AI use makes us overestimate our cognitive

performance

New research warns we shouldn’t blindly trust Large Language Models with logical reasoning –– stopping at one prompt limits ChatGPT’s usefulness more than users realize

Aalto University

FacebookXLinkedInWeChatBlueskyMessageWhatsAppEmail

 

image: 

Researcher Robin Welsch.

view more 

Credit: Matti Ahlgren / Aalto University

New research warns we shouldn’t blindly trust Large Language Models with logical reasoning –– stopping at one prompt limits ChatGPT’s usefulness more than users realise.

When it comes to estimating how good we are at something, research consistently shows that we tend to rate ourselves as slightly better than average. This tendency is stronger in people who perform low on cognitive tests. It’s known as the Dunning-Kruger Effect (DKE) –– the worse people are at something the more they tend to overestimate their abilities and the “smarter” they are, the less they realise their true abilities.

However, a study led by Aalto University reveals that when it comes to AI, specifically, Large Language Models (LLMs), the DKE doesn’t hold, with researchers finding that all users show a significant inability to assess their performance accurately when using ChatGPT. In fact, across the board, people overestimated their performance. On top of this, the researchers identified a reversal of the Dunning-Kruger Effect ––   an identifiable tendency for those users who considered themselves more AI literate to assume their abilities were greater than they really were.

‘We found that when it comes to AI, the DKE vanishes. In fact, what’s really surprising is that higher AI literacy brings more overconfidence,’ says Professor Robin Welsch. ‘We would expect people who are AI literate to not only be a bit better at interacting with AI systems, but also at judging their performance with those systems – but this was not the case.’

The finding adds to a rapidly growing volume of research indicating that blindly trusting AI output comes with risks like ‘dumbing down’ people’s ability to source reliable information and even workforce de-skilling. While people did perform better when using ChatGPT, it’s concerning that they all overestimated that performance.

‘AI literacy is truly important nowadays, and therefore this is a very striking effect. AI literacy might be very technical, and it’s not really helping people actually interact fruitfully with AI systems’, says Welsch.

‘Current AI tools are not enough. They are not fostering metacognition [awareness of one’s own thought processes] and we are not learning about our mistakes,’ adds doctoral researcher Daniela da Silva Fernandes. ‘We need to create platforms that encourage our reflection process.’

The article was published on October 27th in the journal Computers in Human Behavior.

Why a single prompt is not enough

The researchers designed two experiments where some 500 participants used AI to complete logical reasoning tasks from the US’s famous Law School Admission Test (LSAT). Half of the group used AI and half didn’t. After each task, subjects were asked to monitor how well they performed –– and if they did that accurately, they were promised extra compensation.

‘These tasks take a lot of cognitive effort. Now that people use AI daily, it’s typical that you would give something like this to AI to solve, because it’s so challenging’, Welsch says.

The data revealed that most users rarely prompted ChatGPT more than once per question. Often, they simply copied the question, put it in the AI system, and were happy with the AI’s solution without checking or second-guessing.

‘We looked at whether they truly reflected with the AI system and found that people just thought the AI would solve things for them. Usually there was just one single interaction to get the results, which means that users blindly trusted the system. It’s what we call cognitive offloading, when all the processing is done by AI’, Welsch explains.

This shallow level of engagement may have limited the cues needed to calibrate confidence and allow for accurate self-monitoring. Therefore, it’s plausible that encouraging or experimentally requiring multiple prompts could provide better feedback loops, enhancing users’ metacognition, he says.

So what’s the practical solution for everyday AI users?

‘AI could ask the users if they can explain their reasoning further. This would force the user to engage more with AI, to face their illusion of knowledge, and to promote critical thinking,’ Fernandes says.

Full article: ‘AI makes you smarter but none the wiser: The disconnect between performance and metacognition’, Computers in Human Behavior, 10.1016/j.chb.2025.108779

---

Journal

Computers in Human Behavior

Research has found that using AI gives us a false sense of confidence.

