Can a robot be creative?

Patrick Tresset builds robots, each named Paul, that draw a sitter from observation using camera eyes and a pen held in a physical arm. Asked directly about their creativity, Tresset told Right Click Save: "I don't see these systems as artists, but as machines that are perceived as artists. They are simulacra, but when I work with them, I forget that." He describes their apparent intent as an effect of design, saying his algorithms create the impression of intention by using different drawing methods in sequence, and notes that the arms' physical characteristics influence the work's style.

The cognitive scientist Margaret Boden, in The Creative Mind: Myths and Mechanisms, distinguishes combinational creativity, exploratory creativity and transformational creativity. On combination she writes that combining ideas creatively is not like shaking marbles in a bag, and that the ideas have to come together because of some intelligible though previously unnoticed link. Exploratory creativity works within a conceptual space; the deepest cases, she writes, involve someone thinking something which, with respect to the conceptual spaces in their minds, they could not have thought before. She has also argued that computers lack the embodied needs that drive human creative work.

Stephen Thaler sought to register a copyright in a work he attributed entirely to his AI system. In Thaler v. Perlmutter, No. 23-5233, argued 19 September 2024 and decided 18 March 2025, the US Court of Appeals for the District of Columbia Circuit affirmed against him, reported at 130 F.4th 1039. The ruling is summarised by Baker Donelson as holding that authors must be human beings and that human authorship is a bedrock requirement of copyright, reasoning from provisions of the Copyright Act whose terms only a human can satisfy. The Supreme Court denied Thaler's petition for certiorari on 2 March 2026. The case concerned a disembodied AI system rather than a robot.

The US Copyright Office published Copyright and Artificial Intelligence, Part 2: Copyrightability on 29 January 2025. Its conclusions state that the use of AI tools to assist rather than stand in for human creativity does not affect the availability of copyright protection for the output, that whether human contributions to AI-generated outputs are sufficient to constitute authorship must be analyzed on a case-by-case basis, and that based on the functioning of current generally available technology, prompts do not alone provide sufficient control. Surveying other jurisdictions, the report states that those that have addressed this issue so far have agreed that copyright requires human authorship. The report concerns generative AI software, not robots.

Sotheby's catalogued "A.I. God. Portrait of Alan Turing", mixed media on canvas, executed in 2024 and signed "A" by Ai-Da Robot, with an estimate of $120,000 to $180,000 in its Digital Art Day Auction on 7 November 2024. It sold for $1.08 million including buyer's premium, the first work by a humanoid robot sold at auction. Ai-Da paints and draws using cameras and algorithms that move her arm to the canvas. The attribution on the auction record is worth reading closely: the lot is credited to "Ai-Da Robot (Aidan Meller)", naming the human alongside the machine.

Ahmed Elgammal, Bingchen Liu, Mohamed Elhoseiny and Marian Mazzone proposed Creative Adversarial Networks. Their abstract states that the system generates art by looking at art and learning about style, and becomes creative by increasing the arousal potential of the generated art by deviating from the learned styles. It reports that human subjects could not distinguish art generated by the proposed system from art generated by contemporary artists and shown in top art fairs, and that human subjects even rated the generated images higher on various scales. This is a software system with no physical body, included here as contrast to the robots above.

Harold Cohen spent nearly fifty years on a single program, AARON, and gave it a body. His early system put drawings on paper using a drawing turtle, a small robot equipped with a marker, which the Computer History Museum records creating images on the museum floor at the San Francisco Museum of Modern Art in 1979. A 1995 version was almost eight feet long, used a vacuum table to hold the paper while an arm drew and coloured the forms, and mixed paints on a palette to create custom colours which it applied with a variety of brushes. It was shown at Boston's Computer Museum as The Robotic Artist: AARON In Living Color.