Thursday, February 27, 2025

Professional artists viewed as more creative than AI programs

People prefer artwork generated from prompts by experienced artists, study finds

American Psychological Association

In the rapidly developing contest between human creativity and artificial intelligence algorithms, professional artists still have an edge in producing more creative AI-assisted artwork than the AI programs themselves or novice artists, according to research published by the American Psychological Association.

The rapid advancement of AI raises some existential questions about the nature of creativity, said lead researcher Paul Seli, PhD, an assistant professor of psychology and neuroscience at Duke University.

“Creativity has long been considered a uniquely human ability that is tied to personal experiences, emotions and the drive to communicate meaning,” he said. “But now that AI can generate complex, aesthetically compelling artwork, we have to ask if human creativity still holds a distinct edge or is AI fundamentally changing the way we think about creative expression?”

At least for now, professional artists still have the upper hand, according to the study findings, which were published online in Psychology of Aesthetics, Creativity, and the Arts.

The researchers selected 15 professional artists with more than five years of experience and 15 people with little or no artistic training. All of the participants wrote prompts of up to 15 words each that were entered in the DALL-E 3 AI program to generate artwork images.

The same instructions for creating a prompt also were entered 15 times in the ChatGPT AI program to create prompts that also were used to generate artwork images in DALL-E 3.

Another group of 299 online participants viewed all 45 of the generated artwork images and rated them on creativity. Artwork created with prompts by the professional artists was rated as significantly more creative than the AI-generated artwork, followed by the novice artists' work in last place.

Additional analysis found that the professional artists and the ChatGPT program used more words in their prompts, which could help generate more creative artwork.

The professional artists also used words in their prompts that were significantly higher in semantic distance than ChatGPT, followed by the novice artists. Semantic distance in this study meant the words used in a prompt were not commonly seen as related to each other, and semantic distance has been linked to creativity in previous studies. For example, one prompt from a professional artist stated: “A madman trapped in a straitjacket made of toilet paper.” By contrast, a prompt from a novice artist stated: “A frog using a leaf as an umbrella.”

“Whether AI is truly ‘creative’ depends on how we define creativity,” Seli said. “AI undeniably generates images and other outputs that people perceive as creative. But if creativity is tied to human experience, emotions and intentionality, then AI appears to fall short.”

The professional and novice artists had no experience using AI image generators so their creativity ratings might have been higher if they had more experience with the AI program or had the opportunity to edit their prompts to fine tune the digital images.

While the professional artists fared well in this study, the tables could be turned soon as AI programs advance at exponential rates, Seli said.

“If AI becomes better at understanding artistic intent or mimicking human-like intuition, the gap between AI and professional artists could narrow or even disappear,” he said. “However, it’s possible that the uniquely human aspects of creativity - such as emotional depth and lived experience - will continue to set human artists apart. Whether that distinction matters in the long run remains an open question. Only time will tell.”

Article: “Beyond the Brush: Human Versus AI Creativity in the Realm of Generative Art,” Paul Seli, PhD, Lucas Bellaiche, MA, and Sarah Spooner, BA, Duke University; Anya Ragnhildstveit, MA, University of Cambridge; William Orwig, PhD, Harvard University; and Nathaniel Barr, PhD, Sheridan College; Psychology of Aesthetics, Creativity, and the Arts, published online Feb. 27, 2025.

Contact: Paul Seli, PhD, may be contacted at paul.seli@duke.edu.

The American Psychological Association, in Washington, D.C., is the largest scientific and professional organization representing psychology in the United States. APA’s membership includes 173,000 researchers, educators, clinicians, consultants and students. Through its divisions in 54 subfields of psychology and affiliations with 60 state, territorial and Canadian provincial associations, APA works to advance the creation, communication and application of psychological knowledge to benefit society and improve lives.

