Recent developments in artificial intelligence (AI) have made the long-considered idea of automated healthcare more feasible. Some suggest AI can deliver more empathic responses than humans and have even integrated it into various mental health protocols. However, given AI’s lack of inner emotional life, others have claimed it cannot genuinely deliver certain aspects of human empathy. In line with this claim, my previous research indicates that perceived AI involvement reduces perceived empathy, and that emotional sharing and care have unique added value when perceived as human. My next line of research aims to explore the mechanism behind this phenomenon, using a novel interdisciplinary approach. I will use eye-tracking to examine the focus of reading patterns for empathic responses presented as AI- generated or human-authored. This will indicate whether participants approach and process empathic responses differently given their perceived source, actual source, and the interaction of the two. I hypothesize that both the entire response, and emotional words specifically, will garner more focus when presented as human-authored. Following this line, I plan to investigate people’s empathic needs in a variety of situations. I will then use experience sampling methods to test the differences in perceived empathy between human-authored and AI-generated responses, congruent or incongruent with these needs, over recurring emotional interactions. These are crucial steps, before my final project – integrating AI into the therapeutic process, where appropriate. Given empathy’s crucial role in emotional relationships, the expected results will highlight key considerations for AI’s proper use, especially in mental health.

Supervisor: Anat Perry