Research Projects
Interpersonal Concepts and AI
We are often encouraged describe AI with human-like language: AI "reasons"; AI is "friendly","courteous", or "sycophantic"; we "trust" or "distrust" AI. In ongoing empirical projects, I am exploring with colleagues at the University of Kent the extent to which people's understanding of interpersonal concepts such as friendship, trust, and labor mean these concepts can be felicitously applied to AI. Future work will take the further step and ask whether our findings suggest need to rethink our relationship with AI by changing our language to better demarcate the differences between our relationships with humans our relationship with and AI.
​
​
Reasons and Deference in Moral Advice
We often ask others for moral advice, but not all ways of learning morality from others is equally responsible. Many philosophers argue we should not defer to others on matters of morality. That is, we should not believe we ought to do something merely because someone tells us that is the case. It is, however, usually responsible to evaluate the moral reasons they provide and change our minds accordingly. In this project, I am measuring whether a) people defer to AI-generated moral advice and b) change their minds based on the reasons contained in AI-generated moral advice.
​
For more, see my blog post at the APA: Learning from AI's Bullshit
Virtual Trolley Cases
and
Written Trolley Cases
People do not make the same judgements in reaction to written trolley case as they do trolley cases in VR. This project is asking what this says about the epistemology of thought experiments and VR. Using qualitative analysis, Kathryn Francis and I are testing how professional philosophers react to VR trolley cases. See my post at the Junkyard for the inspiration for the project and some of my thoughts on the epistemic virtues of thought experiments.
Experimental Conceptual Engineering
At rock bottom, conceptual engineering is about improving how people talk or think. This project explores how conceptual engineering needs to be an empirical process in order to succeed. One strand is developing an experimental framework to test conceptual revision and replacement. Another strand is arguing, from the armchair, that all conceptual engineering projects need to be sensitive to empirical issues.
​​
See my trilogy of papers on the topic:
1) The need for experimental conceptual engineering
2) Inventing experimental implementation