Credit

Matti Ahlgren / Aalto University

DOI

10.1016/j.chb.2025.108779 

Article Title

AI makes you smarter but none the wiser: The disconnect between performance and metacognition

Article Publication Date

27-Oct-2025

AI produces shallower knowledge than web search

PNAS Nexus

FacebookXLinkedInWeChatBlueskyMessageWhatsAppEmail

Learning about a topic by interacting with AI chatbots like ChatGPT rather than following links provided by web search can produce shallower knowledge. Advice given on the basis of this shallow knowledge tends to be sparser, less original, and less likely to be adopted by others. Shiri Melumad and Jin Ho Yun conducted seven experiments with thousands of online participants who were randomly assigned to learn about various topics, including how to plant a vegetable garden, how to lead a healthier lifestyle, or how to cope with financial scams, using either large language models (LLMs) or traditional Google web search links. Participants then wrote advice based on what they learned. Participants who used LLMs spent less time engaging with search results and reported developing shallower knowledge compared to those using web links, even when the underlying facts were identical. When forming advice, LLM users invested less effort and produced content that was objectively shorter, contained fewer factual references, and showed greater similarity to other participants' advice. In an experiment with 1,501 independent evaluators, recipients—who were unaware of where the advice came from—rated advice written after LLM searches as less helpful, less informative, and less trustworthy compared to advice based on web search, and were less willing to adopt the LLM-derived advice. While LLMs are undeniably efficient, relying on pre-synthesized summaries can transform learning from an active quest to a passive activity. According to the authors, LLMs are thus potentially less useful than web search if the goal is developing procedural knowledge—an understanding of how to actually do things.

---

Journal

PNAS Nexus

Article Title

Experimental evidence of the effects of large language models versus web search on depth of learning

Article Publication Date

28-Oct-2025

Pusan National University researchers show how AI can help in fashion trend prediction

Using a novel prompting technique, researchers showcase the potential of ChatGPT to capture upcoming fashion trends

Pusan National University

FacebookXLinkedInWeChatBlueskyMessageWhatsAppEmail

 

image: 

Using the novel top-down prompting strategy and expert guidance, ChatGPT can become a valuable tool for fashion trend prediction, especially for fashion students and small brands.

view more 

Credit: Yoon Kyung Lee from Pusan National University

Fashion trend forecasting helps companies predict which clothes will be popular in upcoming seasons. Traditionally, this has relied on experts’ intuition, experience and creativity. More recently, big-data analysis has been incorporated, offering deeper insights into consumer behavior. However, such methods pose technical barriers and remain out of reach for fashion students or small brands.

Recent developments in artificial intelligence (AI) can balance the scales. Large language models (LLMs) like ChatGPT have made big data analysis readily available to the public.  LLMs draw from vast societal and cultural data and can potentially be used for predicting fashion trends. However, given their current limitations, such as hallucinations and factual errors, it is imperative to verify their suitability and to develop structured prediction methods.

In a new study, Assistant Professor Yoon Kyung Lee and Master’s student Chaehi Ryu, from the Department of Clothing and Textiles at Pusan National University, South Korea, developed a new approach for predicting fashion trends using ChatGPT. “Rather than simply asking ‘What fashion will be popular in the future?’, we designed a systematic strategy for prompting the AI for more specific and consistent answers,” explains Dr. Lee. “We also compared ChatGPT's predictions to an actual trend agency’s report.” Their findings were published in the Clothing and Textiles Research Journal on September 26, 2025.

The researchers first verified ChatGPT’s characteristics in fashion trend prediction using general prompts about 2023 fall/winter men’s fashion trends. Based on the results, they developed a new Top-Down Prompting (TDP) technique based on the Lotus Blossom (LB) brainstorming approach, for more accurate and specific forecasting.

TDP starts with a central problem prompt, a general query to predict ‘Fashion Trends’, and then expands into sub-problem prompts, including queries on “Silhouette,” “Materials,” “Key Items,” “Garment Details,” “Decorative Elements,” “Color,” “Moods,” and “Prints and Patterns.” 

The researchers used this approach to predict men’s fashion trends for fall/winter 2024 using ChatGPT-3.5 and ChatGPT-4 Classic. The responses were compared and validated against the fall/winter 2024 men’s fashion trend predictions by the Official Fashion Trend Information Company (OFTIC), and the analysis was reviewed by two fashion experts.

Analysis showed that ChatGPT’s predictions mostly reflected established or generalized fashion ideas, rather than forward-looking or innovative designs. Moreover, it accurately identified only 9 out of 39 trends predicted in OFTIC’s report. Notably, however, both models predicted emerging themes, including gender fluidity and statement coats.

“While the prediction accuracy of ChatGPT is low, what's intriguing is that it captured new trends not found in existing data,” notes Dr. Lee. “AI can sense cultural shifts and open up new creative directions.”

Although ChatGPT should not yet be viewed as a definitive forecasting tool, it can effectively complement expert-led analysis. This is especially valuable for fashion students and small brands, helping them achieve more accurate and nuanced trend forecasts. For fashion education, the researchers also developed a TDP-based hybrid conceptual framework for fashion trend forecasting, integrating both AI-analysis and expert-knowledge.