Journal

Psychology of Aesthetics Creativity and the Arts

DOI

10.1037/aca0000743

Method of Research

Experimental study

Subject of Research

People

Article Title

Beyond the Brush: Human Versus AI Creativity in the Realm of Generative Art

Article Publication Date

27-Feb-2025

Closing the gaps — MXene-coating filters can enhance performance and reusability

Drexel University

MXene-coated air filter — image:
Researchers from Drexel University are reporting that a coating of MXene nanomaterial can improve air filters and make them easier to reuse.
view more
Credit: Drexel University

Despite improvements to air filtration technology in the aftermath of the COVID-19 pandemic, some of the smallest particles — those of automobile and factory emissions — can still make their way through less efficient, but common filters. An interdisciplinary team of researchers from Drexel University’s College of Engineering have introduced a new way to improve textile-based filters by coating them with a type of two-dimensional nanomaterial called MXene.

Recently published in C Journal of Carbon Research, the team’s research reports that a non-woven polyester textile — a low-cost material with low filtration efficiency — coated with a thin layer of MXene nanomaterial can turn it into a potent filter capable of pulling some of the finest nanoparticles from the air.

“It can be challenging for common filters to contend with particles less than 100 nanometers, which include those emitted by industrial processes and automobiles,” said Michael Waring, PhD, a professor in Drexel’s College of Engineering, and coauthor of the research. “Being able to augment a filter, through a simple coating process, to make it effective against these emissions is a significant development.”

The researchers report that a polyester textile coated with a titanium carbide MXene can reach approximately 90% filtration efficiency for particles as small as 15-30 nanometers — the size of viruses and the finest dust particles — meaning that it could be effective as an augmentation process to air filters located in urban or industrial environments.

MXenes, a family of nanomaterials discovered by Drexel researchers in 2011, have previously demonstrated proficiency in filtration applications, including water treatment, kidney dialysis, and hydrogen capture. The materials have also shown that they can enhance filters that remove airborne viruses in medical settings.

“With increasing manufacturing volume and decreasing price, MXenes are finding an increasing number of applications,” said Yury Gogotsi, PhD, Distinguished University and Bach professor in the College of Engineering, who led the material’s development and was among the researchers who discovered MXenes and founded MXene, Inc., a company that now manufactures them. “Particularly in the fields that require large amounts of material.”

The latest discovery is a significant step in the exploration of the nanomaterials because it shows their capability to contend with some of the smallest particles in the air and that they can easily be integrated into a filter manufacturing process.

“Our ongoing research continues to reveal the potential of MXene coatings,” Gogotsi said. “The fact that this highly conductive nanomaterial is also hydrophilic means that it can be dispersed in water to produce a coating that can easily be applied to virtually any substrate, including air filters. We are just scratching the surface of its capabilities.”

In the processes of testing the MXene-coated filters, the team made an additional discovery, that pretreating the filters with magnesium salt assisted in the MXene coating process and improved the filter’s performance by 25% — to achieve a maximum efficiency of about 90% for virus-size nanoparticles, which many common filters don’t capture.

The addition of alkaline earth metal ions, such as magnesium, improved the MXene coating process. According to the researchers, this sort of chemical preparation further activates the surface of the MXene, which helps the material spread uniformly across the filter, creating a thicker coating and more complex channels through the material, which all contribute to better filtration capability.

The researchers tested untreated; MXene-coated; and MXene-coated and magnesium ion-treated filters in a vacuum-sealed chamber containing aerosolized sodium chloride (rock salt), and measured removal for particles ranging from 5.6 to 560 nanometers. The filters that had been MXene-coated and magnesium ion-treated performed significantly better than the other two at capturing particles of all sizes in the range, down to 15 nanometers.

In addition to their ability to improve filtration, MXenes are also highly conductive – a trait the researchers theorized could be leveraged to enable the filters to clean themselves. They tested the idea by applying an electric current to the MXene-coated filter, which raised its temperature to 100 degrees Celsius — enough to carefully burn off some particles and debris on the filter and restore its original filtration quality.

