PhD Studentship

Project summary

Recent advances in artificial intelligence (AI) have fuelled the idea of building “AI scientists”, i.e., systems that can autonomously perform part or all of the research process. However, to realise this vision AI systems need specific abilities. One of these abilities is troubleshooting: experiments often go wrong and researchers constantly need to deal with and learn from experimental error. Thus, if AI systems are supposed to perform research autonomously, they need to have error-reasoning ability. This fully funded PhD Studentship forms part of the Error-Reasoning Agents (ERAs) project, which aims to develop the tools required for assessing the error-reasoning ability of AI systems and to thereby foster the careful and effective introduction of AI agents into the research process.

The goal of this PhD project is to develop a novel account of what it means to learn from error in science. In what sense is science “self-correcting”, and what does this process entail? How are mistakes or errors used to advance the research process? Answering these fundamental questions will not only represent a key contribution to debates in philosophy of science but also provide a foundation for thinking about how novel AI tools might help or hinder the research process. In the long term, the research will also help inform the design of novel ERAs.

The PhD candidate will be based at the Egenis Centre for the Study of the Life Sciences, a leading centre for interdisciplinary research on the biosciences and related disciplines. Egenis has a long tradition of pursuing a practice-focused approach to philosophy and offers a lively and diverse academic community that brings together philosophers, scientists, and social scientists.

Specific research tasks the candidate will pursue include the analysis of historical case studies from the biological sciences; engagement with current literature in philosophy of science; as well as field work in bioscience laboratories at the University of Exeter. The student will also have funding available to travel to conferences to present and discuss their findings.

Suitable candidates will preferably have a background in both philosophy and the natural sciences, but the latter is not required. An interest in interdisciplinary communication and engagement as well as an understanding of how AI is currently deployed in science are desirable but not essential.

Advert information