Developing an AI Concept Envisioning Toolkit to Support Reflective Juxtaposition of Values and Harms

Early-stage AI concept envisioning often treats values and potential harms as abstract concepts addressed too late to meaningfully shape design decisions. To solve this, researchers Pitch Sinlapanuntakul, Soyun Moon, Yuri Kawada, Yeha Chung, and Mark Zachry developed the AI Concept Envisioning Toolkit through a Research-through-Design (RtD) approach. The toolkit comprises three components: an AI Capability Library that maps technical possibilities, 24 Value-Harm Cards that force designers to juxtapose positive values with potential negative consequences, and a Value-Tension Map to visualize conflicts. These tools introduce 'productive friction' into workflows, making ethical reasoning a concrete, iterative part of the design process rather than a checkbox exercise.

The team validated the toolkit through a survey with 30 designers and in-depth interviews with 12 participants. Results showed the toolkit is clear and perceived as valuable: it encourages deep value reflection, helps anticipate harms early, and makes ethical trade-offs transparent. Designers reported that the physical cards and mapping tools turned abstract ethical concepts into actionable discussion points. The study demonstrates that structured, tangible tools can embed ethics into AI workflows from the very first concept sketch, rather than leaving it as a late-stage audit. This work offers a practical template for any team building AI products to systematically consider impacts before coding begins.