Understanding Foundation Models: Are We Back in 1924?

TL;DR

This paper critically examines the development of Foundation Models in AI, highlighting their reasoning capabilities, training techniques, and the challenges in understanding their inner workings compared to the human brain.

Contribution

It offers a position on the nature of FMs, emphasizing the role of novel training methods over size and discussing the difficulties in benchmarking and interpreting their functions.

Findings

Reasoning abilities are linked to training techniques, not size.

Benchmarking FMs remains a significant challenge.

FMs share some structural similarities with the human brain but differ fundamentally.

Abstract

This position paper explores the rapid development of Foundation Models (FMs) in AI and their implications for intelligence and reasoning. It examines the characteristics of FMs, including their training on vast datasets and use of embedding spaces to capture semantic relationships. The paper discusses recent advancements in FMs' reasoning abilities which we argue cannot be attributed to increased model size but to novel training techniques which yield learning phenomena like grokking. It also addresses the challenges in benchmarking FMs and compares their structure to the human brain. We argue that while FMs show promising developments in reasoning and knowledge representation, understanding their inner workings remains a significant challenge, similar to ongoing efforts in neuroscience to comprehend human brain function. Despite having some similarities, fundamental differences between FMs and the structure of human brain warn us against making direct comparisons or expecting neuroscience to provide immediate insights into FM function.