AI, Take the Wheel: Using Autonomous Agents as Teammates in Sports Betting and Planning

One of the promises of artificial intelligence (AI) is making sense of large datasets and knowledge bases and communicating this to human counterparts. An increasing number of online platforms have started to offer conversational AI, such as ChatGPT, allowing individuals to talk with AI while engaging in many different activities. But will people be inclined to interact with AI agents in sports betting like other types of AI interactions? How much trust are individuals willing to instill in AI when researching sports teams and betting activities? The present research explores the factors that influence the perceptions of AI within online sports betting in comparison to other contexts. In Study 1, we investigated the extent to which participants viewed an AI agent as a human-like teammate when offered within different activities. Participants were presented with a description of the AI within one of four contexts:

sports betting, employee hiring, travel assistant, and personal assistant. Within each context, participants were randomly assigned to read a description where the AI was described as a human-like teammate (high teaming) or an instrumental tool (low training). Afterwards, the scales of the autonomous agent teammate-likeness (AAT) questionnaire were completed to determine whether perceptions of the AI varied by teaming and context condition.

For the high teaming condition, similar ratings were observed across all activities and scales. For low teaming, lower ratings were observed for sports betting compared to travel for the agency scale (perceived as having autonomy), whereas lower ratings were observed for hiring in the teammate scale (perceived as being a teammate). This suggests that, when described as a human-like teammate, AI was perceived similarly for sports betting as well as other activities. In addition, when described as being tool-like, the sports betting AI was perceived as being even less human-like than the travel assistant AI. In Study 2, we investigated how perceptions of AI in these different activities changed with experience. Participants viewed simulated text-message conversations between AI and a human user. We observed that accumulated conversations contributed to participants viewing the AI as more like a human-like teammate; this pattern was similar for sports betting, personal assistant, and job hiring activity contexts. Overall, this research suggests that sports betting app users are willing to interact with a built-in AI assistant as they make wagers.