From measurement to meaning: new research introduces a learning architecture for the age of AI

Redesigning educational feedback systems to integrate human judgment with AI-supported learning architectures

ECNU Review of Education

Facebook

XLinkedInWeChatBlueskyMessageWhatsAppEmail

SHANGHAI–As artificial intelligence (AI) becomes increasingly embedded in education, schools now have more data about learning than ever before. Yet a paradox remains: more measurement does not necessarily lead to deeper understanding.

In a new study titled “A Distributed Architecture Integrating Educational Philosophy and AI-Driven Learning Design: The PDP–ICEE Learning System,” Ruojun Zhong (YEE Education) argues that modern education has fallen into what she describes as an “assessment trap.” As systems become more sophisticated at collecting and analyzing performance data, learning itself risks being reduced to what can be observed, quantified, and compared.

“The challenge is not that we lack data,” Zhong explains. “It is that our feedback systems often stop at measurement. They produce results, but those results rarely return to reshape how learning is understood or designed.”

The study proposes a shift from evaluation-centered education to what Zhong calls “learning from learning.” Rather than focusing solely on outcomes, the proposed model redesigns feedback loops so that data becomes interpretable insight—helping learners, educators, and institutions continuously adapt.

At the core of the model is a human-in-the-Loop principle. While AI can detect patterns across large-scale learning data, human judgment remains essential for interpretation, context, and ethical direction. By embedding human meaning-making within AI-supported systems, the model seeks to transform assessment from a terminal judgment into an ongoing process of reflection.

The research introduces a distributed learning architecture that integrates educational philosophy with AI-driven design. Instead of treating learning as a linear sequence of tasks and scores, the system organizes learning as evolving action pathways and reflective growth patterns. These mechanisms aim to make long-term development visible without reducing it to standardized metrics.

Importantly, the study does not position AI as a replacement for educators. Rather, it reframes AI as a cognitive partner—supporting schools in building feedback systems that are adaptive, interpretable, and human-centered. As automation expands across sectors, the paper argues that education must move beyond simply collecting more data.

The future of AI in schools depends not on how much learning can be measured, but on whether educational systems can develop the capacity to understand and evolve through their own feedback.

“In the age of AI,” Zhong concludes, “the real question is whether education can design systems that remain responsive to meaning—not just to metrics.”

Reference
DOI: https://doi.org/10.1177/20965311261422768

---

Journal

ECNU Review of Education

DOI

10.1177/20965311261422768 

Method of Research

Computational simulation/modeling

Subject of Research

Not applicable

Article Title

A Distributed Architecture Integrating Educational Philosophy and AI-Driven Learning Design: The PDP–ICEE Learning System

Article Publication Date

17-Mar-2026

 

Guidance for safer AI-enabled medical devices: Dresden researchers highlight the importance of human factors

Technische Universität Dresden

Facebook

XLinkedInWeChatBlueskyMessageWhatsAppEmail

 

image: 

Lead author Rebecca Mathias.

view more 

Credit: EKFZ - Anja Stübner

AI-enabled medical devices promise improved medical care and support for healthcare professionals. However, the safety and performance of such systems not only depends on algorithms or technical specifications. It is equally important how people use these devices and applications. In a recent publication in the scientific journal NEJM AI, a research team led by Prof. Stephen Gilbert from Else Kröner Fresenius Center (EKFZ) for Digital Health at TUD Dresden University of Technology systematically analyzes risks that can arise in human-AI interactions and makes recommendations for manufacturers and regulatory evaluators.

The authors show that existing regulatory requirements for approval have so far only partially addressed many of these so-called “human factors-related risks”. This can create gaps that impact the safety and quality of care. To address these, the researchers identify seven key risks and develop practical recommendations for action that can be integrated into existing regulatory and documentation processes.

Risks in the use of AI systems

AI-based medical devices can be used in various areas of clinical environments. In radiology, for example, they assist in detecting cancer. Clinical decision support systems help select personalized therapies for patients.  AI can also support real-time monitoring and early warning systems, as well as chatbots for applications such as patient communication and software that automatically generate medical reports or summarize findings. The analysis focuses on risks that may arise in the practical use of such AI systems. These include, for example, an increased likelihood of outputs being misunderstood or misinterpreted due to the sometimes-opaque nature of AI systems. Problems can also occur when trust in the application is miscalibrated: resulting in users either relying too heavily on AI assistance or ignoring relevant recommendations. The researchers also point to the risk of automation bias: the tendency to uncritically adopt recommendations from automated systems, potentially overlooking errors or forgoing independent judgment. Additional risks include potential deskilling, technostress among users, an unchecked expansion of indications beyond the originally intended scope (indication creep), and errors related to system changes or different operating modes. Such factors can create additional burdens or unexpected failures in clinical practice – even when the technical performance of a system itself is strong.

