Doctoral Dissertation Research: The Effects of Anthropomorphic Linguistic Framing on the Online Sentence Processing of Texts about AI

As artificial intelligence (AI) large language models (LLMs) gain public attention, it has become clear that people are likely to anthropomorphize them – that is, to think of them as humanlike agents, not just machines. The anthropomorphism of AI raises unique ethical considerations because it affects people’s perception of responsibility. For example, people are less outraged when gender discrimination is caused by AI than when it is caused by a human and are less likely to consider the company who developed the AI responsible for the discrimination. Importantly, the way people talk about AI changes how people think about it. Research has shown that describing non-humans anthropomorphically can dramatically change how people assign agency and responsibility to them. Very likely, anthropomorphic discourse is shaping how these powerful, new pieces of technology will fit into the world. However, there has been no research investigating how anthropomorphic language affects the ability to reason about AIs specifically. This doctoral dissertation project investigates how anthropomorphic linguistic framing affects people's conscious and unconscious reasoning about AIs. This project also benefits society by providing educational opportunities to students and by communicating the project results to lay audiences. This doctoral dissertation project uses eye-tracking technology to study how people understand anthropomorphic texts about AI. Specifically, by observing participants’ eye movements, this project investigates how surprised people are when a text changes from anthropomorphic (“The AI tried to help patients”) to non-anthropomorphic ("The AI was used to help patients”) or vice versa. By comparing these results to participants’ conscious opinions about AI, this study supports a better understanding of how reading anthropomorphic language affects people’s conscious and unconscious reasoning about AI. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.