News Contact

As more legal documents are drafted with large language models (LLMs), falsified citations are becoming a problem for the people who rely on them. Georgia Tech researchers have created a tool to help lawyers verify their work.  

Through a donation from IBM, Georgia Tech’s Center for Scientific Software Engineering (CSSE) and Open Source Program Office (OSPO) held a fellowship competition for students to work with mentors at CSSE and OSPO to create the tool. Master’s students Adrian Lozada and Di Wu were selected as fellows to work on the project.  

The Mellea Legal Requirements Checker (Mellea-LRC) can quickly identify and accurately verify case law citations, specifically checking for citations that are hallucinated by artificial intelligence (AI). One fellowship goal was to give lawyers and others in the legal field a tool to easily verify legal citations. Mella-LRC does this by using both LLMs and IBM’s Mellea open-source framework to verify and repair LLM prompts.  

The team tested the tool on 26 public court-filing documents containing falsified citations. Mellea-LRC can process a document in about eight minutes. The tool first extracts the case law citations from the legal document. It then verifies that each citation is authentic and supports the document's content.  

“IBM’s Mellea framework was the foundation we built it on. Mellea’s job is to make model calls predictable: you write the rules the output must satisfy as ordinary code, and the framework retries until LLM output meets them. Our job was to leverage the framework to build the legal application from end to end,” Wu said.  

An important aspect of the project was ensuring the tool’s code was open source. The team supported this goal by also creating and releasing a new test and validation dataset on Hugging Face – a popular open-source AI development community – for future benchmarking studies. 

“Agentic AI is growing quickly, and much of the growth happens in open-source projects, where anyone can read, use, and improve the work,” Lozada said.  

OSPO Director and Principal Research Scientist Jeff Young noted that IBM’s Mellea framework was critical to this work in that its “instruct, verify, repair” paradigm allows for the creation of reliable LLM loops as well as the use of small language models (SLMs), such as IBM’s Granite, which can be run on a laptop. 

“This project demonstrates the power of open-source models like IBM Granite as well as the importance of using novel software engineering and algorithmic techniques from the Mellea project to improve the reliability of our existing LLMs and SLMs.  

“Mellea-LRC provides a compelling example of AI’s potential to evolve from a source of error in the legal profession into a trusted ally by helping lawyers work faster and cite precedents more accurately,” Young said.  

News Contact

Imagine a vehicle powered by artificial intelligence (AI) whose object detection system has been secretly modified to misidentify stop signs whenever a particular visual pattern appears in its camera feed.

Unlike traditional AI backdoors, which manipulate a model's learned parameters, architectural backdoors hide malicious instructions directly in the code that defines how the model operates.

However, everyday image-processing operations, such as cropping, adjusting contrast, or adding noise, can change a trigger pattern and prevent an attack from working, according to new research from Georgia Tech.

The findings challenge assumptions about how practical these attacks are. Although previous research showed that architectural backdoors could work under controlled conditions, the Georgia Tech team found that existing approaches struggle to remain effective and difficult to detect when AI systems process images under realistic conditions.

“The biggest misconception is that demonstrating an architectural backdoor under controlled conditions means it will actually work in the real world,” said David Oygenblik, Ph.D. student and lead author of the study.

The study details the development of FAULT, a framework for evaluating architectural backdoors under realistic conditions.

The findings reveal a tradeoff. Attackers can make a backdoor more sensitive, so it recognizes its trigger after an image has been changed. But that also increases the chance that ordinary images will accidentally activate the backdoor, potentially hurting the AI system’s performance.

“For the architectural backdoors we study, this tradeoff is fundamental,” Oygenblik said.

The team found that this problem persisted across five methods of implementing architectural backdoors. Even when attackers knew how the system would process images, they struggled to make their attacks both reliable and difficult to detect.

The research did not test attacks against real vehicles or other production systems, nor did it demonstrate outcomes such as crashes or ransomware. Instead, it examined whether existing architectural backdoors could reliably activate under different image-processing conditions.

However, the findings highlight a potential risk in the AI software supply chain. If an attacker inserted malicious code into an AI model used in a real system, the backdoor could potentially cause incorrect predictions when its trigger was detected. The consequences would depend on how the system used the model’s output.

Because architectural backdoors hide in code, examining a model’s learned parameters or retraining it may not be enough to uncover or remove the threat.

Oygenblik said developers should test AI systems under realistic conditions, watch for unexpected drops in accuracy, and examine model code for suspicious trigger-detection logic.

