Artificial Philosophy: Using AI Systems to Support Philosophical Research

Recent advances in artificial intelligence, particularly large language models (LLMs), present novel opportunities for philosophical research. In this talk, I will outline some of these opportunities, focusing on the effort to build 'philosophically-minded' LLMs. The talk will include a discussion of recent work to build tailored LLMs centred on the works of philosophers, such as Andy Clark, Daniel Dennett, and Keith Frankish. In addition to showing how LLMs can be used to support the exploration of philosophical ideas, I will discuss how the practical project to build such systems alters our thinking in regard to prominent philosophical problems, such as the problem of constitutive relevance (in the philosophy of science) and our philosophical understanding of the extended mind. In addition, the talk will explore how suitably configured LLMs might be used to build AI systems that assist with the effort to understand the mind from a predictive processing (or active inference) standpoint. 

One of the offshoots of this work is a novel account of trustworthiness that accommodates existing accounts of human trustworthiness (in philosophy), while also informing our understanding of how to build trustworthy AI systems (in computer science). The upshot is a potential blurring of the traditional disciplinary boundaries between philosophy and engineering. While the project to build 'philosophically-minded' LLMs is easily understood as a form of engineering, it is also, I suggest, a form of philosophical practice. At a minimum, this marks a point of techno-philosophical interest that provides opportunities for interdisciplinary collaboration.

More broadly, however, AI systems raise important questions about the nature of philosophical inquiry and the methodologies used to tackle philosophical problems. By exploring these issues, the talk aims to illuminate the evolving relationship between philosophy and AI, highlighting both its theoretical significance and its practical implications for the future of philosophical research.