The unreasonable effectiveness of pattern matching

It has become common to explain the abilities of large language models (LLMs) as “mere” pattern matching. Because pattern matching is thought to be fragile and highly dependent on having exactly right training data, real thinking and reasoning of the kind that humans do is assumed to be implemented by qualitatively different mechanisms. I am going to argue that our intuitions about the limits of pattern matching are mistaken. At the core of this argument is a demonstration of an astonishing ability of LLMs to make sense of "Jabberwocky" language--texts in which most or all content words have been randomly replaced by nonsense strings--e.g., translating "He dwushed a ghanc zawk" to "He dragged a spare chair". This ability highlights the unreasonable effectiveness of pattern matching and suggests that pattern matching is not an alternative to real intelligence, but its key ingredient.

(This talk is part of the weekly Machine Learning Lunch Meetings (MLLM), held every Tuesday from 12:15 to 1:15 p.m.  Professors from Computer Sciences, Statistics, ECE, the iSchool, and other departments will discuss their latest research in machine learning, covering both theory and applications. This is a great opportunity to network with faculty and fellow researchers, learn about cutting-edge research at our university, and foster new collaborations. For the talk schedule, please visit https://sites.google.com/view/wiscmllm/home. To receive future weekly talk announcements, please subscribe to our UW Google Group at https://groups.google.com/u/1/a/g-groups.wisc.edu/g/mllm.)