Thanks a Bot: Leveraging Artificial Intelligence for Improved Workplace Appreciation

摘要

Generative artificial intelligence (e.g., OpenAI’s ChatGPT), is regularly tasked with navigating the complexities of human-to-human interaction, including peer-to-peer mental health support and improving interpersonal emotion regulation within customer service settings. Qualitative findings question whether artificial intelligence deserves a space in this area (e.g., how can robots be empathic?); however, many recent studies indicate that artificial intelligence may improve human-to-human interaction. In the current study, and across two samples (NGivers = 522; NReceivers = 580; NTotal = 1,102) gathered to detect small effects, we examine a specific use case of generative artificial intelligence in the workplace and whether it can be leveraged to improve prosocial human-to-human messaging above and beyond human-written prompts. In particular, we train and examine the impact of AI on participants’ perceptions of employee recognition messages in the workplace. Notably, we found that participants: a) believed the trained generative artificial intelligence was in line with Human Resources (HR) industry best practice; b) learned from the generative artificial intelligence, believing it would lead them to change their behavior; and c) would instead give and receive messages written by the generative artificial intelligence over human-written messages. Further, recognition “receiver” participants believed the generative artificial intelligence messages to be more in line with HR industry best practice than human-written messages, and these messages were longer, positively valenced; and after controlling for message length, had a more nouns, adjectives, and verbs than human-written messages. Consistencies with the current literature and the place generative artificial intelligence has in human-to-human interaction moving forward are discussed.