AI for Mathematics: A Cognitive Science Perspective

Cedegao E. Zhang,Katherine M. Collins,Adrian Weller,Joshua B. Tenenbaum
2023-10-19
Abstract:Mathematics is one of the most powerful conceptual systems developed and used by the human species. Dreams of automated mathematicians have a storied history in artificial intelligence (AI). Rapid progress in AI, particularly propelled by advances in large language models (LLMs), has sparked renewed, widespread interest in building such systems. In this work, we reflect on these goals from a \textit{cognitive science} perspective. We call attention to several classical and ongoing research directions from cognitive science, which we believe are valuable for AI practitioners to consider when seeking to build truly human (or superhuman)-level mathematical systems. We close with open discussions and questions that we believe necessitate a multi-disciplinary perspective -- cognitive scientists working in tandem with AI researchers and mathematicians -- as we move toward better mathematical AI systems which not only help us push the frontier of the mathematics, but also offer glimpses into how we as humans are even capable of such great cognitive feats.
Neurons and Cognition,Artificial Intelligence
What problem does this paper attempt to address?
The problem this paper attempts to address is the construction of artificial intelligence systems capable of understanding and practicing mathematics at or beyond the level of human mathematicians. Despite significant progress made by large language models (LLMs) in many fields, their performance in mathematics remains lagging. The paper explores, from a cognitive science perspective, how to combine knowledge from cognitive science, artificial intelligence, and mathematics to advance the development of mathematical AI systems. This not only promotes research at the forefront of mathematics but also reveals how humans achieve such remarkable cognitive feats. Specifically, the paper focuses on the following aspects: 1. **Sample-efficient learning**: Investigating how humans learn new concepts, knowledge, and problem-solving strategies with limited data, which has important implications for the design of mathematical AI systems. 2. **Concepts, representations, and world models**: Analyzing how humans construct complex conceptual representations through core knowledge systems (such as "number sense") and form rich models of the world. 3. **Goals, planning, and agency**: Discussing the planning and goal-oriented nature of human problem-solving and how to incorporate these characteristics into AI systems. 4. **Cognitive limitations and resource rationality**: Studying human rational behavior under resource constraints and the impact of these limitations on AI system design. 5. **Communication and explanation**: Emphasizing the importance of effective communication within the mathematical community and how communication can facilitate learning and understanding. The paper argues that through interdisciplinary collaboration, combining knowledge from cognitive science, artificial intelligence, and mathematics, more flexible and efficient automated mathematician systems can be developed.