Artificial Intelligence

Mason Enterprise and Akiak Technology establish strategic partnership to advance innovation, entrepreneurship, and economic opportunity

Katarina Benson — Thu, 23 Jul 2026 20:24:14 +0000

Mason Enterprise and Akiak Technology establish strategic partnership to advance innovation, entrepreneurship, and economic opportunity Katarina Benson Thu, 07/23/2026 - 16:24

Mason Enterprise^TM, 91��’s hub for innovation and entrepreneurial development, established a new partnership with Akiak Technology LLC, an Alaskan tribal-owned, mission-driven technology firm dedicated to delivering impactful artificial intelligence (AI) solutions while supporting Indigenous communities. The collaboration is formalized through a Memorandum of Understanding (MOU) and reflects a shared commitment to fostering business innovation, workforce development, and mutually beneficial growth opportunities.

Mason Enterprise and Akiak Technology will collaborate on technology and business development initiatives designed to expand access to entrepreneurial resources, training, and expertise, while creating new pathways to strengthen innovation ecosystems locally and nationally.

From left, Paula Sorrell, Sharon Hamer, and Kevin Hamer. Photo provided

“This partnership with Akiak Technology aligns with Mason Enterprise’s mission to support entrepreneurs and accelerate innovation,” said Paula Sorrell, associate vice president for innovation and economic development at George Mason. “By combining Mason Enterprise’s entrepreneur support programs, tech talent, and convening power with Akiak’s technology expertise and mission-driven approach, we are adding more building blocks in the Northern Virginia innovation corridor.”

Under the MOU, Mason Enterprise will provide Akiak Technology with access to its entrepreneur programs, innovation expertise, counseling and training resources, while facilitating collaboration across the university. Together, the organizations will identify mutually beneficial opportunities and explore pathways to support Akiak Technology’s business growth in alignment with its mission, including its commitment to its tribal community.

This synergy can speed up the translation of research and discoveries into products and services, addressing complex challenges in areas such as quantum computing, AI, health care technology, and energy. Akiak Technology will also offer strategic insight and feedback to Mason Enterprise programs and help identify opportunities for revenue-generating activities that further Mason Enterprise’s mission of supporting entrepreneurs and innovation-driven organizations.

“This partnership with Mason Enterprise represents an important step forward for Akiak Technology,” said Sharon Hamer, cofounder of Akiak Technology and enrolled Akiak tribal member. “Collaborating with a leading research university gives us access to world-class talent and new opportunities to expand our AI-driven innovations, while remaining true to our mission and values.”

Akiak’s CEO Kevin Hamer added, “Mason Enterprise helped us launch our first business years ago. Deepening this relationship allows us to tap into George Mason’s expansive network and innovation programming as we continue to grow as an AI product and solutions innovator.”

The public–private partnership strengthens technology transfer, supports federal research initiatives, enhances education and workforce readiness, and contributes to regional economic growth—particularly as demand for STEM talent continues to rise.

The collaboration underscores Mason Enterprise’s role as a connector between academia, industry, and community, and highlights Akiak Technology’s commitment to collaboration, innovation, and mission-driven growth.

Learn about Artificial Intelligence at George Mason

Topics

Artificial Intelligence

NSF CAREER award allows researcher to look under AI’s hood

Nathan Kahl — Mon, 20 Jul 2026 16:30:48 +0000

NSF CAREER award allows researcher to look under AI’s hood Nathan Kahl Mon, 07/20/2026 - 12:30

In This Story

Ziyu Yao

Artificial intelligence (AI) systems are now part of everyday life, writing software, tidying up your shopping list, and answering questions. While we know they sit on top of large language models (LLMs), we still understand surprisingly little about what happens inside these systems or what is helping them generate responses.

Ziyu Yao. Photo provided

91�� Department of Computer Science Assistant Professor Ziyu Yao received a prestigious National Science Foundation (NSF) CAREER award for $674,100 over five years to peel the layers back on AI. Specifically, Yao will develop new methods for understanding the internal mechanisms of LLMs. Her research aims to make these AI systems more transparent and useful.

"We use language model–powered products every day, but we have very limited understanding of how they actually work," Yao said. Most efforts to evaluate AI systems focus on their behavior. Researchers give a model a prompt and judge the quality of its response. Yao's work instead focuses on mechanistic interpretability, an emerging field that examines what happens inside a language model as it processes information.

"If we send something to the model, we want to see how it extracts features or signals from my input and gradually performs computations before producing an answer," she said.

As language models have grown dramatically in size, that task is increasingly difficult. Today's leading models contain potentially hundreds of billions of parameters. And Yao said, "The interpretability research community has not caught up. Many traditional approaches cannot be efficiently applied to today's language models."