Overall, this study shows how AI tools can make fashion trend forecasting more systematic and accessible.

 

***

 

Reference
DOI: 10.1177/0887302X251371969  
 

About Pusan National University
Pusan National University, located in Busan, South Korea, was founded in 1946 and is now the No. 1 national university of South Korea in research and educational competency. The multi-campus university also has other smaller campuses in Yangsan, Miryang, and Ami. The university prides itself on the principles of truth, freedom, and service and has approximately 30,000 students, 1,200 professors, and 750 faculty members. The university comprises 14 colleges (schools) and one independent division, with 103 departments in all.
Website: https://www.pusan.ac.kr/eng/Main.do
 

About Assistant Professor Yoon Kyung Lee
Dr. Yoon Kyung Lee is an Assistant professor of Fashion Design for Dept. Clothing and Textiles at Pusan National University. Her research interests include Sustainability in fashion, Creativity, Innovative design education, Digital fashion technologies, AI & neuroscience in fashion. Before coming her current position, she worked as an Assistant research professor at Seoul National University. She completed the Postdoctoral training at DeLong’s lab at University of Minnesota. She received a PhD in Aesthetics in Dress from Seoul National University and an MFA in Fashion and Textile Design from the Institute European of Design (IED) Milan, Italy.

---

Journal

Clothing and Textiles Research Journal

DOI

10.1177/0887302X251371969 

Method of Research

Computational simulation/modeling

Subject of Research

Not applicable

Article Title

How the Field of Fashion can use ChatGPT to Predict Fashion Trends

Single-cell atlas reveals gene networks shaping peanut peg formation and subterranean pod development

Maximum Academic Press

FacebookXLinkedInWeChatBlueskyMessageWhatsAppEmail

Using single-nucleus RNA sequencing, the researchers mapped over 50,000 individual cells from both aerial and subterranean pegs, identifying distinct cell types and gene expression patterns that guide the downward growth of fertilized ovaries into the soil. Their analysis revealed that hormone signaling, particularly the auxin pathway, plays a critical role in regulating gravity-responsive growth and tissue differentiation.

Peanut (Arachis hypogaea L.) produces flowers above ground, but its fertilized ovaries extend downward to form subterranean pods—a process known as pegging. The peg, or gynophore, is a positively geotropic organ that senses gravity and directs seed formation underground. Previous studies have implicated plant hormones such as auxin, ethylene, cytokinin, and gibberellins in this process, but the molecular details have remained elusive. Advances in single-cell RNA sequencing now allow scientists to dissect cell-specific transcriptional activity and reconstruct developmental trajectories. Given the peg’s hybrid nature—combining stem-like structure with root-like behavior—cell-level resolution is critical to reveal the genetic and hormonal mechanisms underlying its formation.

A study (DOI:10.48130/ph-0025-0019 ) published in Plant Hormones on 16 September 2025 by Hao Liu’s & Wenyi Wang’s team, Guangdong Academy of Agricultural Sciences & South China Agriculture University, provides a detailed molecular blueprint of peg formation, offering valuable insights into the mechanisms that ensure successful pod development and potentially paving the way for higher-yield peanut varieties.

To dissect how peanuts drive fertilized ovaries into the soil to form underground pods, the researchers combined single-nucleus/single-cell RNA sequencing (snRNA-seq) and bulk RNA sequencing on both aerial and subterranean pegs, generating high-resolution transcriptional maps of 23,539 aerial peg cells and 27,364 subterranean peg cells. They clustered these cells (20 aerial clusters and 22 subterranean clusters) using UMAP and t-SNE, then annotated nine major cell types in each peg type through marker-gene expression and RNA in situ hybridization, including ovule, cortex, xylem, epidermis/exodermis, and phloem/cambium. With these cell identities established, they reconstructed developmental “pseudo-time” trajectories to model how peg cells differentiate as the peg grows downward and enters the soil. In aerial pegs, this analysis resolved 11 developmental states and 7,086 pseudo-time differentially expressed genes (DEGs), including 87 key transcription factors (TFs) such as members of the MYB, AP2/ERF, WRKY, and AGL families. Network analysis highlighted TFs like AP2/ERF and WRKY41/53/70 as central regulators of aerial peg growth. A similar pseudo-time reconstruction in subterranean pegs identified 4,830 DEGs and 42 core TFs, with enriched pathways in secondary metabolism, fatty acid degradation, and flavonoid biosynthesis, suggesting metabolic reprogramming after soil penetration. Integrating single-cell and bulk datasets revealed cell type–specific regulators of cortex differentiation, including auxin transport and response genes such as PIN3 and ARF8, and showed that hormone signaling pathways—especially auxin signaling—shift sharply between aerial and subterranean pegs. Hormone profiling confirmed major differences in indole-3-acetic acid (IAA), ABA, salicylic acid, and cytokinins between aboveground and belowground tissues. Together, these methods produced the first cell-by-cell regulatory atlas of peanut peg development and showed that auxin-driven transcriptional programs underlie gravity-guided peg penetration and pod initiation.

