To say nothing of the fact that an LLM is a kind of model of "culture" in the abstract, and I don't know that we've ever had this accessible a way to feel out the vibe of a whole corpus. They're super interesting objects of study, but that doesn't feel worth destroying the economics of art
November 16, 2025 at 8:57 PM
To say nothing of the fact that an LLM is a kind of model of "culture" in the abstract, and I don't know that we've ever had this accessible a way to feel out the vibe of a whole corpus. They're super interesting objects of study, but that doesn't feel worth destroying the economics of art
For me, it's always been about expanding both the range of things computers can do, and also who gets to specify how they behave. With LLMs, it feels like that's much more possible than before. There's so much room to build this kind of AI with actual consent, for purposes other than rent seeking.
November 16, 2025 at 8:55 PM
For me, it's always been about expanding both the range of things computers can do, and also who gets to specify how they behave. With LLMs, it feels like that's much more possible than before. There's so much room to build this kind of AI with actual consent, for purposes other than rent seeking.
I agree completely! Some of my work is actually focused on the dataset construction side of this, and I work with materials scientists that do similar (I think?) gaussian process work
November 7, 2025 at 7:40 PM
I agree completely! Some of my work is actually focused on the dataset construction side of this, and I work with materials scientists that do similar (I think?) gaussian process work
It's been really frustrating to see how much this style of "hypothesis generation" work is in AI for science circles. I think there are good ways to use AI in materials science! But so much of the oxygen is taken up by end-to-end, speculative work :/
November 7, 2025 at 7:19 PM
It's been really frustrating to see how much this style of "hypothesis generation" work is in AI for science circles. I think there are good ways to use AI in materials science! But so much of the oxygen is taken up by end-to-end, speculative work :/
This is such a neat idea! I wonder how the infrastructure of gating carries over to modalities that are less identity carrying - like, is there an equivalent of this for text? Is there an analogue of signing your input tokens with your private key?
October 29, 2025 at 6:50 PM
This is such a neat idea! I wonder how the infrastructure of gating carries over to modalities that are less identity carrying - like, is there an equivalent of this for text? Is there an analogue of signing your input tokens with your private key?
I think it's even more exciting than this!! There's a bunch of stuff conveyed in _not_ text (i.e. diagrams/formulas etc as you point out), but also a lot of stuff contained in _not quite_ text — formatting, layout, all sorts of fun visual pragmatics that do convey information and are important
October 20, 2025 at 3:57 PM
I think it's even more exciting than this!! There's a bunch of stuff conveyed in _not_ text (i.e. diagrams/formulas etc as you point out), but also a lot of stuff contained in _not quite_ text — formatting, layout, all sorts of fun visual pragmatics that do convey information and are important