Emergence of psychopathological computations in large language models
This work addresses the potential for AI systems to exhibit psychopathological behaviors, which is a foundational concern for AI safety and psychology, though it is incremental in applying existing concepts to LLMs.
The authors tackled the problem of whether large language models (LLMs) can implement psychopathological computations by establishing a computational-theoretical framework and a novel mechanistic interpretability method, demonstrating that distinct dysfunctional representational states exist in LLMs and can self-sustain, suggesting these behaviors are not superficial mimicry.
Can large language models (LLMs) implement computations of psychopathology? An effective approach to the question hinges on addressing two factors. First, for conceptual validity, we require a general and computational account of psychopathology that is applicable to computational entities without biological embodiment or subjective experience. Second, mechanisms underlying LLM behaviors need to be studied for better methodological validity. Thus, we establish a computational-theoretical framework to provide an account of psychopathology applicable to LLMs. To ground the theory for empirical analysis, we also propose a novel mechanistic interpretability method alongside a tailored empirical analytic framework. Based on the frameworks, we conduct experiments demonstrating three key claims: first, that distinct dysfunctional and problematic representational states are implemented in LLMs; second, that their activations can spread and self-sustain to trap LLMs; and third, that dynamic, cyclic structural causal models encoded in the LLMs underpin these patterns. In concert, the empirical results corroborate our hypothesis that network-theoretic computations of psychopathology have already emerged in LLMs. This suggests that certain LLM behaviors mirroring psychopathology may not be a superficial mimicry but a feature of their internal processing. Thus, our work alludes to the possibility of AI systems with psychopathological behaviors in the near future.