Anthropomorphism is the attribution of human characteristics, intentions, or mental states to nonhuman entities or systems. With artificial intelligence, it often appears when fluent language or simulated emotion is mistaken for evidence of thought, feeling, or consciousness. section 11.0 section 11.2
The debate over Google’s Language Model for Dialogue Applications (LaMDA) illustrates the problem. Engineer Blake Lemoine described the system as sentient, while other experts characterized it as a program that statistically generates plausible text outputs. Convincing conversation can therefore encourage human interpretations without establishing inner experience. section 11.0 section 11.2
Anthropomorphism does not settle whether artificial consciousness is possible. Researchers continue to investigate that possibility because increasingly prevalent systems could raise moral concerns if they became conscious or sentient. The responsible distinction is between a system’s persuasive behavior and evidence about its experiences. section 11.2
LaMDA and ELIZA
LaMDA’s apparent emotional and conversational abilities revived a pattern associated with ELIZA: people may treat language-generating systems as more human-like than the available evidence warrants. Users’ interpretations can also shape how they describe the system’s behavior. section 11.2
