Assisting Learning and Insight in Design Using Embodied Conversational Agents

PI: Larry Leifer


Reflective practices have been shown to correlate with both insight in engineering design projects (Currano 2015) and learning in engineering education (Chew et al. 2016). Schön (1992) observed that designers engage in a reflective conversation with design materials as they sketch and prototype their ideas. However, students can struggle when learning domain knowledge and design practices simultaneously. We hypothesize that (1) making the conversation between student designers and their prototyping materials literal and explicit, rather than metaphorical, can assist in reflection, and (2) the effect can be enhanced by embodying a conversational agent within the prototype itself. We propose to develop an embodied conversational agent as a tool to elicit reflection, leading to greater insight and learning, during a hands-on prototyping and design activity. The agent will answer questions, offer comments, and ask deep reasoning and generative design questions, which correlate with convergent and divergent phases of the design process (Eris 2003). We will compare conditions with and without conversational agents, and agents that are embodied and not embodied, to evaluate the impact of (1) conversational agents, and (2) embodiment, on learning and insights gained in a basic mechatronics prototyping task and a paper robot redesign task.


David Sirkin, Rebecca Currano