Anthropomorphization of AI: Opportunities and Risks

Abstract

Anthropomorphization, which is the tendency to attribute human-like traits to non-human entities, is prevalent in many social contextschildren anthropomorphize toys and adults do so with brands. It is also a versatile tool in science, with behavioral psychology and evolutionary biology meticulously documenting its consequences. With widespread adoption of AI systems, and the push to make it human-like through alignment techniques, human voice, and avatars, the tendency for users to anthropomorphize it increases significantly. We take a dyadic approach to understanding this phenomenon with large language models (LLMs) by studying (1) the objective legal implications, as analyzed through the lens of the recent blueprint of AI bill of rights 1 and the (2) subtle psychological aspects of customization and anthropomorphization. We find that anthropomorphized LLMs customized for different user bases violate multiple provisions in the legislative blueprint and raise corporate personhood confusions. In addition, we point out that anthropomorphization of LLMs affects the influence they can have on their users, thus establishing potential for manipulation and negative influence. With LLMs being hyper-personalized for vulnerable groups like children and patients among others, we propose a conservative strategy for the cautious use of anthropomorphization to improve trustworthiness of AI systems.