English
Related papers

Related papers: Misalignment from Treating Means as Ends

200 papers

Reinforcement Learning from Human Feedback (RLHF) aligns Large Language Models (LLMs) with human preferences, yet the underlying reward signals they internalize remain hidden, posing a critical challenge for interpretability and safety.…

Machine Learning · Computer Science 2026-01-21 Nyal Patel , Matthieu Bou , Arjun Jagota , Satyapriya Krishna , Sonali Parbhoo

A misspecified reward can degrade sample efficiency and induce undesired behaviors in reinforcement learning (RL) problems. We propose symbolic reward machines for incorporating high-level task knowledge when specifying the reward signals.…

Artificial Intelligence · Computer Science 2022-04-22 Weichao Zhou , Wenchao Li

Learning rewards from human behaviour or feedback is a promising approach to aligning AI systems with human values but fails to consistently extract correct reward functions. Interpretability tools could enable users to understand and…

Artificial Intelligence · Computer Science 2024-10-16 Jan Wehner , Frans Oliehoek , Luciano Cavalcante Siebert

Designing rewards for Reinforcement Learning (RL) is challenging because it needs to convey the desired task, be efficient to optimize, and be easy to compute. The latter is particularly problematic when applying RL to robotics, where…

Machine Learning · Computer Science 2020-05-28 Yiming Ding , Carlos Florensa , Mariano Phielipp , Pieter Abbeel

Recent advances in deep learning have brought attention to the possibility of creating advanced, general AI systems that outperform humans across many tasks. However, if these systems pursue unintended goals, there could be catastrophic…

Machine Learning · Computer Science 2024-11-25 Dylan Xu , Juan-Pablo Rivera

Behavioral skills or policies for autonomous agents are conventionally learned from reward functions, via reinforcement learning, or from demonstrations, via imitation learning. However, both modes of task specification have their…

This research explores how human-defined goals influence the behavior of Large Language Models (LLMs) through purpose-conditioned cognition. Using financial prediction tasks, we show that revealing the downstream use (e.g., predicting stock…

General Finance · Quantitative Finance 2026-05-07 Sean Cao , Wei Jiang , Hui Xu

Humans do not just find mistakes after the fact -- we often catch them mid-stream because 'reflection' is tied to the goal and its constraints. Today's large language models produce reasoning tokens and 'reflective' text, but is it…

Artificial Intelligence · Computer Science 2025-10-24 Sion Weatherhead , Flora Salim , Aaron Belbasis

People are remarkably capable of generating their own goals, beginning with child's play and continuing into adulthood. Despite considerable empirical and computational work on goals and goal-oriented behavior, models are still far from…

Artificial Intelligence · Computer Science 2025-05-20 Guy Davidson , Graham Todd , Julian Togelius , Todd M. Gureckis , Brenden M. Lake

The aim of Reinforcement Learning (RL) in real-world applications is to create systems capable of making autonomous decisions by learning from their environment through trial and error. This paper emphasizes the importance of reward…

Machine Learning · Computer Science 2024-12-31 Sinan Ibrahim , Mostafa Mostafa , Ali Jnadi , Hadi Salloum , Pavel Osinenko

Reinforcement Learning (RL) agents require the specification of a reward signal for learning behaviours. However, introduction of corrupt or stochastic rewards can yield high variance in learning. Such corruption may be a direct result of…

Machine Learning · Computer Science 2018-11-09 Joshua Romoff , Peter Henderson , Alexandre Piché , Vincent Francois-Lavet , Joelle Pineau

Optimization in machine learning typically deals with the minimization of empirical objectives defined by training data. However, the ultimate goal of learning is to minimize the error on future data (test error), for which the training…

Machine Learning · Statistics 2021-11-08 Bernhard Stankewitz , Nicole Mücke , Lorenzo Rosasco

Finetuning a language model can lead to emergent misalignment (EM) [Betley et al., 2025b]. Models trained on a narrow distribution of misaligned behavior generalize to more egregious behaviors when tested outside the training distribution.…

Machine Learning · Computer Science 2026-04-29 Jan Dubiński , Jan Betley , Anna Sztyber-Betley , Daniel Tan , Owain Evans

This work handles the inverse reinforcement learning (IRL) problem where only a small number of demonstrations are available from a demonstrator for each high-dimensional task, insufficient to estimate an accurate reward function. Observing…

Artificial Intelligence · Computer Science 2017-10-16 Kun Li , Joel W. Burdick

Because human preferences are too complex to codify, AIs operate with misspecified objectives. Optimizing such objectives often produces undesirable outcomes; this phenomenon is known as reward hacking. Such outcomes are not necessarily…

Artificial Intelligence · Computer Science 2026-04-27 Henrik Marklund , Alex Infanger , Benjamin Van Roy

Ensuring long-term fairness is crucial when developing automated decision making systems, specifically in dynamic and sequential environments. By maximizing their reward without consideration of fairness, AI agents can introduce disparities…

Machine Learning · Computer Science 2025-01-03 Sahand Rezaei-Shoshtari , Hanna Yurchyk , Scott Fujimoto , Doina Precup , David Meger

Frontier models have demonstrated exceptional capabilities following the integration of task-reward-based reinforcement learning (RL) into their training pipelines, enabling systems to evolve from pure reasoning models into sophisticated…

Machine Learning · Computer Science 2026-04-20 Sarthak Mittal , Leo Gagnon , Guillaume Lajoie

Multimodal learning often relies on aligning representations across modalities to enable effective information integration, an approach traditionally assumed to be universally beneficial. However, prior research has primarily taken an…

Machine Learning · Computer Science 2025-11-26 Wanlong Fang , Tianle Zhang , Alvin Chan

Multi-task reinforcement learning (RL) aims to simultaneously learn policies for solving many tasks. Several prior works have found that relabeling past experience with different reward functions can improve sample efficiency. Relabeling…

Machine Learning · Computer Science 2020-02-26 Benjamin Eysenbach , Xinyang Geng , Sergey Levine , Ruslan Salakhutdinov

This paper focuses on robotic reinforcement learning with sparse rewards for natural language goal representations. An open problem is the sample-inefficiency that stems from the compositionality of natural language, and from the grounding…

Machine Learning · Computer Science 2022-09-12 Frank Röder , Manfred Eppe , Stefan Wermter