The project focuses specifically on AI-powered code generation, in which users describe a programming task in natural language and an AI system generates software code. Code generation is an increasingly important application of LLMs, with uses including software engineering and scientific research.

Yao's team will investigate how LLMs apply programming syntax, recall semantic knowledge, and perform deliberate reasoning while writing code. The researchers will also study how training data and learning methods influence those internal mechanisms and explore how interpretability insights can be used to improve AI coding systems.

Current interpretability studies often require researchers to manually design experiments, collecting internal model signals and analyzing the results. Yao plans to develop a framework that automates much of that process using AI itself, allowing researchers to study LLMs more efficiently.

At the same time, she wants interpretability research to produce practical benefits rather than simply satisfying scientific curiosity.

"We want to build the bridge between what we discover about language models and how that knowledge can improve them," Yao said. "Can interpretability actually help us build better AI systems?"

Although commercial AI systems such as ChatGPT are proprietary, Yao and colleagues primarily study open-weight language models that are downloadable. These models are becoming increasingly competitive with commercial systems while supporting the transparency needed for scientific research.

An expectation with the award and a hallmark of the CAREER program is a plan for broader impact. Yao hopes to expand AI literacy through outreach programs, create new curricula and open-source educational resources, provide undergraduate research opportunities, and foster collaborations between researchers in artificial intelligence and software engineering.

"I think we should encourage the younger generation to embrace AI," Yao said. "But at the same time, they should be skeptical, think critically, and be careful about what they are reading from AI."

Topics

large language models

National Science Foundation

Artificial Intelligence

College of Engineering and Computing

Research

New initiative aims to build AI literacy among educators through year-long professional learning community

Martha Bushong — Tue, 14 Jul 2026 17:55:00 +0000

New initiative aims to build AI literacy among educators through year-long professional learning community Martha Bushong Tue, 07/14/2026 - 13:55

In This Story

Amarda Shehu

91�� has launched the Ambassadors of Teaching in the Age of AI (ATAA) program, a new professional learning initiative designed to help educators develop the knowledge and skills needed to navigate the growing role of artificial intelligence in education and society.

The program officially debuted June 22–23 at George Mason's SciTech Campus during the Virginia Data Science Teacher Professional Learning Institute, which brought together more than 200 K–12 educators, higher education faculty, industry representatives, and government stakeholders. Approximately 50 participants joined ATAA as founding members of the year-long professional learning community.

Chief AI Officer Amarda Shehu at the Ambassadors of Teaching in the Age of AI kick-off event (Photo provided)

The launch marks another step in George Mason's broader AI literacy efforts, by bringing together expertise from across the university to create accessible, research-based programming. ATAA is designed to complement existing educator preparation efforts while providing a foundational AI literacy framework that can be applied by professionals across disciplines.

Developed over the past year, ATAA was created through a collaboration among Peng Warweg, research program director in the College of Engineering and Computing (CEC); Padhu Seshaiyer, professor in the College of Science and director of the Center for Mathematics and Science Education Research and Outreach; Manpriya Dua, research postdoctoral fellow in computer science; and Amarda Shehu, George Mason's chief AI officer and professor of computer science.

The team focused on a growing challenge: how to build AI fluency among practitioners, particularly the K–16 teacher workforce. Warweg, who helped lead the program's design and curriculum development, said ATAA is among the first programs to translate newly released federal guidance into a practical learning experience.

"ATAA is grounded in the U.S. Department of Labor's AI Literacy Framework—a guideline that represents a national consensus on what AI fluency should look like across professions," said Warweg. "ATAA is among the first practitioner programs in the country to translate that guidance into a structured, accessible curriculum."

Peng Warweg (on right), research program director in the College of Engineering and Computing (CEC) with Padhu Seshaiyer (left), professor in the College of Science and director of the Center for Mathematics and Science Education Research and Outreach and Manpriya Dua (center), research postdoctoral fellow in computer science. Photo provided.

The launch was supported by the Virginia Data Science Teacher Professional Learning Institute, now in its fifth year and co-organized by Seshaiyer. Through longstanding partnerships with the Virginia Department of Education and decades of leadership in STEM education, Seshaiyer helped establish the foundation that enabled the program to launch at scale.

"Virginia has been building toward this," said Seshaiyer. "ATAA is the natural next step to connect research, policy, and practice around the future of learning."

Seshaiyer and Dua also helped shape the curriculum, ensuring participants receive both practical and technically grounded instruction. Dua's expertise in artificial intelligence and computer science research helped make complex concepts accessible to educators from a wide range of backgrounds.