This single-cell atlas provides a powerful resource for understanding the gene regulatory networks controlling peanut reproduction. By elucidating the hormonal and transcriptional mechanisms that mediate peg differentiation and soil penetration, the findings offer molecular targets for breeding peanut cultivars with improved pod-setting efficiency and yield stability under variable environmental conditions. The data also open possibilities for manipulating hormone signaling pathways to optimize gravity responses and enhance reproductive success in other legume crops.

###

References

DOI

10.48130/ph-0025-0019

Original Source URL

https://doi.org/10.48130/ph-0025-0019

Funding Information

This work was supported by the Guangdong Provincial Key Research and Development Program-Modern Seed Industry (2022B0202060004), the National Key Research and Development Project (2023YFD1202800), the Open Competition Program of Top 10 Critical Priorities of Agricultural Science and Technology Innovation for the 14th Five-Year Plan in Guangdong Province (2022SDZG05), The National Natural Science Foundation of China (32272121, 32172051, 32301869), the Guangdong Science and Technology Plan Project (2023B1212060038), the Guangdong Basic and Applied Basic Research Foundation (2021A1515010811, 2023A1515010098, 2023A1515010569, 2023A04J0776), the China Agriculture Research System of MOF and MARA (CARS-13), the Special Fund for Scientific Innovation Strategy-Construction of High Level Academy of Agriculture Science (R2020PY-JX004, R2020PY-JG005, R2021PY-QY003, R2022YJ-YB3025), the Foundation of Director of Crop Research Institute of Guangdong Academy of Agriculture Sciences (202101, 202201, 202306), Special Funds for the Revitalization of Agriculture through Seed Industry under the Provincial Rural Revitalization Strategy (2022-NPY-00-022), the Project of Collaborative Innovation Center of GDAAS (XTXM202203), the Special Support Program of Guangdong Province (2021TX06N789), Science and Technology Planning Project of Heyuan City (Heyuan She Nong Da Zhuan Xiang 2022002), and the Science and Technology Project of Qingyuan City 2023 (2023KJJ002).

About Plant Hormones

Plant Hormones (e-ISSN 3067-221X) is an open access, online-only, academic journal publishing rigorously peer-reviewed original articles, reviews, break-through methods, editorials, and perspectives on broad aspects of plant hormone biosynthesis, signal transduction, and crosstalk. The journal primarily publishes fundamental research that represents significant advances or new insight into specialized areas of plant hormones, and review articles that provide comprehensive and critical review of current research areas and offer directions or perspectives for future research. The journal publishes applied research that has significant implications for the development of agriculture, horticulture, and forestry. Plant Hormones also provides a community forum by publishing editorials and perspective papers for expressing opinions on specific issues or new perspectives about existing research on particular topics. Plant Hormones is hosted by Chongqing University, and published by Maximum Academic Press.

---

Journal

Plant Hormones

DOI

10.48130/ph-0025-0019 

Method of Research

Experimental study

Subject of Research

Not applicable

Article Title

Single-cell resolution transcriptional atlases of aerial and subterranean pegs reveals auxin signaling associated with peg development in peanut (Arachis hypogaea L.)

 

Arts-based methods are transforming leadership education – new study highlights the importance of creativity and empathy for future leaders

University of Eastern Finland

FacebookXLinkedInWeChatBlueskyMessageWhatsAppEmail

Arts-based methods can significantly enhance leadership education and better prepare students to address the major societal challenges of our time, a new study from the University of Eastern Finland shows. The study examined a course on innovation cultures delivered at the University of Eastern Finland between 2013 and 2023, where students employed visual arts, music, drama and play as part of their learning process.

According to the study, arts-based methods support the development of students’ emotional intelligence, creativity and critical thinking. They encourage students to reflect on their experiences, question their assumptions and explore leadership from new perspectives. Arts-based methods also promote collaborative learning and create learning environments that mirror the complexity and demands of workplaces.

“Students felt that the course helped them gain a deeper understanding of innovation cultures, while encouraging them to apply creative thinking both in their studies and in their future careers. Many reported increased confidence and greater courage to experiment with new things, even in cases where they did not have a strong background in the arts,” says University Lecturer Pasi Hirvonen, one of the course teachers.

