Human-AI teaming
Human-AI teaming and collaboration in engineering design is a multi-faceted problem that benefits from perspectives in psychology and management as well as AI and engineering. Our research areas focus on designing AI agents for collaboration and engineering purpose, structuring human-AI interaction, creating human-AI hybrid teams, and leveraging use-inspired research.
Designing AI agents for collaboration and engineering purpose
A key tenet of the initiative is that AI alone is rarely the right solution—rather, a human+AI solution tends to be better. However, much work in AI and design computation has traditionally been automation-focused. For that reason, our research often starts with advancing the state-of-the-art in AI and machine learning to create AI agents that are designed from the ground up to interact with human designers and engineers. Often, these AI agents are one of three types: tools, intended to be invoked at specific times by a user; partners, which proactively and interactively collaborate with a teammate; and managers, which make recommendations to a team or individual regarding their design process. At the same time agents need to understand and solve engineering problems, and such technology creation is a foundational focus of the initiative.
Structuring human-AI interactions
As we advance the state-of-the-art in collaborative AI agents for design, it becomes vital to assess how those agents interact with human designers, and what effect these interactions have on the designer’s mental state and patterns of work. AIs solving hard problems will not be perfect. That imperfection is reflected on the human partner, in part due to their inability to solve the problem alone, the form and outcomes of design decisions by human and AI convolve in sometimes unexpected ways. Understanding this interaction, and identifying ways to inform the human partner of such limitations (or overcome them) is an active area of research for the Initiative.
Creating human-AI hybrid teams
Design never happens in a vacuum—it almost always happens in a team! Understanding the social context of design is an inextricable piece of the puzzle. It’s important to examine how the integration of AI agents within this social context impacts the behavior and performance of the team. We conduct both theoretical and empirical research on hybrid teams to build a more detailed understanding of the challenges and opportunities posed. In some cases, these teams perform exceedingly well, but in other cases teams perform worse after the introduction of AI. Hybrid teaming is far from being a one-size-fits-all solution.
Leveraging use-inspired research
As we discover new knowledge in AI, design, and collaboration, we also seek to apply these insights across several domains of interest. By engaging in these application areas, we also gain a better understanding of the opportunities and challenges of human+AI approaches in practice. The following application areas are samples of the operational context of our work:
- Design for Additive Manufacturing
- Generative Manufacturing
- Co-Design of UAVs and Operational Logistics
- Design of Civil Infrastructure
- Proactive Operation and Maintenance of Civil Infrastructure
- Digital Transformation of Healthcare
- Search and Rescue Operations