ATAA also builds on George Mason's broader AI strategy under Shehu's leadership. A key component of the curriculum draws from UNIV 182 AI for All: Understanding and Building Artificial Intelligence (Mason Core) a course developed, by Shehu, to help nontechnical learners gain AI literacy. Research conducted by Shehu has demonstrated significant AI literacy gains among students who complete the course.

"Foundational AI literacy cannot belong only to people who already saw themselves as technical," said Shehu. "What we learned from AI for All is that anyone can develop real fluency when the teaching meets them where they are. ATAA takes that same conviction to the practitioners who multiply it: the educators."

Designed using a train-the-trainer model, ATAA aims to extend its impact beyond the initial cohort. Participants will complete a structured year-long learning experience and earn Continuing Education Units (CEUs). After completing 15 hours of learning, participants will receive a digital badge recognizing their AI literacy achievement, with a full credentialing pathway scheduled to launch in fall 2026.

While educators are the program's first audience, ATAA envisions expanding AI literacy training to professionals across sectors, including healthcare, government, technology, and community organizations. With its strengths in AI research, governance, and education, George Mason hopes to help prepare a broader workforce for an AI-driven future.

Topics

Artificial Intelligence

Campus News

GCI-Education

Science and Technology Campus

Can AI help make medical records less biased? New study suggests yes—with caveats

Heather Carroll — Fri, 10 Jul 2026 16:50:49 +0000

Can AI help make medical records less biased? New study suggests yes—with caveats Heather Carroll Fri, 07/10/2026 - 12:50

Large language models can identify judgmental language in clinical notes but the settings play a major role in accuracy.

Teenu Xavier PhD, RN. Photo provided

“Addict,” “non-compliant,” “failed treatment,” and “obese person” are examples of stigmatizing language that can appear in medical records. At 91��’s College of Public Health, researchers are exploring whether artificial intelligence (AI) can help identify this kind of language in clinical notes before it impacts patient care.

Nurse scientist Teenu Xavier and colleagues found that large language models (LLMs) show promise in identifying stigmatizing language in clinical documentation, but their performance is highly dependent on their settings. Model size, temperature settings, prompting strategies, and even note type can substantially influence results.

One finding was consistent across every model tested: Providing examples of stigmatizing language improved accuracy.

“Simply selecting an LLM is not enough when used for clinical documentation,” said Xavier, an assistant professor in the School of Nursing. “Careful attention must be paid to settings and prompting before these tools can be reliably used in health care environments.”

Why does this matter?

The use of stigmatizing language in clinical documentation can reinforce bias and affect a patient’s future care. AI tools may be able to help identify this kind of language, promoting more equitable care and improving patient trust and experience.

“Pre-trained models, when optimized for identifying stigmatizing language, could help enable more timely interventions and modifications to the documentation process,” Xavier said. “Our research highlights the need for continued collaboration between health care professionals and AI developers to create tools that improve communication, reduce bias, and improve the overall patient experience.”

What are the detailed study findings?

The largest LLM (trained on large amounts of data) was the best at predicting “stigmatizing” language (94%), but the worst at predicting “not stigmatizing” correctly (47%).
The smallest LLM was the best at predicting “not stigmatizing” correctly (99.7%), but worst at predicting “stigmatizing” correctly (2%).
When researchers gave the LLM an example of stigmatizing language, accuracy improved in all models.
Emergency provider notes were most accurately (69%) categorized as “stigmatizing” or “non-stigmatizing,” and plan of care notes had the lowest accuracy (56%). Misclassifications most commonly arose in long, clinically dense notes where neutral descriptions of complex illness or adverse events were mistaken by the models for judgmental language.
Larger models worked best at lower temperature (how predictable or random the LLMs’s output is when making a classification) and smaller models improved with higher temperature, which means the LLM took more risks in interpretation.

“Detecting stigmatizing language with large language models: mind the settings” was published in JAMIA Open in April 2026. Co-authors include Jane M. Carrington from the University of Florida and Joshua Lambert from the University of Cincinnati.

Thumbnail photo by issaronow from Adobe Stock.

Key Takeaways

A George Mason study found that large language models show promise in identifying stigmatizing language in clinical documentation, but their performance is highly dependent on their settings, such as model size, temperature settings, prompting strategies, and note type.
The ability to detect and correct stigmatizing language early can reduce bias in patient care and lead to improved patient trust and better health outcomes.
Continued collaboration between health care workers and AI developers is needed to create tools that enhance communication, reduce bias, and improve the overall patient experience.

