The Artificial Scientist: Logicist, Emergentist, and Universalist Approaches to Artificial General Intelligence
Michael Timothy Bennett, Yoshihiro Maruyama
TL;DR
The paper investigates what it would take to build an Artificial Scientist, arguing that no single AGI paradigm suffices. It compares Logicist, Emergentist, and Universalist approaches, outlining their strengths, limitations, and implications for hypothesis formation, reasoning, and explainability. The authors advocate a unifying perspective that combines these paradigms, drawing on Kantian cognitive architecture and the notion of solutions to any task to achieve general, interpretable, and controllable scientific reasoning. This hybrid framework aims to enable efficient hypothesis generation, robust causal understanding, and natural-language explainability, with practical significance for advancing AGI research and automated scientific discovery.
Abstract
We attempt to define what is necessary to construct an Artificial Scientist, explore and evaluate several approaches to artificial general intelligence (AGI) which may facilitate this, conclude that a unified or hybrid approach is necessary and explore two theories that satisfy this requirement to some degree.