Source-linked AI summary
The Debate Over Understanding in AI's Large Language Models
Melanie Mitchell, David C. Krakauer
TL;DR
The paper examines whether large language models understand language and the physical and social situations it describes. It surveys arguments on both sides and proposes developing a broader science of intelligence focused on distinct forms of understanding and their integration.
Problem
The paper addresses whether large language models understand natural language and the physical and social situations language can describe.
Method
The paper surveys arguments for and against attributing understanding to large pre-trained language models and considers available cognitive-science approaches.
Results
Large language models can display theory-of-mind performance, humanlike reasoning abilities, conversation, and apparent reasoning despite lacking conceptual understanding needed for humanlike functional language abilities.
Takeaways & Limitations
Progress requires new benchmarks and probing methods to illuminate mechanisms of diverse forms of intelligence and support broader conceptions of understanding in humans and machines.
Takeaways & Limitations
Current cognitive-science-based methods for gaining insight into diverse forms of intelligence and understanding are limited.
Abstract
from arXiv · showhide
We survey a current, heated debate in the AI research community on whether large pre-trained language models can be said to "understand" language -- and the physical and social situations language encodes -- in any important sense. We describe arguments that have been made for and against such understanding, and key questions for the broader sciences of intelligence that have arisen in light of these arguments. We contend that a new science of intelligence can be developed that will provide insight into distinct modes of understanding, their strengths and limitations, and the challenge of integrating diverse forms of cognition.