The current consensus for attributing consciousness to artificial entities like Large Language Models relies on computational functionalism, which proposes that consciousness should be ascribed based on computational equivalence. This opinion piece criticizes that approach and argues for an alternative 'behavioral inference principle', whereby consciousness is attributed only when doing so helps explain and predict observed behaviors. The authors believe this principle offers an epistemologically valid and operationalizable criterion for assessing 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.
The current consensus on attributing consciousness to large language models (LLMs) relies on computational functionalism, which ascribes consciousness based on computational equivalence. This opinion piece argues against that approach and proposes a behavioral inference principle instead: consciousness should be attributed only when it is useful for explaining and predicting observed behaviors. The authors contend that this principle offers an epistemologically unbiased and operationalizable standard for assessing machine consciousness, avoiding the speculative nature of the computational equivalence view.