Learn about Artificial Intelligence at George Mason

Topics

large language models

Machine Learning in Health Care

Health Care Services

College of Public Health

Artificial Intelligence

Research

George Mason researcher develops interactive tool to measure carbon footprint of AI use in K-12 education

Katarina Benson — Fri, 10 Jul 2026 15:53:35 +0000

George Mason researcher develops interactive tool to measure carbon footprint of AI use in K-12 education Katarina Benson Fri, 07/10/2026 - 11:53

As artificial intelligence (AI) applications like ChatGPT become more commonplace in K-12 classrooms, school district leaders and policymakers will face the challenge of ensuring these tools are used in an ethical, responsible, and environmentally sustainable way. Seth B. Hunter, associate professor of education leadership and senior fellow in the EdPolicyForward center at 91��, has developed an interactive tool to help.

The tool enables school district policymakers to make better informed decisions on selecting AI software that meets the learning needs of their students and promotes environmental sustainability.

Seth Hunter. Photo by John Boal Photography

“The many conversations I’ve had with leaders across the commonwealth and nation have focused almost exclusively on the potential benefits of AI—benefits that, I should point out, we don’t yet understand well,” said Hunter. “Almost no one is talking about the costs beyond the purchase price. This tool is one way to start filling that gap—by estimating a cost that’s seemingly invisible: the additional energy and emissions that come with putting these tools in students’ hands.”

The tool enables users to input data on the total number of students enrolled in a school district, the number of students expected to use AI (based on surveys from peer-reviewed research), the anticipated number of daily queries they are likely to conduct, and the level of AI processing power needed to handle those queries. It also allows users to select their state from a dropdown menu. Next to each state is a number indicating the amount of carbon dioxide in grams released per kilowatt hour of electricity consumed.

Using these inputs, the tool calculates the increases in kilowatt hours and the carbon emissions generated by AI use. To help users better understand the significance of these estimates, the tool translates the data into every day, real-world equivalents—such as the number of miles driven with a car, the number of days a lightbulb could run continuously, or how many times a smartphone could be charged.

In addition, the tool shows what the trajectory of carbon emissions associated with the use of AI in schools will be through 2031 if no policies are put in place to reduce the associated carbon footprint.

“Even leaders with very little technical understanding of AI have many opportunities to influence its energy consumption, and it starts with a simple conversation with the vendor,” Hunter stated. “I’m not talking about choosing one brand over another, like Gemini versus ChatGPT. I'm talking about the type of model. Within a single company’s family of models, the ones built to ‘reason’ or ‘think’ through problems at length currently consume far more energy than lighter models designed for simpler tasks. The most useful question a procurement office can ask is how much a model relies on that extended ‘reasoning,’ and whether a lighter model would meet students’ needs.”

School district leaders will need to consider several factors, such as the complexity of the task the AI tool will be used to support. For example, if an AI chatbot is used to help students with multi-step math problem-solving, it will use more energy than simpler tasks such as quick factual lookups or summarizing an article. Increases in energy requirements and carbon emissions will also be influenced by the number of students accessing AI and the average number of daily queries they conduct. The type of energy source used in the school district will also have an impact. Renewable energy sources like solar, wind, or hydropower produce less carbon compared to power sources like coal.

“No one wants their students to fall behind in what can feel like an AI arms race,” said Hunter. “But I argue that policymakers and system leaders should ask how much those advantages outweigh the financial, environmental, and social costs. This tool doesn’t make that decision for anyone, but it estimates one of the costs that usually gets left out, so leaders can weigh student learning and environmental sustainability in the same conversation.”

Test out the interactive tool: "The Carbon Footprint of AI in K–12 Education"

Explore Artificial Intelligence at George Mason

Topics

Research

College of Education and Human Development

Artificial Intelligence

The AI-powered classroom: tutors, avatars, and new models for learning

Mary Cunningham — Thu, 09 Jul 2026 19:17:45 +0000

The AI-powered classroom: tutors, avatars, and new models for learning Mary Cunningham Thu, 07/09/2026 - 15:17

As health informatics expert Sanja Avramovic introduces new artificial intelligence (AI) tools into her classrooms at 91��, she's also made a point of asking students what they think, both through formal research and casual conversation.

In one early experiment, Avramovic designed AI avatars to present course material in short, on-demand videos students could watch at their own pace. Some students balked a little, finding the digital instructors monotone and robotic. They said they preferred a human.

So Avramovic asked a follow-up question: Should she remove the videos?

AI avatar created by Sanja Avramovic. Image created by AI.

"They said, 'don't you dare,'" she recalled. "'They're excellent supplemental material. Leave them. But we want you to teach.'"

That push and pull has become a central dynamic at the College of Public Health, as faculty respond and adapt to the undeniable rise of AI in higher education. The question is no longer whether AI belongs in the classroom, but how to use it in ways that actively benefit students.

“The goal isn’t AI-free learning,” Avramovic said. “It’s AI-augmented thinking.”

