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 POST-Agents Proposal (PAP) is an idea for ensuring that advanced artificial agents never resist shutdown. A key part of the PAP 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
The end of economic growth? Unintended consequences of a declining population – Charles I. Jones (Stanford University)
In many models, economic growth is driven by people discovering new ideas. These models typically assume either a constant or growing population. However, in high income countries today, fertility is already below its replacement rate: women are having fewer than two children on average. It is a distinct possibility — highlighted in the recent book, Empty Planet — that global population will decline rather than stabilize in the long run. …
Quadratic Funding with Incomplete Information – Luis M. V. Freitas (Global Priorities Institute, University of Oxford) and Wilfredo L. Maldonado (University of Sao Paulo)
Quadratic funding is a public good provision mechanism that satisfies desirable theoretical properties, such as efficiency under complete information, and has been gaining popularity in practical applications. We evaluate this mechanism in a setting of incomplete information regarding individual preferences, and show that this result only holds under knife-edge conditions. We also estimate the inefficiency of the mechanism in a variety of settings and show, in particular, that inefficiency increases…
‘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…