Is AI actually inventive? Examine exhibits how visibility of course of shapes notion

April 7, 2025

The GIST Editors' notes

This text has been reviewed in response to Science X's editorial course of and insurance policies. Editors have highlighted the next attributes whereas guaranteeing the content material's credibility:

fact-checked

trusted supply

proofread

Is AI actually inventive? Examine exhibits how visibility of course of shapes notion

What makes folks assume an AI system is inventive? New analysis exhibits that it relies on how a lot they see of the inventive act. The findings have implications for a way we analysis and design inventive AI techniques, they usually additionally elevate basic questions on how we understand creativity in different folks.

The work is printed within the journal ACM Transactions on Human-Robotic Interplay.

"AI is taking part in an more and more giant function in inventive follow. Whether or not which means we should always name it inventive or not is a distinct query," says Niki Pennanen, the examine's lead writer. Pennanen is researching AI techniques at Aalto College and has a background in psychology. Along with different researchers at Aalto and the College of Helsinki, he did experiments to seek out out whether or not folks assume a robotic is extra inventive in the event that they see extra of the inventive act.

Within the examine, contributors had been initially requested to guage the creativity of robots based mostly solely on nonetheless life drawings they’d made. They had been informed the robots had been pushed by AI, however in actual fact it had been programmed to breed drawings that the researchers had commissioned from an artist. This deception made it potential to measure folks's notion of creativity with out requiring the robotic to be inventive, which might have launched an excessive amount of variability between the drawings.

Subsequent, the examine contributors evaluated how inventive the drawings had been after they noticed not solely the ultimate product but in addition a video of the drawing course of––the traces showing on the web page, however not the robotic creating them. Within the ultimate stage, contributors scored the drawings after they may see all three parts: the ultimate product, the method, and the robotic making the drawing.

The findings confirmed that the drawings had been seen as extra inventive as extra parts of the inventive act had been revealed.

"The extra folks noticed, the extra inventive they judged it to be," says Christian Guckelsberger, assistant professor of inventive applied sciences at Aalto and the examine's senior writer. "So far as I'm conscious, we're the primary to check the results of perceiving product, course of and producer in a separate and managed method, not solely within the context of AI but in addition extra usually."

The ability of notion

Understanding how folks assess the creativity of robots or different synthetic techniques is essential in interested by find out how to design them––but it surely's not completely clear what the suitable design decisions can be.

"The examine means that revealing extra concerning the course of and producer may be conducive to folks's notion of the techniques' creativity," says Guckelsberger. "But when we added parts to make AI techniques appear extra inventive although the system is in actual fact performing the identical means, we may query whether or not that's really an excellent factor."

In some instances, that might be useful––for instance, it may be a means to assist folks keep engaged with a co-creative system. However in different contexts, it may give folks a misleading impression of how inventive a man-made system actually is.

"Our findings assist deal with this battle by giving us a greater concept of our personal human biases. This analysis makes them a bit extra clear, which can also be essential from the person's perspective, for us to know how a system's design impacts our notion of it," says Guckelsberger.

Along with these social and design implications, the findings even have significance for analysis on inventive AI techniques. If our judgment of creativity relies on how a system is offered, then future research ought to management for that issue. Likewise, current analysis must be reevaluated in mild of those findings––evaluating the creativity of various techniques with out accounting for variations of their presentation may have led to false conclusions.

One other intriguing query posed by this analysis is what it tells us about ourselves.

"Now that we've discovered this about folks's notion of AI creativity, does it additionally apply to folks's notion of different folks?" asks Guckelsberger.

Does form matter?

The researchers additionally carried out the experiments with two totally different robotic designs. Their objective was to check whether or not folks scored the creativity otherwise relying on the robotic's form, as a result of earlier work had advised a hyperlink between form and perceived creativity.

The group examined whether or not folks noticed totally different ranges of creativity when a nonetheless life was drawn by a smooth arm-like robotic or a extra mechanistic plotter robotic. Retaining the drawings constant between the robots and from one participant to a different was fairly difficult.

"I believe our greatest issue was the bodily robots themselves. We did lots of work with the robots and the drawing course of to attempt to hold all the pieces equivalent so we may do a scientifically rigorous comparability," says Pennanen.

The researchers had been shocked to seek out no vital distinction in how folks scored the 2 robots. They're planning future work to look additional into this counterintuitive end result, in addition to what different parts affect our notion of creativity.

"We're serious about doing extra analysis about what sorts of biases have an effect on our analysis of inventive and embodied AI techniques and the way these results occur," says Pennanen.

The findings must also be confirmed for various creative genres, in addition to different types of artwork and inventive expression. To make it simpler for others to copy their work and construct on it, the researchers adopted strict open science practices. As synthetic techniques turn into commonplace, understanding the components shaping our notion of their creativity is significant for efficient design––and it might additionally shed some mild on how we acknowledge creativity in people.

Extra info: Niki Pennanen et al, From Product to Producer: The Influence of Perceptual Proof and Robotic Embodiment on the Human Evaluation of AI Creativity, ACM Transactions on Human-Robotic Interplay (2025). DOI: 10.1145/3711939

Supplied by Aalto College Quotation: Is AI actually inventive? Examine exhibits how visibility of course of shapes notion (2025, April 7) retrieved 7 April 2025 from https://techxplore.com/information/2025-04-ai-creative-visibility-perception.html This doc is topic to copyright. Aside from any truthful dealing for the aim of personal examine or analysis, no half could also be reproduced with out the written permission. The content material is offered for info functions solely.

Discover additional

Examine: Wanting contained in the black field of gender variations in creativity 0 shares