Memory Augmented Large Language Models Are Computationally Universal

Very interesting. Cognition and Turing equivalence are not things I automatically assume to be the same, so the paper's title is a question I hadn't stopped to consider previously.

Makes intuitive sense, though.