Finding FAULTs in Architectural Backdoors: Why Trigger Detection Fails Under Real-World Conditions, was accepted to the ACM Conference on Computer and Communications Security. The authors will present the paper in The Hague in November.

Oygenblik conducted the research with undergraduate students Teja Akella and Tanmay Gupta, Ph.D. student Yizhi Huang, and Associate Professor Brendan Saltaformaggio of Georgia Tech’s Cyber Forensics Innovation Laboratory. The lab is a collaboration between the School of Cybersecurity and Privacy and the School of Electrical and Computer Engineering. 

News Contact

ATLANTA — Machine learning PhD students from across Georgia Tech showed off their work on Friday, October 16, 2026, at the ML@GT Student Conference. The Machine Learning Center (ML@GT) hosted the event, which brought together faculty and students from labs and disciplines across campus.

The conference had three goals: to showcase research by ML@GT's PhD students, to show students what's happening in other labs and fields, and to build more collaboration across the Center.

The program included 12 student presentations across four research themes. More than 20 posters were on display during the poster session. Topics ranged from efficient prediction and protein modeling to brain-computer interfaces, seizure detection, and acoustic perception.

Attendees, including Georgia Tech faculty and students, voted for their favorite presentations and posters. Prizes went to the top three in each category.

Presentation Awards

• 1st Place: Hangyu Zhou, "Learning Computational Structure for Sample-Efficient Prediction" - School of Computer Science, advised by Steve Mussmann
• 2nd Place: Darin Tsui, "Protein Circuit Tracing via Cross-layer Transcoders" - School of Electrical and Computer Engineering, advised by Amirali Aghazadeh
• 3rd Place: Pranav Deevi, "FETTUCCINE: Fast and efficient brain-to-text decoding on mobile devices" - School of Biomedical Engineering, advised by Chethan Pandarinath

Poster Awards

• 1st Place: Hyunju Kim, "Spatiotemporal Hyperedges for EEG Seizure Detection and Prediction" - School of Computational Science and Engineering, advised by Nabil Imam
• 2nd Place: Benjamin Burns, "When, why, and how do diffusion posterior samplers fail? A finite-sample lens" - School of Electiral and Computer Engineering, advised by Sara Fridovich-Keil
• 3rd Place: Tianwei Jiang, "Intelligent Acoustic Perception: From Sound Recognition to Multimodal Spatial Understanding" - School of Electrical and Computer Engineering, advised by David Anderson

The winning projects show the range of machine learning research at Georgia Tech. They include basic work on sample efficiency and generative models as well as applications in neuroscience, healthcare, biology, and audio.

The ML@GT Student Conference gives students a chance to share their research and make connections across the Center. ML@GT congratulates all of this year's presenters and award winners.

This article was drafted with the help of Microsoft Copilot.

As doctors increasingly rely on computational approaches to examine and treat patients, Georgia Tech is building its expertise at the intersection of healthcare, computer vision, and artificial intelligence (AI).

Jiayun (Peter) Wang began his tenure-track career on Oct. 15 as an assistant professor in the School of Computational Science and Engineering (CSE).

At Georgia Tech, Wang will develop new AI tools to help doctors make better use of complex medical data. His research aims to transform information from medical images, clinical measurements, and patient records into clearer insights that improve diagnosis, treatment, and patient care.

“I’m excited to build a lab at Georgia Tech, and to work with students and collaborators on new ideas,” Wang said. “I hope our research can connect fundamental advances in AI with real challenges in healthcare and science.”

Before starting at Georgia Tech, Wang conducted postdoctoral research at Caltech and worked as a senior research scientist at Accenture. He earned his Ph.D. in 2023 from the University of California, Berkeley.

Much of Wang’s healthcare research builds on computer vision, which enables machines to interpret images and videos.

For example, lung ultrasound is a fast, portable, and radiation-free way to check breathing problems. However, a drawback is that images can be hard to read.

Wang led the development of LUNA, an AI method that converts raw ultrasound signals into maps showing how air is distributed in the lungs. This approach makes bedside lung exams more consistent and easier to interpret.

Wang has also used AI to improve another common type of medical imaging: MRI.

MRI scans can take time to complete, and analyzing the results can take even longer. Patients may find it uncomfortable to lie still in a scanner and waiting for results can add stress during an already difficult time. Long scans and reviews can also make it harder for doctors to schedule procedures and consult specialists.

