Beyond Computational Equivalence: The Behavioral Inference Principle for Machine Consciousness

The authors argue that the current consensus for attributing consciousness to large language models (LLMs) is flawed. This consensus, based on computational functionalism, suggests consciousness should be ascribed when a system's computations are equivalent to those of a conscious being. The authors propose an alternative "behavioral inference principle": consciousness should be attributed to an artificial entity only when doing so helps explain and predict its observable behaviors. They contend this principle offers a more epistemologically sound and operationalizable criterion for assessing machine consciousness, moving beyond abstract computational equivalence to a testable, behavior-based approach.

Study at a glance

Abstract

Large Language Models (LLMs) have rapidly become a central topic in AI and cognitive science, due to their unprecedented performance in a vast array of tasks. Indeed, some even see 'sparks of artificial general intelligence' in their apparently boundless faculty for conversation and reasoning, Their sophisticated emergent faculties, which were not initially anticipated by their designers, has ignited an urgent debate about whether and under which circumstances we should attribute consciousness to artificial entities in general and LLMs in particular. The current consensus, rooted in computational functionalism, proposes that consciousness should be ascribed based on a principle of computational equivalence. The objective of this opinion piece is to criticize this current approach and argue in favor of an alternative “behavioral inference principle”, whereby consciousness is attributed if it is useful to explain (and predict) a given set of behavioral observations. We believe that a behavioral inference principle will provide an epistemologically valid and operationalizable criteria to assess machine consciousness.