A Toolkit to Enable the Design of Trustworthy AI

Abstract

Technological progress in artificial intelligence (AI) and machine learning (ML) has an enormous impact on our society, economy and environment. And although the urgent need for creating sustainable and ethical AI technology is admitted, there exists a lack of design tools and expertise to facilitate this advancement. This study investigates how to help designers design for the value of trust in AI systems. A literature review unveiled a myriad of ethical AI principles as well as gathered existing tools addressing the research area. Iterative reviews together with an expert on trust in technology evaluated these guidelines and a first iteration of the toolkit containing 28 design principles had been created. Through multiple participatory design workshops the next iteration of the toolkit was co-designed in collaboration with design professionals. The result is an iterated toolkit comprising 16 principles relevant in the design for trust in AI systems, and providing tool suggestions for each principle.