Against the Current Debate on Digital Suffering

During this presentation, I will argue against the fruitfulness of the current debate concerning digital suffering in the philosophy of artificial intelligence. Rather than contesting the three assumptions on which this debate rests, the essay grants all of them (A1: one day we will be able to create digital systems with conscious experience; A2: if A1 is true, then digital minds might be able to suffer; A3: if A2 is true, we have to address this in some way) and argues that the debate is nonetheless structurally problematic. Two problems are identified. First (P1), we do not know when and how digital minds will be created (Saad and Bradley, 2022; Dung, 2023) and, given our complete lack of familiarity with anything resembling this scenario (unlike nuclear war or engineered pandemics), we are unable to formulate non-trivial proposals for dealing with it. Second (P2), there is a serious version of the problem of mindreading that needs to be addressed (Saad and Bradley, 2022): since digital minds will likely differ from us mentally and physically as much as or more than the most exotic animals, the standard theories of mindreading (Theory-Theory, Simulation Theory, and Direct Perception) may be insufficient to detect when a digital mind suffers. The conclusion is that the debate concerning digital suffering should either be suspended until more elements become available on which to base non-trivial proposals or radically restructured. References Dung, L. (2023) ‘How to deal with risks of AI suffering’, Inquiry: An Interdisciplinary Journal of Philosophy, 68(7), pp. 2281–2309. Saad, B. and Bradley, A. (2022) ‘Digital suffering: why it’s a problem and how to prevent it’, Inquiry: An Interdisciplinary Journal of Philosophy, 68(7), pp. 2110–2145.