Research

My current work focuses on epistemic norms, skepticism, virtual reality, machine mentality, and the philosophy of artificial intelligence. The projects below are works in progress.

Epistemic Norms and Epistemic Evaluations

Under review

This project addresses the tension between traditional evidential epistemic norms and zetetic norms that guide inquiry. It argues that traditional epistemic norms primarily help us evaluate other agents in an epistemic network rather than guide epistemic actions that are often spontaneous and involuntary. Drawing on a distinction between knowledge generation and knowledge distribution, the project places the two kinds of norms in different normative domains.

Virtual-Inclusive Categories and the Simulation Hypothesis

Under review

David Chalmers argues that virtual-inclusive categories allow the simulation hypothesis to avoid a Putnam-style semantic argument against brains in vats. This project develops a functional account of virtual-inclusive artifact categories in place of a structural account. Because the relevant functional properties are observer-relative, it argues that simulations are not symmetric with the external world and that both hypotheses remain answerable to causal constraints on reference.

Understanding How It Works Defeats Mental Attribution

Under review

This project proposes a mechanistic opacity condition for justified attributions of mentality: we are justified in attributing mentality only to systems whose internal workings we cannot fully understand. For a mechanistically transparent system, a mechanistic explanation will be preferable to a mental explanation. The account supports skepticism about important attributions of consciousness, intelligence, intentionality, and mentality to current AI systems while avoiding biological chauvinism.

Alignment Is All You Need: No Catastrophic Risk to Worry at the Intentional Level

In preparation

This project examines two theories of machine desire and argues that meaningful interpretations of a machine’s desires ultimately derive from the desires of its developers. On this view, AI safety concerns should be addressed at the level of system design rather than at an independent intentional level.

The Personal Identity Assumption in External World Skeptical Scenarios

In preparation

Dreams, brain-in-a-vat cases, and simulations often assume that the subject in the real scenario is numerically identical to the subject of the first-person experience. This project argues that prominent theories of personal identity do not support that assumption. If the identity relation fails, standard external-world skeptical scenarios do not generate the epistemological threat they are usually taken to pose.