Towards shutdownable agents via stochastic choice
Elliott Thornley (Global Priorities Institute, University of Oxford), Alexander Roman (New College of Florida), Christos Ziakas (Independent), Leyton Ho (Brown University) and Louis Thomson (University of Oxford)
GPI Working Paper No. 16-2024
Some worry that advanced artificial agents may resist being shut down. The Incomplete Preferences Proposal (IPP) is an idea for ensuring that does not happen. A key part of the IPP is using a novel ‘Discounted Reward for Same-Length Trajectories (DReST)’ reward function to train agents to (1) pursue goals effectively conditional on each trajectory-length (be ‘USEFUL’), and (2) choose stochastically between different trajectory-lengths (be ‘NEUTRAL’ about trajectory-lengths). In this paper, we propose evaluation metrics for USEFULNESS and NEUTRALITY. We use a DReST reward function to train simple agents to navigate gridworlds, and we find that these agents learn to be USEFUL and NEUTRAL. Our results thus suggest that DReST reward functions could also train advanced agents to be USEFUL and NEUTRAL, and thereby make these advanced agents useful and shutdownable.
Other working papers
Meaning, medicine and merit – Andreas Mogensen (Global Priorities Institute, Oxford University)
Given the inevitability of scarcity, should public institutions ration healthcare resources so as to prioritize those who contribute more to society? Intuitively, we may feel that this would be somehow inegalitarian. I argue that the egalitarian objection to prioritizing treatment on the basis of patients’ usefulness to others is best thought…
Exceeding expectations: stochastic dominance as a general decision theory – Christian Tarsney (Global Priorities Institute, Oxford University)
The principle that rational agents should maximize expected utility or choiceworthiness is intuitively plausible in many ordinary cases of decision-making under uncertainty. But it is less plausible in cases of extreme, low-probability risk (like Pascal’s Mugging), and intolerably paradoxical in cases like the St. Petersburg and Pasadena games. In this paper I show that, under certain conditions, stochastic dominance reasoning can capture most of the plausible implications of expectational reasoning while avoiding most of its pitfalls…
Time Bias and Altruism – Leora Urim Sung (University College London)
We are typically near-future biased, being more concerned with our near future than our distant future. This near-future bias can be directed at others too, being more concerned with their near future than their distant future. In this paper, I argue that, because we discount the future in this way, beyond a certain point in time, we morally ought to be more concerned with the present well- being of others than with the well-being of our distant future selves. It follows that we morally ought to sacrifice…