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
How to resist the Fading Qualia Argument – Andreas Mogensen (Global Priorities Institute, University of Oxford)
The Fading Qualia Argument is perhaps the strongest argument supporting the view that in order for a system to be conscious, it does not need to be made of anything in particular, so long as its internal parts have the right causal relations to each other and to the system’s inputs and outputs. I show how the argument can be resisted given two key assumptions: that consciousness is associated with vagueness at its boundaries and that conscious neural activity has a particular kind of holistic structure. …
Misjudgment Exacerbates Collective Action Problems – Joshua Lewis (New York University) et al.
In collective action problems, suboptimal collective outcomes arise from each individual optimizing their own wellbeing. Past work assumes individuals do this because they care more about themselves than others. Yet, other factors could also contribute. We examine the role of empirical beliefs. Our results suggest people underestimate individual impact on collective problems. When collective action seems worthwhile, individual action often does not, even if the expected ratio of costs to benefits is the same. …
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.