A practical guide for manufacturers and evaluators

For their analysis, the research team evaluated existing standards on usability and safety, regulatory guidelines, alongside the scientific literature on AI in healthcare. In addition, expert discussions from the fields of clinical application, regulation, and human factors were incorporated. The result is a practical guide, that fills a gap in current standards, with seven recommendations. These are intended to support manufacturers and evaluators both before and after a product is placed on the market. The aim is to identify AI-specific risks in interaction with human users at an early stage and to address them systematically.

The framework recommends developing and deploying AI-based medical devices in a way that clearly defines the users, in which context the systems are applied, and which tasks are assigned to humans and which to the system. Furthermore, results should be presented in a way that is easy to understand, integrated into existing clinical workflows, and supplemented by training where needed as well as safe fallback options in the event of system failures. The authors emphasize the importance of continuous monitoring after market entry. Usage patterns, potential misuse, or overreliance on AI systems should be systematically observed and corrected as needed. Changes to the systems must also be communicated transparently so that work processes can be adjusted accordingly.

The recommendations are deliberately formulated in general but regulatory-aligned terms so that they can be applied to different AI-enabled medical devices and application scenarios. In a next step, the researchers aim to test and further develop their recommendations based on concrete pilot applications with AI-enabled medical devices. In the long term, human factors should be systematically considered in the regulation and evaluation of AI-based health technologies – reducing avoidable risks while supporting safe innovation in medicine.

The article was authored by researchers from TU Dresden (EKFZ for Digital Health, Chair of Industrial Design Engineering, and Faculty of Business and Economics), in collaboration with experts from the University of Oxford (United Kingdom) and Geneva University Hospital (Switzerland).

 

Publication

Rebecca Mathias, Anne Schmitt, Mateo Campos, Baptiste Vasey, Sebastian Lorenz, Peter McCulloch, Stephen Gilbert: Evaluation of Human Factors-Related Risks in AI-Enabled Medical Devices: A Practical Guide, NEJM AI, 2026. Link: https://ai.nejm.org/doi/full/10.1056/AIpc2501297

 

Else Kröner Fresenius Center (EKFZ) for Digital Health

The EKFZ for Digital Health at the Faculty of Medicine at TUD Dresden University of Technology and University Hospital Carl Gustav Carus Dresden was established in September 2019. It receives funding of around 40 million euros from the Else Kröner Fresenius Foundation for a period of ten years. The center focuses its research activities on innovative, medical and digital technologies at the direct interface with patients. The aim here is to fully exploit the potential of digitalization in medicine to significantly and sustainably improve healthcare, medical research and clinical practice.

---

Journal

NEJM AI

DOI

10.1056/AIpc2501297 

Subject of Research

Not applicable

Article Title

Evaluation of Human Factors-Related Risks in AI-Enabled Medical Devices: A Practical Guide

Article Publication Date

26-Mar-2026

 

Artificial intelligence learns to make sense of childhood cancer survivors’ health care needs

St. Jude Children's Research Hospital

Facebook

XLinkedInWeChatBlueskyMessageWhatsAppEmail

 

image: 

(L to R) Co-authors Kiri Ness, PT, PhD, FAPTA, and Melissa Hudson, MD, with corresponding author I-Chan Huang, PhD, all of the St. Jude Department of Epidemiology & Cancer Control.

view more 

Credit: St. Jude Children's Research Hospital

(MEMPHIS, Tenn. – March 26, 2026) Artificial intelligence (AI) could help physicians determine if survivors of childhood cancer need extra support — and the more information included in AI prompting, the better its performance. This finding, published today in Communications Medicine by scientists from St. Jude Children’s Research Hospital, may guide future integration of AI into clinical workflows.