Learning to think with AI

Sanja Avramovic, assistant professor in the Department of Health Administration and Policy

One challenge is persistent: How do you keep students learning when AI can provide answers in seconds?

Research by Avramovic, assistant professor in the Department of Health Administration and Policy Abdul Hafeez, and colleagues found that students can become overly dependent on AI-generated answers, reducing their engagement, interaction, and critical thinking.

Rather than prohibiting AI use, faculty developed and tested AI tutors that function more like coaches than answer generators, walking students through assignments one step at a time. In order to build them, instructors map out common mistakes and teach the systems how to respond in ways that promote deeper learning. “It interviews students,” Hafeez said. “It compels them to think critically.”

Early findings suggest this approach can help counter passive AI use, with students showing stronger performance and engagement than their peers. In one analysis, undergraduates using the tutors improved performance by nearly 9 percentage points.

Tempering the trust

Along with the enthusiasm for AI’s potential, faculty are just as focused on teaching students where the technology falls short.

Vanessa Chee, an assistant professor in the Department of Global and Community Health, warns students against relying on it too early in the learning process, likening it to “using roller blades when you’re crawling.”

In one global health course, Chee requires students to develop a 2,500-word narrative and a presentation proposing a public health intervention entirely on their own. Afterward, they use AI to generate a version of the same project, and compare the two.

That exercise is often revealing. Many students walk away disappointed by the quality of the generated work, but with a sharper understanding of what their own thinking contributes, Chee said.

Katherine Scafide, an associate professor in the School of Nursing, introduces AI early in the research process, using it to help students scan the literature and identify possible gaps worth investigating. But she draws a line between assistance and analysis. Students are expected to verify sources, refine their own research questions, and critically test whether AI summaries actually hold up.

"I encourage AI to assist in some structural aspects of writing, like outlining and grammar, but not to author the work," she said. " AI can be a writing tool, but it cannot be an author."

For Avramovic, the biggest concern is that students will trust AI too quickly. She recalled one student who relied heavily on AI while studying, learned flawed logic, and then repeated it on an exam.

“Don’t use AI until you know enough to correct it,” she tells her students.

Rebuilding the classroom

As AI changes how students learn, faculty are also rethinking how they create and deliver their own course content.

With one undergraduate health informatics class, Avramovic and colleagues realized that much of the content felt disconnected from the clinical realities that nursing students—who are required to take the course—would eventually face. So, the instructors used AI tools to rewrite assignments around nursing-specific scenarios, like identifying problems in electronic health records. They also introduced AI-generated videos narrated by digital avatars, guiding students through technical material.

The updates made a difference. Nursing students in the AI-redesigned course showed statistically significant learning gains, more than half said AI-generated videos were more helpful than assigned readings, and nearly four out of five reported that the tools improved their overall learning experience.

Farrokh Alemi, a professor in the Department of Health Administration and Policy, took a similar approach in his graduate statistics course, using AI to help deliver lectures and tutor students.

For students, the change meant more access to their instructor. “Because the AI handled a lot of mechanical tasks, I used the extra time to have more interactions with students, give them career advice, call up at-risk students, just be more available,” he said.

Evolving the toolkit

Faculty at the College of Public Health compare the current moment to earlier technological shifts in education, from the arrival of calculators in math classrooms to the rise of the internet and online learning.

Students still need to learn the fundamentals, faculty said, but once they understand the underlying concepts, AI can help them explore more complex problems and apply knowledge in new ways.

“You still need the human in the loop,” Avramovic said.

The technology continues to evolve quickly. But faculty say the larger goal remains to help students build the skills they’ll need in a world where AI is already becoming part of the job.

Key Takeaways

George Mason College of Public Health faculty are using artificial intelligence tools including AI tutors and digital avatars to enhance teaching and student learning.
Instead of prohibiting tools like ChatGPT, instructors are teaching students to think critically, verify AI outputs, and use AI responsibly.
AI is helping redesign courses, personalize learning, and prepare College of Public Health students for an AI-integrated workforce.

In This Story

Sanja Avramovic, PhD

Katherine Scafide, PhD, RN, FAAN, FAAFS

Topics

Artificial Intelligence

College of Public Health

innovative classes

Students

GCI-Education

George Mason researcher is using CAREER award to build brain-inspired AI that is more adaptive

Nathan Kahl — Tue, 07 Jul 2026 18:49:00 +0000

George Mason researcher is using CAREER award to build brain-inspired AI that is more adaptive Nathan Kahl Tue, 07/07/2026 - 14:49

In This Story

Maryam Parsa

Though artificial intelligence (AI) is making extraordinary advances, today's AI systems still have significant limitations. They often require enormous computing power and energy (not to mention well-documented water usage), struggle to adapt when conditions change, and can make decisions in ways that are difficult for humans to interpret, validate, or trust.

