Can We Understand How Large Language Models Reason? – Communications of the ACM
Some of the data collected by this provider is for the purposes of personalization and measuring advertising effectiveness. Some of the data collected by this provider is for the purposes of personalization and measuring advertising effectiveness. Membership in ACM includes a subscription to Communications of the ACM (CACM), the computing industry's most trusted source for staying connected to the world of advanced computing. Researchers are using tools from causality theory to understand the keys to an algorithm working at a higher level of abstraction. Credit: Kravchenko Tetiana / Getty Images Large language models can write essays, solve math problems, and generate computer code, but it’s not fully understood how they do it. Researchers can observe the billions of parameters inside these systems changing during training, yet the internal logic of the models remains largely hidden. In a sense, the engineering is ahead of the science. Can science catch up and make LLMs and other deep