“Studies like this are encouraging for real-world application of MXenes in air filtration,” said Prastuti Upadhyay, a Materials Science and Engineering undergraduate student in the College of Engineering, who was mentored by Drexel postdoctoral researcher Stefano Ippolito, PhD, and was the lead author of the paper. “But it should be noted that our air filters could still be improved by focusing on optimizing the MXene structure, pretreatment ions and the filter substrate. This leaves room for many exciting possibilities for this line of research.”

DOI

10.3390/c11010013

Method of Research

Experimental study

Article Title

Nanoparticle Air Filtration Using MXene-Coated Textiles

Study finds support for solar energy has become politically polarized

Sentiment Analysis of Solar Energy in U.S. Cities: A 10-Year Analysis Using Transformer-Based Deep Learning

North Carolina State University

Support for Solar Energy Has Become Politically Polarized — image:
A new analysis of social media posts finds that public support for solar energy remains high, though that support declined significantly from 2016 to 2022. The study also found that solar power has become an increasingly polarized issue, with the shift in support being driven largely by opposition to solar power among people in regions that lean Republican.
view more
Credit: Becky Kirkland, NC State University

A new analysis of social media posts finds public support for solar energy remains high, though that support declined significantly from 2016 to 2022. The study also found solar power has become an increasingly polarized issue, with the shift in support being driven largely by opposition to solar power among people in regions that lean Republican.

“The U.S. saw significant growth in the solar energy sector between 2013 and 2022, but that growth has not been spread evenly across the country – some areas have seen more deployment of solar energy technologies than other areas,” says Serena Kim, corresponding author of the study and an assistant professor of public administration at North Carolina State University. “We wanted to assess attitudes toward solar energy at both the local and regional level, as well as how those attitudes have changed over time.

“These attitudes can influence policies that have a significant impact on sustainable energy, and our findings suggest that attitudes are largely being driven by politics,” Kim says.

For this study, the researchers collected more than 8 million posts related to solar energy from the social media platform previously known as Twitter, spanning the years 2013-2022. These posts are from users who had included location data on their public-facing accounts.

The researchers then used analytical tools to characterize the sentiment of each post as positive, negative or neutral toward solar energy.

“We found that a majority of people support solar energy efforts – that has not changed,” Kim says. “However, that majority shrank quite a bit.”

Support for solar energy peaked in 2016, with 65% of posts conveying a positive sentiment toward solar energy, while only 7% were negative. By 2022, 58% of posts were positive, and the number of negative posts had increased to 28.4%.

“One of the telling findings here is that the proportion of neutral posts decreased from 41.9% in 2016 to 13.3% in 2022,” Kim says. “This tells us that sentiment toward solar energy has become significantly more polarized.”

The researchers also found big differences in support for solar power from state to state. On average, the five states that were most positive toward solar energy between 2013 and 2022 were Vermont, Massachusetts, New York, New Mexico and Colorado. The five states that were most negative toward solar power were Alabama, Louisiana, Oklahoma, Idaho and North Dakota.

“In addition to regional differences, we also found significant differences over time associated with an area’s political preferences,” Kim says. “For example, in 2016, support for solar power in Republican-leaning municipalities was only 2% lower than that in Democratic-leaning municipalities. But by 2022, support for solar power in Republican municipalities was 30% lower.

“If we are interested in expanding cleaner and decentralized energy resources, we need to understand public sentiment toward technologies like solar energy,” Kim says. “This work is a step in that direction, offering insight into how that sentiment is changing – and suggestions as to what is driving that change.”

The researchers have made city- and state-level data on public perceptions of solar energy publicly available at https://solarsentiment.org.

A paper on the work, “Sentiment Analysis of Solar Energy in U.S. Cities: A 10-Year Analysis Using Transformer-Based Deep Learning,” is published open access in Journal of Computational Social Science. The paper was co-authored by Crystal Soderman of the University of Colorado Denver; and Lan Sang of the University of Colorado Boulder.

This work was done with support from the University of Colorado Denver’s Presidential Initiative on Urban and Place-Based Research and the Infrastructure Informatics Grand Challenge Initiative.