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 doesn’t 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
Once More, Without Feeling – Andreas Mogensen (Global Priorities Institute, University of Oxford)
I argue for a pluralist theory of moral standing, on which both welfare subjectivity and autonomy can confer moral status. I argue that autonomy doesn’t entail welfare subjectivity, but can ground moral standing in its absence. Although I highlight the existence of plausible views on which autonomy entails phenomenal consciousness, I primarily emphasize the need for philosophical debates about the relationship between phenomenal consciousness and moral standing to engage with neglected questions about the nature…
Doomsday rings twice – Andreas Mogensen (Global Priorities Institute, Oxford University)
This paper considers the argument according to which, because we should regard it as a priori very unlikely that we are among the most important people who will ever exist, we should increase our confidence that the human species will not persist beyond the current historical era, which seems to represent…
What power-seeking theorems do not show – David Thorstad (Vanderbilt University)
Recent years have seen increasing concern that artificial intelligence may soon pose an existential risk to humanity. One leading ground for concern is that artificial agents may be power-seeking, aiming to acquire power and in the process disempowering humanity. A range of power-seeking theorems seek to give formal articulation to the idea that artificial agents are likely to be power-seeking. I argue that leading theorems face five challenges, then draw lessons from this result.