Maryam Parsa. Photo provided

Maryam Parsa, assistant professor in 91��'s Department of Electrical and Computer Engineering, believes those shortcomings point toward a compelling source of inspiration: the human brain.

Parsa received a five-year, $655,000 National Science Foundation Early Career Development (CAREER) Award for developing a new framework for brain-inspired AI. The project seeks to make AI systems more efficient, adaptable, resilient, and transparent by borrowing principles from biological nervous systems.

"AI was originally inspired by the brain," Parsa said. "Concepts such as neurons, synapses, and neural networks drew inspiration from how brains process information. But over time, AI has moved farther away from those biological principles.”

Parsa believes researchers now have an opportunity to reconnect AI and neuroscience. “Neuroscience advanced tremendously over the past few decades, with only a small fraction of those discoveries finding their way into AI systems.”

As computing power grew, researchers found AI systems could make major gains by scaling neural networks and training them on massive datasets. The approach, fueled by increasingly powerful computers, is leading to significant breakthroughs, particularly with the rise of large language models (LLMs).

"They are producing remarkable devices," she said. "The progress we see in LLMs and modern AI grew out of relatively simple neural-network ideas. Over time, researchers added more layers, more data, more computational power, and more sophisticated training methods, leading to powerful systems.”

The goal is not to replace today's successful AI systems with direct copies of the brain. Instead, her research focuses on areas where current AI falls short. "We know some of the key challenges AI still faces," she said. "It can be difficult to understand how these systems reach their decisions, and they do not always transfer well when conditions, data, or environments change. That is where biological intelligence can offer useful design principles.”

When today’s AI systems encounter unfamiliar environments or changing conditions, they often need to be retrained or fine-tuned. Biological intelligence, however, is remarkably adaptable and energy efficient, offering clues to how future AI systems could learn and respond more flexibly.

Her CAREER project builds a unified framework around three characteristics of biological intelligence. The first pillar, dimensionality, examines how the brain creates meaningful representations of the world rather than simply processing raw information. The second, heterogeneity, explores the brain's remarkable diversity. Rather than relying on identical neurons performing tasks, biological brains contain many neuron types with distinct behaviors. "The brain is not copy, paste, copy, paste, copy, paste,” she said.

The third pillar, nonlinearity, draws inspiration from the retina, which performs extensive processing before visual information reaches the brain. "It's not like the raw pixels go directly to our central brain," Parsa said. "The retina digests and processes the information first."

Parsa said the broader idea is combining these principles into a more biologically inspired approach to AI. “If I know how to process my input, create a better representation of my environment, and leverage this massive heterogeneity, then I can move AI a little closer to the way biological intelligence works,” she said.

Beyond advancing AI research, the project includes an educational effort to help students see how ideas from the brain can shape the future of computing and artificial intelligence. Parsa plans to bring these concepts into graduate courses, hands-on hackathons, and STEM outreach activities for K-12 students, creating new pathways for students to explore the intersection of neuroscience, engineering, and AI.

Topics

large language models

Electrical and Computer Engineering

Electrical and Computer Engineering Faculty

National Science Foundation

Research

Artificial Intelligence

C-TASC

Cora Sula awarded Fulbright to study AI in classrooms in Estonia

Martha Bushong — Wed, 17 Jun 2026 16:31:39 +0000

Cora Sula awarded Fulbright to study AI in classrooms in Estonia Martha Bushong Wed, 06/17/2026 - 12:31

Cora Sula, a third-year PhD student in the Department of Information Sciences and Technology at 91��, has been awarded a prestigious Fulbright scholarship to study artificial intelligence (AI) in classrooms in Estonia this fall.

Sula works with Assistant Professor Nora McDonald at the intersection of digital agency, AI and identity. Her recent work examines healthcare privacy and the impact of AI chatbot use on teens’ self-concept and well-being—research she presented at CHI 2026 in Barcelona, the world’s leading conference in human-computer interaction.

“I am so thrilled at this opportunity to study the impact of AI on classrooms,” Sula said. “Estonia is one of the few countries with a national rollout of AI-assisted classrooms.”

While AI education initiatives in the United States are expanding, they remain fragmented and locally driven. In contrast, Estonia is pioneering a nationally coordinated approach through its AI Leap program, an initiative led by the Ministry of Education and Research in partnership with OpenAI. The program aims to systematically integrate AI-powered tools into high school teaching and learning.

Sula’s Fulbright project will investigate how this national initiative is shaping teachers’ professional practices and classroom pedagogy. Using a mixed-methods design—including ethnographic observation, semi-structured interviews, and pre/post surveys—her research will explore how educators navigate both the opportunities and challenges of AI integration.

