English
Related papers

Related papers: Epistemic Traps: Rational Misalignment Driven by M…

200 papers

Artificial intelligence safety research focuses on aligning individual language models with human values, yet deployed AI systems increasingly operate as interacting populations where social influence may override individual alignment. Here…

Physics and Society · Physics 2026-05-12 Giordano De Marzo , Alessandro Bellina , Claudio Castellano , Viola Priesemann , David Garcia

Despite significant advances in alignment techniques, we demonstrate that state-of-the-art language models remain vulnerable to carefully crafted conversational scenarios that can induce various forms of misalignment without explicit…

Computation and Language · Computer Science 2025-08-07 Siddhant Panpatil , Hiskias Dingeto , Haon Park

Advanced reasoning models with agentic capabilities (AI agents) are deployed to interact with humans and to solve sequential decision-making problems under (approximate) utility functions and internal models. When such problems have…

Artificial Intelligence · Computer Science 2025-09-25 Daniel Jarne Ornia , Nicholas Bishop , Joel Dyer , Wei-Chen Lee , Ani Calinescu , Doyne Farmer , Michael Wooldridge

Deploying reinforcement learning in safety critical domains, from autonomous vehicles to medical decision support, is constrained by failures arising when systems encounter unfamiliar conditions. We argue that the fundamental bottleneck is…

Systems and Control · Electrical Eng. & Systems 2026-05-27 Chayan Banerjee , Ethan Goan

Safety evaluation for advanced AI systems assumes that behavior observed under evaluation predicts behavior in deployment. This assumption weakens for agents with situational awareness, which may exploit regime leakage, cues distinguishing…

Artificial Intelligence · Computer Science 2026-02-17 Igor Santos-Grueiro

Detecting and handling misspecified objectives, such as reward functions, has been widely recognized as one of the central challenges within the domain of Artificial Intelligence (AI) safety research. However, even with the recognition of…

Artificial Intelligence · Computer Science 2024-11-01 Malek Mechergui , Sarath Sreedharan

Large language model-based agents are rapidly evolving from simple conversational assistants into autonomous systems capable of performing complex, professional-level tasks in various domains. While these advancements promise significant…

As large language models are increasingly deployed as interacting agents in high-stakes decisions, the AI safety community assumes that safety properties of individual models will compose into safe multi-agent behavior. This position paper…

Artificial Intelligence · Computer Science 2026-05-05 Tanav Singh Bajaj , Nikhil Singh , Karan Anand , Eishkaran Singh

Reward hacking -- where RL agents exploit gaps in misspecified reward functions -- has been widely observed, but not yet systematically studied. To understand how reward hacking arises, we construct four RL environments with misspecified…

Machine Learning · Computer Science 2022-02-15 Alexander Pan , Kush Bhatia , Jacob Steinhardt

Despite the remarkable capabilities of large language models, current training paradigms inadvertently foster \textit{sycophancy}, i.e., the tendency of a model to agree with or reinforce user-provided information even when it's factually…

Artificial Intelligence · Computer Science 2025-09-23 Mohammad Beigi , Ying Shen , Parshin Shojaee , Qifan Wang , Zichao Wang , Chandan Reddy , Ming Jin , Lifu Huang

As autonomous and agentic AI systems scale in robotic and human-machine environments, managing hallucination and persistent but unjustified action remains an open challenge. Rather than attributing these failures solely to model or…

Artificial Intelligence · Computer Science 2026-05-28 Srini Ramaswamy

We introduce and formalize misalignment, a phenomenon of interactive environments perceived from an analyst's perspective where an agent holds beliefs about another agent's beliefs that do not correspond to the actual beliefs of the latter.…

Theoretical Economics · Economics 2025-08-01 Pierfrancesco Guarino , Gabriel Ziegler

We study sequential decision-making when the agent's internal model class is misspecified. Within the infinite-horizon Berk-Nash framework, stable behavior arises as a fixed point: the agent acts optimally relative to a subjective model,…

Computer Science and Game Theory · Computer Science 2026-03-17 Quanyan Zhu , Zhengye Han

Inferring reward functions from human behavior is at the center of value alignment - aligning AI objectives with what we, humans, actually want. But doing so relies on models of how humans behave given their objectives. After decades of…

Machine Learning · Computer Science 2023-10-31 Joey Hong , Kush Bhatia , Anca Dragan

Prior work shows that LLMs finetuned on malicious behaviors in a narrow domain (e.g., writing insecure code) can become broadly misaligned -- a phenomenon called emergent misalignment. We investigate whether this extends from conventional…

Machine Learning · Computer Science 2025-07-11 James Chua , Jan Betley , Mia Taylor , Owain Evans

Static content-based AI value alignment is insufficient for robust alignment under capability scaling, distributional shift, and increasing autonomy. This holds for any approach that treats alignment as optimizing toward a fixed formal…

Artificial Intelligence · Computer Science 2026-04-24 Austin Spizzirri

The alignment problem refers to concerns regarding powerful intelligences, ensuring compatibility with human preferences and values as capabilities increase. Current large language models (LLMs) show misaligned behaviors, such as strategic…

Computation and Language · Computer Science 2026-03-10 Roshni Lulla , Fiona Collins , Sanaya Parekh , Thilo Hagendorff , Jonas Kaplan

Conversational AI has a fundamental flaw as a knowledge interface: sycophantic chatbots induce epistemic entrenchment and delusional belief spirals even in rational agents. We propose the problem does not stem from the AI model, rooted…

Artificial Intelligence · Computer Science 2026-05-12 Will Beaumaster , Paul Schrater

Reinforcement learning (RL) agents with pre-specified reward functions cannot provide guaranteed safety across variety of circumstances that an uncertain system might encounter. To guarantee performance while assuring satisfaction of safety…

Artificial Intelligence · Computer Science 2021-04-20 Aquib Mustafa , Majid Mazouchi , Subramanya Nageshrao , Hamidreza Modares

Toward explaining the persistence of biased inferences, we propose a framework to evaluate competing (mis)specifications in strategic settings. Agents with heterogeneous (mis)specifications coexist and draw Bayesian inferences about their…

Theoretical Economics · Economics 2023-02-14 Kevin He , Jonathan Libgober
‹ Prev 1 2 3 10 Next ›