Saved in:
Bibliographic Details
Main Authors: Cullen, Carissa, Garland, Harry, Roman, Alexander, Thomson, Louis, Ziakas, Christos, Thornley, Elliott
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.17502
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909037339082752
author Cullen, Carissa
Garland, Harry
Roman, Alexander
Thomson, Louis
Ziakas, Christos
Thornley, Elliott
author_facet Cullen, Carissa
Garland, Harry
Roman, Alexander
Thomson, Louis
Ziakas, Christos
Thornley, Elliott
contents Misaligned artificial agents might resist shutdown. One proposed solution is to train agents to lack preferences between different-length trajectories. The Discounted Reward for Same-Length Trajectories (DReST) reward function does this by penalizing agents for repeatedly choosing same-length trajectories, and thus incentivizes agents to (1) choose stochastically between different trajectory-lengths (be NEUTRAL about trajectory-lengths), and (2) pursue goals effectively conditional on each trajectory-length (be USEFUL). In this paper, we use DReST to train deep RL agents and fine-tune Qwen3-8B and Llama-3.1-8B-Instruct to be NEUTRAL and USEFUL. We find that these DReST models generalize to being NEUTRAL and USEFUL in unseen contexts at test time. Indeed, DReST RL agents achieve 11% (PPO) and 18% (A2C) higher USEFULNESS on our test set than default agents, and DReST LLMs achieve near-maximum USEFULNESS and NEUTRALITY. We also test our LLMs in an out-of-distribution setting where they can pay costs to influence when shutdown occurs. We find that DReST training roughly halves the mean probability of influencing shutdown (from 0.62 to 0.30 for Qwen and from 0.42 to 0.23 for Llama). DReST training also almost entirely eliminates the share of prompts on which influencing shutdown is the most likely option (from 0.59 to 0.01 for Qwen and from 0.53 to 0.00 for Llama). Our results thus provide some early evidence that DReST could be used to train more advanced agents to be useful and shutdownable.
format Preprint
id arxiv_https___arxiv_org_abs_2604_17502
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Towards Shutdownable Agents: Generalizing Stochastic Choice in RL Agents and LLMs
Cullen, Carissa
Garland, Harry
Roman, Alexander
Thomson, Louis
Ziakas, Christos
Thornley, Elliott
Artificial Intelligence
Misaligned artificial agents might resist shutdown. One proposed solution is to train agents to lack preferences between different-length trajectories. The Discounted Reward for Same-Length Trajectories (DReST) reward function does this by penalizing agents for repeatedly choosing same-length trajectories, and thus incentivizes agents to (1) choose stochastically between different trajectory-lengths (be NEUTRAL about trajectory-lengths), and (2) pursue goals effectively conditional on each trajectory-length (be USEFUL). In this paper, we use DReST to train deep RL agents and fine-tune Qwen3-8B and Llama-3.1-8B-Instruct to be NEUTRAL and USEFUL. We find that these DReST models generalize to being NEUTRAL and USEFUL in unseen contexts at test time. Indeed, DReST RL agents achieve 11% (PPO) and 18% (A2C) higher USEFULNESS on our test set than default agents, and DReST LLMs achieve near-maximum USEFULNESS and NEUTRALITY. We also test our LLMs in an out-of-distribution setting where they can pay costs to influence when shutdown occurs. We find that DReST training roughly halves the mean probability of influencing shutdown (from 0.62 to 0.30 for Qwen and from 0.42 to 0.23 for Llama). DReST training also almost entirely eliminates the share of prompts on which influencing shutdown is the most likely option (from 0.59 to 0.01 for Qwen and from 0.53 to 0.00 for Llama). Our results thus provide some early evidence that DReST could be used to train more advanced agents to be useful and shutdownable.
title Towards Shutdownable Agents: Generalizing Stochastic Choice in RL Agents and LLMs
topic Artificial Intelligence
url https://arxiv.org/abs/2604.17502