The study shows that arts-based methods are not only beneficial in leadership education, but they also offer broader opportunities to develop pedagogical approaches that support students in becoming responsible and empathetic change leaders. Arts-based methods strengthen students’ ability to navigate uncertainty, work in teams and networks, and harness emotions and creativity as drivers of change.

According to the course teachers, successful implementation of the course requires robust pedagogical expertise, the ability to navigate uncertainty and, above all, institutional support.

“Creating a safe and encouraging learning environment is essential for students to feel confident enough to engage with creative learning methods,” Professor Hanna Lehtimäki notes.

The study highlights the need to reform business education to better respond to current and future challenges. Traditional teaching methods should be supplemented by approaches that support students’ holistic growth.

“Arts-based methods offer students a unique opportunity to develop skills that are urgently needed in modern workplaces. Empathy, critical thinking and creative problem-solving are not just academic goals but essential skills that enable future professionals to thrive in a rapidly changing and increasingly complex world,” Hirvonen concludes.

---

Journal

The International Journal of Management Education

DOI

10.1016/j.ijme.2025.101306 

Article Title

Innovation pedagogy in management education: Student-centered learning with arts-based methods

Article Publication Date

24-Oct-2025

 

Scientists develop an efficient method of producing proteins from E. coli

This development could enhance microbial protein production and drive progress in biomanufacturing

Nagoya University

FacebookXLinkedInWeChatBlueskyMessageWhatsAppEmail

 

image: 

Proteins are synthesized through two processes involving DNA: transcription, which converts DNA into mRNA; and translation, where ribosomes read the mRNA and sequentially link amino acids to form proteins. This image illustrates the translation process accelerated to produce proteins more efficiently.

 

view more 

Credit: Teruyo Ojima-Kato

Proteins sourced from microorganisms are attracting attention for their potential in biomanufacturing a variety of products, including pharmaceuticals, industrial enzymes, and diagnostic antibodies. These proteins can also be used for converting resources into biofuels and bioplastics, which could serve as viable alternatives to petroleum-based fuels and products. Therefore, efficiently producing microbial proteins could make a significant contribution to sustainable manufacturing.

Producing proteins from Escherichia coli (E. coli) has become popular due to its cost-effectiveness and efficiency. However, yields of protein production in E. coli may be reduced depending on the specific gene sequence of the target protein.

In a study published in the journal RSC Chemical Biology, a research group in Japan has successfully developed a new technology that improves the efficiency of protein production in E. coli.

The reduction in protein production is believed to be a contributing factor to  ribosome stalling. Ribosomes synthesize proteins using genetic information carried by messenger RNA (mRNA). However, when ribosomes are unable to continue the translation process for some reason, protein synthesis is halted.

"In our previous research, we found that adding a short peptide sequence composed of four amino acids—serine, lysine, isoleucine, and lysine—to the N-terminus of a protein reduces ribosomal stalling and significantly improves translation efficiency in E. coli," said Associate Professor Teruyo Ojima-Kato from Nagoya University's Graduate School of Bioagricultural Sciences.

Based on this finding, Kato and Professor Hideo Nakano of Nagoya University, in collaboration with researchers from the National Institute of Advanced Industrial Science and Technology and Waseda University, conducted a study to identify short translational-enhancing peptides (TEPs) that can prevent ribosome stalling.

The researchers first created a tetrapeptide library, a collection of random peptide sequences formed by all possible combinations of the four amino acids. This library contains the arrangements of the 20 amino acids that make up proteins, totaling 160,000 distinct tetrapeptides. Using this library, they conducted a comprehensive analysis and identified several novel TEPs that effectively prevent ribosome stalling.

Next, the researchers attempted to evaluate the translation-enhancing effects of the 160,000 tetrapeptides. To assess such large numbers of tetrapeptides, they developed an artificial intelligence (AI) prediction model based on data from about 250 experiments.

The research group conducted three rounds of AI predictions, demonstrating that their AI model accurately predicts the strength of translation enhancement for all 160,000 tetrapeptides. This result suggests that AI-based predictive models could be helpful in the rational design of peptide sequences that can be readily translated into target proteins.

"We present a novel approach for efficient production of proteins using short peptide sequences," said Kato. "This technique can be applied to more efficient production of enzymes, which play a significant role in the biorefinery sector that generates chemicals and fuels from renewable resources. Our findings could provide fundamental technology to support sustainable manufacturing that does not depend on petroleum."

 

---

Journal

RSC Chemical Biology

Article Publication Date

22-Oct-2025