 

The scientists observed how well large language models, a type of AI, could analyze interviews with young survivors and their caregivers to detect multiple symptoms causing severe disruptions in their daily lives. By comparing different prompting approaches, the researchers found that more complex prompts, which provided additional information to the models, performed the best. The results suggest that future efforts to leverage AI to improve survivors’ care should consider these sophisticated prompting strategies over simpler ones.

 

“About 40%-60% of a clinical encounter is a patient talking to their physician about symptoms and related health experiences,” said corresponding author I-Chan Huang, PhD, St. Jude Department of Epidemiology & Cancer Control. “We have provided a proof of concept that large language models could help analyze that underutilized conversational data to detect symptom severity and its functional impact and assist physician decision-making to provide better care to survivors.”

 

Comparing prompting strategies for survivorship

 

Children who have cancer are treated during a critical time in their development, which can have a ripple effect later in life. Cancer- and treatment-related effects can occur long after the initial disease is cured. However, identifying which survivors have symptoms severe enough to need extra, targeted support is difficult for physicians. Much of the data that informs that identification exists in transcripts of conversations and answers to open-ended questions in surveys that cannot be reviewed quickly. Newer language-based AI offers physicians an opportunity to analyze, understand and use that information to help survivors.

 

The researchers interviewed 30 survivors between 8 and 17 years old and their caregivers. Two human experts analyzed the conversation transcripts for signs of excessive pain and fatigue, yielding more than 800 analyzable pieces of information. They categorized the symptoms by severity and their physical, cognitive or social impact. After performing that gold-standard analysis, the scientists gave the same transcripts to two large language models, ChatGPT and Llama, using four styles of prompts. Both models showed an ability to analyze the data in a similar way to the experts, though their performance depended on the prompt used.

 

Prompting is the art of instructing AI to perform a task. The researchers compared four common prompting strategies: two simple and two complex. The simple approaches were zero-shot and few-shot prompting, which provide no or minimal information beyond the basic instructions. These approaches produced unstable and inaccurate results.

 

“We found that simple prompts were not effective,” Huang said. “However, our more sophisticated prompting strategies performed significantly better and had a higher concurrence with our human reviewers.”

 

The two complex strategies were chain-of-thought and generated knowledge prompts. Chain-of-thought uses step-by-step logical instructions, while generated knowledge prompts a model to come up with background information before giving it the instructions. Both complex prompting methods distinguished the physical and cognitive impact of symptoms on survivors well, while having a moderate ability to detect social impacts.

 

Clinical use will require much more testing, but these early results suggest that chain-of-thought, generated knowledge or similar prompting methods should be used in the future. The findings provide one of the first concrete examples of how AI may be able to improve survivorship care.

 

“These AI-driven approaches provide us with a new way to unlock the complex symptom information hidden in the wealth of patient-physician conversations that currently go unused,” Huang said. “By making this information easier to capture and analyze, we can help physicians better identify survivors who need additional support in real time and improve care for this growing population.”

 

Authors and funding

The study’s first author is Jin-ah Sim, formerly of St. Jude. The study’s other authors are Madeline Horan, formerly of St. Jude, now Wake Forest University School of Medicine; Xiaolei Huang, University of Memphis; Minsu Kim, Hallym University; Kumar Srivastava, Kirsten Ness and Melissa Hudson, St. Jude; and Justin Baker, formerly of St. Jude, now Stanford University Medical School.

 

The study was supported by grants from the National Cancer Institute (U01CA195547, R21CA202210, R01CA238368 and R01CA258193), Cancer Center Support (CORE) grant (CA21765) and the American Lebanese Syrian Associated Charities (ALSAC), the fundraising and awareness organization of St. Jude.

 

St. Jude Children's Research Hospital

St. Jude Children’s Research Hospital is leading the way the world understands, treats, and cures childhood catastrophic diseases. As the only National Cancer Institute-designated Comprehensive Cancer Center devoted solely to children, St. Jude advances groundbreaking research and shares its discoveries worldwide to accelerate progress in pediatric medicine. Treatments developed at St. Jude have helped push the overall childhood cancer survival rate from 20% to more than 80% since the hospital opened more than 60 years ago. Through collaboration and innovation, St. Jude is working to ensure that children everywhere have access to the best possible care. To learn more, visit stjude.org, read St. Jude Progress, a digital magazine, and follow St. Jude on social media at @stjuderesearch.

