Designing Explainable and Controllable Artificial Intelligence Systems Together: Inclusive Participation Formats for Software-Based Working Routines in Industry

Abstract

“This is what it would look like if your AI (Artificial Intelligence) system was explainable to and controllable by you.” With this title, a co-creation workshop was hosted by the Berlin-based Institute for Innovation and Technology (iit) on December 1, 2021. The workshop addressed the question of how artificially intelligent systems can be designed to be explainable and controllable in collaboration with different user groups. The workshop tested sociotechnical design methods for the participatory inclusion of potential users in the development process. An interactive matrix helped to collect ideas from participants and differentiated sociotechnical aspects on a technical, organizational, and human level.

As a second method, drafters (a.k.a. graphic recorders) accompanied the workshop and visualized the computer system´s interface according to the discussion. The following article introduces these two workshop methods, presents the results and gives recommendations based on the experiences for future workshops. The workshop format also addresses the challenge of operationalizing ethical principles in the design of AI.