Categorical Linguistics in Quanta
Posted by Tom Leinster
Quanta magazine has just published a feature on Tai-Danae Bradley and her work, entitled
Where Does Meaning Live in a Sentence? Math Might Tell Us.
The mathematician Tai-Danae Bradley is using category theory to try to understand both human and AI-generated language.
It’s a nicely set up Q&A, with questions like “What’s something category theory lets you see that you can’t otherwise?” and “How do you use category theory to understand language?”
Particularly interesting for me is the part towards the end where Bradley describes her work with Juan Pablo Vigneaux on magnitude of enriched categories of texts.
Posted at April 15, 2025 9:10 AM UTC
Re: Categorical Linguistics in Quanta
I’ve found myself marveling at the weird way that LLM’s are, in some modes, the undeniable heralds of the beauty of words, surpassing our poets and pirouetting in service of the least lithe of us.
But they also operate in a mode that moves the center-of-gravity between words and minds away from our minds. Will our minds end up orbiting the black box LLM?
Tai-Danae Bradley’s work might be the kind of analysis that can allow our minds to gain the insight often not available when working with an LLM and thereby better tether our own Language with our own minds.