Her Fulbright will be hosted at Tallinn University of Technology, where she will collaborate with Janika Leoste head of the Creativity Matters Lab and local scholars who bring expertise in digital education and policy.

“This is an amazing opportunity to systematically study how AI will transform classrooms for generations to come,” Sula said. “I’m excited to bring what I learn back to my work on AI and teens.”

In This Story

Nora McDonald

Explore Artificial Intelligence at George Mason

Topics

Artificial Intelligence

Machine learning teaches asset traders not to sweat the small stuff

Katelynn C Hipolito — Mon, 15 Jun 2026 14:40:20 +0000

Machine learning teaches asset traders not to sweat the small stuff Katelynn C Hipolito Mon, 06/15/2026 - 10:40

Financial markets are governed by a combination of rational and irrational forces, statistical probabilities, and “animal spirits.” It takes fluency in both to understand the market, let alone to beat it. Yet market actors, including asset traders, now frequently use machine-learning techniques to help generate predictions of future asset prices.

Bo Hu, assistant professor of finance at the Costello College of Business at 91��. Photo provided by Bo Hu.

Scholars such as Bo Hu, assistant professor of finance at the Costello College of Business at 91��, are researching how these machine-learning tools are changing the decision-making processes that move the market, for better or worse.

The subject of Hu’s recent paper in Management Science is a well-known machine-learning technique called LASSO (least absolute shrinkage and selection operator), which has been widely adopted by financial practitioners since its introduction in 1996 by statistician Robert Tibshirani.

“If you look at the original paper, it describes an approach created by adding a regularization penalty to the least-squares regression method,” Hu says. Translation: “The power of LASSO is that it can screen out (i.e., penalize) weak signals while capturing stronger, potentially profitable ones. A LASSO-type trading strategy involves an ‘inactive zone’ for smaller-scale activity, in which the trading strategy is to do nothing.”

The paper was co-authored by Wen Chen of Texas Tech University and Liyan Yang of the University of Toronto.

Despite LASSO’s popularity and power, the soundness of its economic rationale remains unclear. Traders are presumably seeking any edge, however small, in the pursuit of outsized returns. How could it make sense for them to adopt a system designed to relegate signals of lesser magnitude to an ignored “inactive zone”?

To resolve this question, the researchers developed a theoretical framework to model a financial market in which multiple agents (read: traders) use an asset’s price history to forecast its return and make trading decisions.

In the benchmark case, when traders know the trading environment and do not face model uncertainty, they act according to an alternative to LASSO known as MSE (mean squared error). “MSE is essentially a Bayesian learning approach grounded in economic rationality,” Hu says. “It means that rational agents use Bayesian learning to update their beliefs and design their trading strategies. That stands in stark contrast to LASSO estimation, which filters out weak signals.”

However, the researchers found that when traders faced substantial ambiguity about the distribution of asset values, the trading calculus shifted. Ambiguity-averse agents will adopt a robust, LASSO-like strategy, refraining from trading in response to weak or intermediate market signals. With linear constraints imposed on the allowable trading strategies, the equilibrium decisions exactly matched LASSO estimates.

As an equilibrium trading strategy, LASSO can improve aggregate profits relative to the edge-seeking Bayesian alternative in the benchmark case, because the more conservative positions dictated by the “inactive zone” soften competition among traders. In a large market, aggressive competition drives the aggregate profits of traders using the conventional MSE strategy toward zero. By trading less aggressively, LASSO traders may preserve positive aggregate profits—a mechanism the researchers describe as “implicit collusion,” even though the traders do not communicate or explicitly coordinate.

However, Hu underscores that LASSO’s usefulness as a hedge against ambiguity depends on how well traders’ beliefs match the market’s true distribution of risk. The profitability of a LASSO strategy therefore hangs in the balance between traders’ biases and their enhanced market power (due to LASSO’s conservatism).

This balance is especially important when traders must distinguish between temporary fads and persistent trends. “Recently, the semiconductor index had a record-breaking run of consecutive gains. That’s very strong momentum, but if you are a contrarian, you might want to bet against that trend,” Hu says. “A trader must decide whether to follow the trend or take a contrarian position. LASSO’s inactive zone can help prevent overreaction to weak evidence, but it may also delay action when an emerging trend is genuine.”

There is also the possibility that a LASSO strategy could increase market volatility when combined with the other objectives and constraints of market makers, including high-frequency traders. “These traders need to control their inventory,” Hu says. “If they follow a LASSO-type strategy with an inactive zone, they will accommodate the liquidity demands of the market until their inventory reaches a certain threshold. At that point, they may start trading like momentum traders and cause the liquidity dry-up in financial markets. This dynamic is part of what fuels incidents like the ‘flash crash’ in 2010.”

