Designing and Evaluating Intelligent Agents’ Interaction Mechanisms for Assisting Human in High-Level Thinking Tasks

摘要

In the twenty-first century, people need to learn and apply high-level thinking skills, such as decision-making, problem-solving, and critical thinking, in their daily lives. Traditionally when in need, people can get help from human assistants to complete these thinking tasks, yet, such human experts are not always accessible. Intelligent agents, such as physical robots and virtual chatbots, are promising alternatives to the human assistants in these tasks. My doctoral research focuses on designing intelligent agents to assist human users and evaluating the impact of their interaction mechanisms (i.e., way of behaving) on user experience. Specifically, my works surround this theme by proposing: 1) an anticipation-autonomy framework that models the service robot’s proactivity in decision-making support contexts (decision-making); 2) a bot assistant that helps users to solve the problem of writing poor-quality comments in online mental health communities either by assessing the writing performance or recommending writing examples (problem-solving); and 3) a chatbot in comparison to a non-conversational tool for facilitating people to read academic papers critically (critical thinking). My dissertation contributes to the design of intelligent agents for helping people in high-level thinking tasks and insights for developing appropriate interaction mechanisms for these agents in the future.