---

Journal

Communications Medicine

DOI

10.1038/s43856-026-01499-5 

Article Publication Date

25-Mar-2026

SCIENTOLOGY HAS COMPETITION FROM AI

AI set to transform personality testing, new research finds

Study from the University of East London shows machine learning can deliver quicker, more detailed assessments of human traits

University of East London

Facebook

XLinkedInWeChatBlueskyMessageWhatsAppEmail

Personality tests are widely used in workplaces to shape recruitment, leadership training and team building. But what if artificial intelligence could make them faster, smarter and more accurate? New research from the University of East London (UEL) suggests that machine learning could significantly improve the way organisational psychologists and managers use one of the most widely used personality tools, the DISC assessment.

DISC assessment classifies individuals into four behavioural styles - Dominance, Influence, Steadiness and Conscientiousness - and is commonly used by organisations to understand how people communicate, lead and work in teams. The model’s appeal lies in its simplicity, allowing organisational psychologists and managers to gain quick insights into behavioural tendencies.

However, traditional DISC assessment relies on straightforward scoring rules that assign people to a single category based on their highest score. While efficient, this approach can sometimes oversimplify personality by overlooking individuals whose traits span more than one behavioural style.

The new study explores whether machine learning can provide a more flexible and data-driven way of analysing DISC responses, offering potentially more accurate and nuanced personality insights. Rather than assigning people to a single category, the approach can also identify blended behavioural patterns when individuals show traits from more than one DISC style.

Using responses from over 1,000 participants, researchers tested several machine learning models to predict DISC personality types based on a standard 40-question assessment. The most successful models achieved accuracy rates of more than 93 per cent, demonstrating that artificial intelligence can reliably replicate traditional DISC classifications.

The research also examines whether the questionnaire itself can be streamlined. By identifying the most informative questions within the assessment, the team shows that a much shorter version can still produce highly reliable results.

A model using just 10 carefully selected questions retained accuracy of more than 91 per cent - suggesting that DISC assessments could be delivered far more quickly without losing much of their predictive strength.

Beyond prediction, the researchers also applied clustering techniques to explore how people naturally group together based on behavioural traits. The analysis reveals four clear personality clusters that closely align with the established DISC categories, while also highlighting subtle overlaps between behavioural styles.

Research lead Dr Mohammad Hossein Amirhosseini, Associate Professor in Computer Science and Digital Technologies at UEL, said the findings show how modern data science can strengthen established psychological tools without losing their practical value.

“DISC has long been valued in workplaces because it is simple and easy to apply,” he said. “What our research shows is that machine learning can retain that simplicity while adding a deeper layer of insight, helping organisations understand behavioural patterns with greater accuracy and flexibility.”

Shorter assessments could also make personality profiling easier to use in fast-moving professional environments where time is limited.

“A 10-question assessment tool that still captures the underlying personality structure would make these assessments far more practical in contexts such as recruitment, leadership development and team building,” Dr Amirhosseini said.

The study also suggests that machine learning could help move personality assessment beyond rigid categories by identifying hybrid or blended behavioural profiles that traditional scoring methods may miss.

As organisations increasingly turn to data and artificial intelligence to support decision-making, such approaches could help bring personality assessment into a more flexible and evidence-based era.

“Human personality rarely fits neatly into a single box,” Dr Amirhosseini added. “By using machine learning, we can better reflect the complexity of behaviour while still keeping the clear, practical insights that have made DISC so widely used.”

The study, Reinventing DISC personality assessment: machine learning approaches for deeper insights and greater efficiency, was published in the Journal of Artificial Intelligence & Robotics.

---

Journal

Journal of Artificial Intelligence and Robotics

DOI

10.52768/3067-7947/1037 

Method of Research

Data/statistical analysis

Subject of Research

Not applicable

Article Title

Reinventing DISC personality assessment: machine learning approaches for deeper insights and greater efficiency

 

In a breakthrough for identifying emerging online communities, Stevens researchers use machine learning and social network theory to identify formation patterns in digital forums

The team developed a novel framework that tracks digital communities over time, analyzing how online groups emerge and disappear

Stevens Institute of Technology

LinkedInWeChatBlueskyMessageWhatsAppEmail

HOBOKEN, NJ., March 26 — Researchers at Stevens Institute of Technology used machine learning tools and social network theory — the study of how people connect with each other—to better understand how people interact online. Using data from X, formerly Twitter, researchers probed the complex patterns of relationships and shared interests that link people together across the internet. In particular, they focused on elucidating how people form online communities, interact within those communities or leave them. 

