Towards shutdownable agents via stochastic choice
Elliott Thornley (Global Priorities Institute, University of Oxford), Alexander Roman (New College of Florida), Christos Ziakas (Imperial College, London), Leyton Ho (Brown University) and Louis Thomson (University of Oxford)
GPI Working Paper No. 16-2024
The Incomplete Preferences Proposal (IPP) is an idea for ensuring that advanced artificial agents never resist shutdown. 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 provide some initial evidence that DReST reward functions could train advanced agents to be USEFUL and NEUTRAL. Our theoretical work suggests that these agents would be useful and shutdownable.
Other working papers
Longtermism in an Infinite World – Christian J. Tarsney (Population Wellbeing Initiative, University of Texas at Austin) and Hayden Wilkinson (Global Priorities Institute, University of Oxford)
The case for longtermism depends on the vast potential scale of the future. But that same vastness also threatens to undermine the case for longtermism: If the future contains infinite value, then many theories of value that support longtermism (e.g., risk-neutral total utilitarianism) seem to imply that no available action is better than any other. And some strategies for avoiding this conclusion (e.g., exponential time discounting) yield views that…
Dispelling the Anthropic Shadow – Teruji Thomas (Global Priorities Institute, University of Oxford)
There are some possible events that we could not possibly discover in our past. We could not discover an omnicidal catastrophe, an event so destructive that it permanently wiped out life on Earth. Had such a catastrophe occurred, we wouldn’t be here to find out. This space of unobservable histories has been called the anthropic shadow. Several authors claim that the anthropic shadow leads to an ‘observation selection bias’, analogous to survivorship bias, when we use the historical record to estimate catastrophic risks. …
‘The only ethical argument for positive 𝛿’? – Andreas Mogensen (Global Priorities Institute, Oxford University)
I consider whether a positive rate of pure intergenerational time preference is justifiable in terms of agent-relative moral reasons relating to partiality between generations, an idea I call discounting for kinship. I respond to Parfit’s objections to discounting for kinship, but then highlight a number of apparent limitations of this…