To overcome this shortfall, Wang worked on a team that built a single, flexible AI model that reconstructs high-quality MRI images from fewer measurements across different scanning settings. Along with enabling faster scans, the approach preserves the details doctors need to make informed decisions. 

Surgeon training is another way Wang has helped doctors using computer vision techniques.

During robot-assisted surgery, trainees receive guidance from experienced surgeons. However, it is difficult to know what advice improves techniques. A surgical-training study that Wang co-authored combined video and spoken feedback to assess trainee learning.

Wang has used less serious clinical conditions to improve quality of life through AI. Dry eye is one of the most common reasons people visit eye clinics, yet the condition is difficult to assess. 

Working with collaborators at UC Berkeley, Wang studied AI methods to analyze patient information alongside images of tiny glands in the eyelids. This work contributed to AI tools that assist in diagnosing ocular surface diseases and predicting dry eye onset based on lifestyle factors.

Beyond healthcare, Wang also works to make AI easier to train and use. His research helps models learn from images without extensive human labeling and build strong systems that require less computing power.

“Medical measurements are often noisy, incomplete, or difficult to interpret. Yet they can carry information that matters for doctors and patients,” said Wang. 

“I’m motivated by the opportunity to combine what we know about the physical world with new learning methods to build AI systems that are more reliable and useful.”

Looking ahead, Wang is building the SWI Lab (science, world models, and intelligence) at Georgia Tech. 

The lab develops AI “world models” that learn how physical and biological systems behave, predict how they change, and help researchers draw insights from incomplete measurements. This work connects fundamental AI advances with challenges in medicine and science.

Ph.D. students Xinsong Feng and Wentao Zhou joined SWI Lab this fall as its inaugural members. The lab welcomes prospective students, research interns, and visitors interested in world models and Wang’s work.

“The opportunity to work across computing, engineering, and science attracted me most to Georgia Tech,” Wang said. 

“The most interesting problems often require both new AI ideas and deep knowledge of their applications. I look forward to studying those problems through new relationships and collaborations.”

News Contact

Public health systems often face a difficult challenge when time, funding, and resources are limited: who should receive care first? These decisions are complicated further by differences in patient needs, disease progression, and social factors like access to care.

A team of computer scientists and doctors led by Georgia Tech aims to change that. Using a system of artificial intelligence (AI) models, the team can help clinicians deliver the right intervention to the right patient at the right time.

Their system uses Restless Multi-Armed Bandit (RMAB) models. While the name is technical, the idea is simple. The RMAB framework helps decision-makers choose the best actions over time when conditions are complex and constantly changing.

This project will tackle Type 2 Diabetes (T2D), building on previous experience expanding RMAB models to support maternal health care in India. Success on this front could lead to treating other chronic illnesses, like hypertension, heart disease, kidney disease, and mental health.

Recognizing the RMAB system’s potential, the National Institutes of Health (NIH)  awarded the group an R01 grant. The award will fund the project with $1.2 million through 2029.

“Our goal is to address the immense challenge of allocating limited health intervention resources in optimal ways,” said Kai Wang, an assistant professor in the School of Computational Science and Engineering (CSE). 

“By leveraging rich contextual data and accounting for social determinants, we envision RMAB models that empower health centers to customize interventions to individual patient profiles. Ultimately, this advances precision medicine and ensures equitable healthcare access for diverse populations.”

T2D is a diabetes condition where the body resists insulin or doesn’t produce enough to regulate blood sugar levels. Left unmanaged, high blood sugar leads to severe health problems like heart disease and stroke, damage to organs like eyes and kidneys, nerve damage, infections, and ulcers.

T2D is challenging to treat because patients have different ages, genetics, lifestyles, and body composition, all of which affect the progress of the disease. Social factors, like transportation and family support, matter just as much as biology to treat the disease.

For these reasons, patients who get the same medications may get very different results. While treatment plans now incorporate new GLP-1 drugs, they remain expensive and difficult to distribute equitably.

“Diabetes is a great example of a complicated, multi-faceted problem. It is a natural playground for the kind of high-dimensional machine learning algorithms we want to build and deploy,” said School of CSE Professor B. Aditya Prakash.

“As a chronic disease, diabetes also creates opportunities for sequential decision-making. Providers are not making just one decision to treat the condition, but rather a series of decisions over time.”

True to its multi-armed name, the team’s framework applies RMAB models across three different thrusts. 

