Um agente desenvolvimental: feedback por um agente de inteligência artificial para melhorar os processos interpessoais das equipas

The heart of the present project lies in merging the artificial intelligence’s capabilities in pattern detection and forecasting with the relevance of feeding back to teams’ information on their interpersonal interaction patterns, allowing them to adjust their behavior accordingly. Simultaneously, it allows us to explore how taking on the role of providing feedback impacts the perceptions of an AI agent by team members. The main goal of this project is to capitalize the scientific knowledge on team interaction and interpersonal processes, applying it to the development of an artificial intelligent agent able to provide real-time feedback to teams on interpersonal interaction. At the end of the project, we will be able to explore the impact of agent-led feedback on teams’ interpersonal processes and overall effectiveness, and to understand how differences in the type and form of both the agent and the feedback provided impact perceptions of agent team membership. More specifically, our objectives are: • First, we will develop and train an AI agent for the detection of interaction affective and conflict patterns from multiple-source information types (i.e., verbal language, gestures, tone of voice, etc) in work teams. The project will focus specifically on conflict and affect indicators, aligned with sentiment analysis (including opinion lexicons) and emotion recognition (Liu, 2020). Previous work on sentiment analysis has been mostly done with text (e.g., Volkova et al., 2013, Hamilton et al., 2016) or image (Poria et al., 2017) alone; combining multiple sources of information that feed into the learning of the agent in identifying the affective nature of the team’s interaction is a scientific step forward, aligned with the increasingly growing use of collaborative software. • Second, and narrowing down the focus of the project to the organizational behavior field, we will explore the developmental role of the agent, by defining feedback messages tailored to the interpersonal needs of the team identified previously, and explore their impact of that feedback on the teams’ performance, viability, and interpersonal processes over time. • Third, we will analyze the impact of different types of agent-led feedback, in terms of content and form, to the perception of YODA as a team member by the remaining members of the team, and its relationship to team effectiveness.