Position: Stop Acting Like Language Model Agents Are Normal Agents

TL;DR

This paper argues that Language Model Agents should not be treated as normal agents due to their inherent structural problems, which undermine their reliability, trustworthiness, and perceived agency.

Contribution

It highlights the intrinsic pathologies of LMAs and advocates for measuring their ontological properties throughout deployment to mitigate issues.

Findings

LMAs are ontologically stateless and stochastic

Pathologies destabilize LMA properties like identity and consistency

Measuring ontological properties can help mitigate negative effects

Abstract

Language Model Agents (LMAs) are increasingly treated as capable of autonomously navigating interactions with humans and tools. Their design and deployment tends to presume they are normal agents capable of sustaining coherent goals, adapting across contexts and acting with a measure of intentionality. These assumptions are critical to prospective use cases in industrial, social and governmental settings. But LMAs are not normal agents. They inherit the structural problems of the large language models (LLMs) around which they are built: hallucinations, jailbreaking, misalignment and unpredictability. In this Position paper we argue LMAs should not be treated as normal agents, because doing so leads to problems that undermine their utility and trustworthiness. We enumerate pathologies of agency intrinsic to LMAs. Despite scaffolding such as external memory and tools, they remain ontologically stateless, stochastic, semantically sensitive, and linguistically intermediated. These pathologies destabilise the ontological properties of LMAs including identifiability, continuity, persistence and and consistency, problematising their claim to agency. In response, we argue LMA ontological properties should be measured before, during and after deployment so that the negative effects of pathologies can be mitigated.