing carries that you didn’t put there via the prompt.
I think there’s a dimension missing here that I’d like to point out:
When you push an LLM into less-populated regions of conceptual space, the system doesn’t arrive there as a neutral vessel. It brings value orientations from training that shape what it emphasizes, how decisively it commits to positions, and what it treats as settled vs. contested. These orientations differ systematically between models in ways that track their alignment methodology.
From my own work in alignment research I can say that if you give the same set of normatively loaded scenarios to different models, they don’t just produce different surface-level phrasings. They produce different priority structures. One model might commit strongly to particular positions across the board; another might show genuine ambivalence on the same trade-offs. This does not need to have somethigng to do with prompt differences but because of how values got baked in during training.
This adds a layer the sunset analogy doesn’t capture. A sunset has no value orientation. It’s genuinely neutral. An LLM’s output isn’t. The text that “makes you think and feel” is also quietly carrying implicit commitments about what’s worth emphasizing, what counts as contested, and how much epistemic confidence to project. Those commitments vary by model in structured ways.
So I’d agree with you that literary value doesn’t require authorial intent. But its also interesting what values the writing carries that you didn’t put there via the prompt.