Comparing human against computer generated designs: New possibilities for design activity within agile projects

Abstract

Our research explores the possibility of using intelligent agents to automate some of the design activity and evaluate if this kind of tool can perform as good as a human designer would. Thus, we present an app called Design Thinking Canvas Autonomus that supports and automates the creation of design concepts. The justification to do so comes from the context of design activity being executed within agile projects. From the most prominent themes of this conjunction, Little Design Up-Front stands outs as one of the most important. It corresponds to agile projects’ need of defining a clear objective, while avoiding unnecessary waste in the long run. In order to test if Autonomus could help addressing this need, we tested its performance against a human designer. The results showed that there was not a statistically significant difference between Design Thinking Canvas Autonomus and its human counterpart in all considered criteria. Actually, for suitability factor, Autonomus had a statistically proven advantage compared to human designs (p-value of 0.037), according to specialist designers’ evaluation. It establishes a new frame of reference for agile projects, as the cost and effort to execute Autonomus is significantly lower than a human process.