Artists’ Dilemma

Early research on the nascent field of GenAI creativity has produced intriguing but conflicting results. Some have found it inspires artists to push at the boundaries, while others find it homogenises output. Professor Jie Gong and Professor Yulin Fang of the HKU Business School recently added more nuanced conclusions on GenAI’s effects in separate investigations that looked at the production of art and whether the use of GenAI leads to higher valuation.

Professor Gong and co-authors, Professor Jin Li of HKU, Mr Zibo Zhao and Dr Yvonne Jie Chen, focussed on art production, comparing human output with and without GenAI support to see how AI assistance affects artists’ motivation and the resulting quality.

Some 219 professional and student artists were tasked with producing two works of art inspired by text excerpts from literature in two 150-minute sessions. In the first session, they used only their preferred mediums. In the second, the artists were randomly divided: 168 could also use GenAI while the rest continued as usual as the control group. The whole process was videotaped and independent judges from academia and industry assessed the works at different stages, from initiation to completion.

The findings revealed two things. First, GenAI-assisted works quickly achieved higher quality than the unassisted works. But at the 60-minute mark it plateaued. “Going from reasonably good to excellent was harder to achieve,” Professor Gong said.

The works produced without GenAI, however, kept improving. This dovetailed with the second finding that GenAI had a negative impact on the artists’ motivation. Many of those using GenAI stopped working at the 60-minute mark, making a rational trade-off that the extra effort would offer diminishing returns. Those without GenAI worked right up to the deadline and the quality of their work increased throughout their session. (Interestingly, 20 per cent of artists given AI access chose not to use the tool.)

“Individuals using GenAI can produce good enough outcomes with remarkable speed, only to labour extensively for the additional improvement that marks true quality,” Professor Gong said. “This is the essence of the GenAI paradox: better technology, worse motivation.”

The implications go well beyond the art world. Those working in fields that require creative input, such as marketing, might find it more profitable to trade quality for quantity to meet deadlines and handle workloads. But if excellence is the goal, they may need encouragement to go the extra mile.

“The direct implication is that the pursuit of excellence will be even harder because the technology makes it so efficient to stop working earlier. GenAI is not going to help get the best output. Quite the opposite – it encourages people to settle for ‘good enough’,” she said.

Finding encouragement

Professor Fang’s research focussed on the digital art market, particularly whether GenAI makes artists more creative and whether that adds monetary value to their artwork.

He and his team, Dr Bingjie Qian and Mr Yangchen Mou of HKU’s Institute of Digital Economy and Innovation (IDEI), worked with 415,906 digital paintings by 3,355 artists on a prominent trading platform that requires artists to disclose if they use GenAI, though not how they use it. Existing technologies were used to analyse content creativity (themes and topics) and visual creativity (artist’s personal style). The innovation of the works was also assessed on how much they differed from the artist’s previous work and from other artists. Artist reputation based on how many medals they had been awarded was also factored in, as were measures to validate their truthfulness in GenAI disclosures.

The results showed GenAI could increase creativity, but with some important caveats. “The use of GenAI was more related to an increase in content creativity of those organic works, but not visual creativity,” Professor Fang said.

This had implications on their value. GenAI use did not lead to a higher premium for artworks. Instead, the most important factors were the artist’s reputation and their ‘radical’ visual creativity. (Think of Picasso and Monet, who might paint the same subjects but in very different styles.)

“It’s a paradox. GenAI increases creativity to some extent, but this kind of creativity does not affect sales. GenAI does not affect creativity at a radical level, but that is the very factor that affects sales,” he said.

Professor Fang is also Director of IDEI and he believes there are takeaways for the business world. Behavioural studies in his field do not tend to break creativity down into components. Doing so can give a more fine-grained understanding of both creativity and how GenAI can be used to enhance it.

He also thinks artists can find encouragement. “Our conclusion is that AI helps artists to be more creative, specifically in a digital platform context. So maybe artists do not have to feel overwhelmed by AI. They can use it, get inspired, and may produce more creative artwork,” he said.