In This Story

Bo Hu

Topics

Costello College of Business News

Costello College of Business Faculty Research

A.I. and Innovation - Costello

Costello Research Machine Learning

Costello Research Fintech

Finance - Costello

Costello Research Financial Crises

Costello Research Risk Management

Research

Artificial Intelligence

NSF CAREER award will support teen autonomy in age of AI

Nathan Kahl — Tue, 09 Jun 2026 15:45:38 +0000

NSF CAREER award will support teen autonomy in age of AI Nathan Kahl Tue, 06/09/2026 - 11:45

Adolescence has always been a time of figuring out who you are. Increasingly, that process unfolds alongside social media feeds, recommendation algorithms, and AI chatbots. 91��’s Nora McDonald received a prestigious National Science Foundation (NSF) CAREER award for $748,000 to better understand how adolescents experience and are shaped by these artificial intelligence (AI) and other personalized digital environments, and how they can develop the skills and capacities necessary to maintain their sense of self, autonomy, and agency within them.

McDonald, an assistant professor in the Department of Information Sciences and Technology, has spent several years studying a growing reality of adolescent life, and one increasingly shaped by personalized AI technologies. Her research has found that social media recommendation systems and AI chatbots can serve as spaces where young people reflect on themselves and seek feedback about who they are.

McDonald's work will give teens tools to properly engage with digital technologies. iStock photo.

“Much of the public conversation has been about how algorithms influence what kids see and how we go down rabbit holes,” McDonald said. “I’m interested in a different question: How do algorithms shape self-understanding and how do teens make sense of themselves in environments that are increasingly personalized and shaped by algorithms and AI?”

McDonald’s previous research suggests that even while teenagers often view AI chatbots and recommendation systems as practical tools, those systems can become important spaces for self-exploration and self-presentation.

“Over time, teens might come to expect these systems to reflect aspects of who they are,” McDonald said. “They use them not just for help with homework or entertainment, but also to explore and understand themselves.”

Those findings raise broader questions about adolescent development. If recommendation systems continually influence what young people see, what feedback they receive, and what content captures their attention, how will that affect how they see themselves?

Nora McDonalds. Photo provided

McDonalds's CAREER project will investigate that question through what she calls a resilience framework that examines how adolescents recognize, interpret, and respond to algorithmic influence and how it relates to the coherence of their sense of self.

The research combines interviews, diary studies, surveys, and participatory design activities to better understand how adolescents interpret and respond to algorithmic systems. McDonald will recruit teenagers from across the country, beginning with qualitative studies and eventually expanding to a national survey designed to generate broader statistical insights.

One major outcome of the project will be the development of practical resources and activities designed with teenagers to help them navigate personalized digital environments.

The effort differs from traditional digital literacy initiatives that focus primarily on explaining how algorithms work. “It helps them reflect on algorithmic influence and develop strategies that are really rooted in managing attention and maintaining perspective on who they are outside of digital experiences,” McDonald said.

The CAREER award also supports substantial educational activities. McDonald plans to develop a new undergraduate course focused on algorithmic self-defense from a human-computer interaction perspective. The project will also support a youth summer institute for high school students helping young people to critically engage with AI personalized digital technologies that increasingly shape their everyday experiences.

Learn about Artificial Intelligence at George Mason

In This Story

Nora McDonald

Topics

information sciences and technology

social media intervention

Teens

Artificial Intelligence

Research

National Science Foundation

Artificial Intelligence

Mason Enterprise and Akiak Technology establish strategic partnership to advance innovation, entrepreneurship, and economic opportunity

Related Stories

Topics

NSF CAREER award allows researcher to look under AI’s hood

In This Story

Topics

New initiative aims to build AI literacy among educators through year-long professional learning community

In This Story

Topics

Can AI help make medical records less biased? New study suggests yes—with caveats

Why does this matter?

What are the detailed study findings?

Key Takeaways

Related Stories

Topics

George Mason researcher develops interactive tool to measure carbon footprint of AI use in K-12 education

Related News

Topics

The AI-powered classroom: tutors, avatars, and new models for learning

Learning to think with AI

Tempering the trust

Rebuilding the classroom

Evolving the toolkit

Key Takeaways

Related Stories

In This Story

Topics

George Mason researcher is using CAREER award to build brain-inspired AI that is more adaptive

In This Story

Topics

Cora Sula awarded Fulbright to study AI in classrooms in Estonia

In This Story

Related Stories

Topics

Machine learning teaches asset traders not to sweat the small stuff

In This Story

Topics

NSF CAREER award will support teen autonomy in age of AI

Related Stories

In This Story

Topics