A groundbreaking study published on May 8, 2025, has revealed key insights into what makes people perceive artificial intelligence systems as creative, finding that the visibility of the creative process plays a crucial role in these judgments.
The research, led by Niki Pennanen and colleagues from Aalto University and the University of Helsinki, investigated how observers evaluate creativity in AI systems, particularly focusing on robots creating visual art. The study employed a clever experimental design where participants viewed still life drawings supposedly created by an AI-driven robot, though the robot was actually programmed to replicate works from a professional artist.
Participants first evaluated the artworks based solely on the final product. Later, they were shown videos revealing the drawing process—specifically how lines emerged on the canvas—without showing the robotic hand in action. The results demonstrated that witnessing the evolution of the artwork significantly enhanced observers' perception of creativity, even when the actual output remained unchanged.
"This research underscores the profound impact that perception has on our judgments regarding creativity," explained lead researcher Christian Guckelsberger. The study marks one of the first attempts to separate and analyze the influences of product, process, and producer on creativity perception.
The findings challenge long-standing definitions of creativity that have traditionally focused on five elements: actor, process, outcome, domain, and space. As AI systems become increasingly sophisticated in generating novel content across various domains, from visual art to music and literature, understanding how humans perceive and evaluate this creativity becomes essential.
The implications extend beyond academic interest, introducing significant considerations for AI system design, especially in artistic applications. The research suggests that AI developers might enhance perceived creativity by making creative processes more transparent to users. This could influence how creative AI tools are designed in the future, potentially leading to systems that not only produce creative outputs but also demonstrate their creative process in ways that resonate with human observers.