“A community in this study is not merely a collection of users tweeting about similar topics but an interactional cluster,” explains Stevens Associate Professor Jose Ramirez-Marquez at the Department of Systems Engineering, who studies how communities evolve and interact. “In other words, it’s a networked structure where users are thematically aligned and actively connected through retweets, mentions or replies.”

In the past few decades, the concept of a community has transformed. Throughout history, human communities shaped based on geography, commonly influenced by access to water, fertile soil, other food sources and climate. For most of human history, communities formed based on location, evolving into villages, cities and countries. At the more local levels, communities formed through physical proximity — by people living in the same neighborhood, attending the same schools or working in the same places.  

The arrival of the internet has altered the meaning of community by removing many of the geographic and social barriers, such as for example, discrimination, that once defined it. Today, online platforms allow individuals to form communities based on shared interests, identities or beliefs rather than location. Through social media, forums and online groups, people can connect with others across the world, creating networks that exchange information, support and opinions in real time. “Essentially, the internet has transformed communities from primarily local, place-based groups into dynamic, global networks shaped by digital communication and shared interests.”

However, while social media allowed expanded opportunities for connection and collaboration, it brought new challenges—like anonymity and the ability to reach many people quickly, which can be misused. “As a member of an online community, I can disguise myself, and then I can say all sorts of things without social repercussions,” says Ramirez-Marquez. “In real life there are consequences.” Without digital communication platforms, there are also limits to how many people one can reach.

Some online communities may turn into echo chambers where people primarily interact with others who share similar viewpoints. Some groups may use social networks to spread extremist ideas or promote violence. In certain online communities, aggressive or hateful language can become normalized and encourage movements that challenge mainstream social values, amplifying false narratives and making them harder to correct. Studies have also found that increases in hateful online posts can sometimes occur shortly before rises in real-world hate crimes. 

To understand how communities interact, Ramirez-Marquez and his PhD candidate Amirhossein Dezhboro developed a framework that tracks digital communities over time and classifies the topics they discuss, allowing the researchers to examine how conversations split into smaller subtopics and how these groups emerge and disappear. The research team did this by combining an analysis of what individual users write in their posts with how those users connect with each other, by assesing network data. The team published their findings in the paper titled Community Shaping in the Digital Age: A Risk-Focused Temporal Fusion Framework for Analyzing Information Diffusion and Fragmentation in Online Social Networks, published in the journal of Risk Analysis on March 26, 2026.

The framework they developed leverages machine learning classification models to analyze user posts and interactions, revealing underlying group structures. In the study, the research team also identified several key analytical elements based on social science theories to better understand the structure and dynamics of these online communities and how real-world events influence them. 

“We found that social media interactions can create echo chambers and increase societal polarization, while the framework can help detect emerging misinformation communities and track how narratives spread over time,” Ramirez-Marquez says. 

Researchers emphasize that understanding how online communities work is important not only for scientists who study social interactions, but also for policymakers who make decisions about technology and society. “By studying how these online communities form, grow and interact, we may be able to identify early warning signs of harmful discourse,”Ramirez-Marquez says. “And that may help policymakers develop strategies to reduce potential risks while still supporting the positive aspects of online communication.”

About Stevens Institute of Technology

Stevens is a premier, private research university situated in Hoboken, New Jersey. Since our founding in 1870, technological innovation has been the hallmark of Stevens’ education and research. Within the university’s three schools and one college, more than 8,000 undergraduate and graduate students collaborate closely with faculty in an interdisciplinary, student-centric, entrepreneurial environment. Academic and research programs spanning business, computing, engineering, the arts and other disciplines actively advance the frontiers of science and leverage technology to confront our most pressing global challenges. The university continues to be consistently ranked among the nation’s leaders in career services, post-graduation salaries of alumni and return on tuition investment.

---

Journal

Risk Analysis

Article Title

Community Shaping in the Digital Age: A Risk-Focused Temporal Fusion Framework for Analyzing Information Diffusion and Fragmentation in Online Social Networks

Article Publication Date

26-Mar-2026