Probing the Lack of Stable Internal Beliefs in LLMs
This work addresses a critical problem for developers of interactive dialogue systems by revealing incremental limitations in LLMs' ability to simulate human-like personality traits.
The study investigated whether large language models (LLMs) can maintain stable internal beliefs, specifically 'implicit consistency' in adhering to unstated goals across multi-turn interactions, and found that LLMs struggle to preserve latent consistency without explicit context, highlighting limitations in persona-driven modeling.
Persona-driven large language models (LLMs) require consistent behavioral tendencies across interactions to simulate human-like personality traits, such as persistence or reliability. However, current LLMs often lack stable internal representations that anchor their responses over extended dialogues. This work explores whether LLMs can maintain "implicit consistency", defined as persistent adherence to an unstated goal in multi-turn interactions. We designed a 20-question-style riddle game paradigm where an LLM is tasked with secretly selecting a target and responding to users' guesses with "yes/no" answers. Through evaluations, we find that LLMs struggle to preserve latent consistency: their implicit "goals" shift across turns unless explicitly provided their selected target in context. These findings highlight critical limitations in the building of persona-driven LLMs and underscore the need for mechanisms that anchor implicit goals over time, which is a key to realistic personality modeling in interactive applications such as dialogue systems.