Contextual RMABs use patient histories, biomarkers, demographics, and other data to predict how each individual will respond to an intervention. These models directly provide personalized, data-driven treatments. 

By incorporating patient-specific information, contextual RMAB models can help healthcare providers prioritize the most timely interventions. The goal is to improve patient outcomes while making more efficient use of limited healthcare resources.

Next, the group wants to understand social life and community influence in public health. Families and social networks shape how people manage chronic health conditions.

Network RMABs study and account for the role of social factors in health interventions. For example, the models can recognize when one patient responds better to treatment than others through benefits of their social circle. 

More broadly, hospitals, clinics, government agencies, and nonprofit organizations often share responsibility for the same population in an area. Sometimes, they even compete for the same medications or funding. 

Multi-agent RMABs are the third glass of models aimed at decentralized health systems to make fairer decisions. These tools examine how to coordinate resources and decisions between stakeholders. This in turn leads to policies that encourage cooperation and equitable resource distribution.

“I feel extremely confident in RMABs because we have already seen these frameworks deliver real-world impact,” Wang said. 

“Prior success has proven the value of data-driven decision-making under resource constraints, giving us a strong blueprint to effectively address the nuances and complexities of diabetes.”

Wang previously built RMAB models to expand maternal healthcare for women in India. Working in conjunction with the non-profit organization ARMMAN, the models used healthcare information to identify and reach the most vulnerable populations. 

By integrating RMAB models into ARMMAN mobile health technology, the partnership has reached over 350,000 mothers. This further led to a 31% improvement in engagement with health information.

Compared to maternal healthcare, diabetes is a more nuanced disease with differing causes, symptoms, and treatments. This requires a team effort spanning institutions and backgrounds.  

Dr. Sara Cromer is an endocrinologist at Massachusetts General Hospital (MGH). Experienced diagnosing and treating diabetes, she has studied how social issues and equity influences the disease. MGH will provide much of the patient data to train the RMAB models.

Associate Professor Jian Li studies reinforcement learning at Stony Brook University. His expertise in applying reinforcement learning for resource allocation plays a key part tailoring models for the T2D problem. Wang and Li have previously worked together on RMAB research.

Prakash is an expert in computational epidemiology at Georgia Tech. He has led substantial contributions to the use of AI for networks to predict outbreaks and track disease spread. This experience has led to collaborations with NIH, the Centers for Disease Control and Prevention (CDC), National Science Foundation, and others.

“This project speaks to the interdisciplinary problem-solving we try to achieve in CSE. Despite having different backgrounds, our interest in applying our strengths to real-world healthcare problems unites us together,” Prakash said.

“These problems require experts from all sides. Without computational scientists, we can’t take advantage of available data. We lean on clinical experts to provide the data and expertise needed to guide our methods and make sense of results. Ultimately, these approaches need to be clinically validated if we want to make real-world progress.”

News Contact

Panama has launched its National Artificial Intelligence Strategy, a 10-year road map designed to strengthen the country’s position in artificial intelligence (AI) and advance its broader digital transformation.

The strategy, announced in July 2026 as part of Panama’s National Agenda on Advanced Technologies, was developed through a partnership between Panama’s National Secretariat of Science, Technology, and Innovation (SENACYT) and Tech AI, Georgia Tech’s AI solutions developer. The effort brought together leaders from government, academia, industry, and civil society.

The strategy identifies priorities for AI across economic development, workforce preparation, digital infrastructure, governance, research, and international collaboration.

Pascal Van Hentenryck, the A. Russell Chandler III Chair and professor at Georgia Tech and director of the National Science Foundation (NSF) Artificial Intelligence Institute for Advances in Optimization (NSF AI4OPT), helped lead the development of the strategy.

“Panama has a unique opportunity to leverage its position as a global logistics and financial hub to become a trusted leader in AI,” Van Hentenryck said. “The strategy lays out what is needed to develop the talent, infrastructure, data, and institutions to support that ambition.”

Panama’s focus on artificial intelligence comes as the country expands its role in Latin America’s technology ecosystem. The 2025 Latin American Artificial Intelligence Index classified Panama as an “Adopter” nation and reported a 1.47-point increase in the country’s score from the previous edition.

Building the strategy

Van Hentenryck and Timothy M. Brown, interim managing director of Tech AI at Georgia Tech, authored the strategy following months of engagement with stakeholders throughout Panama.

The process included workshops, interviews, and a nationwide survey that gathered input from more than 6,700 Panamanians. Government officials, researchers, industry representatives, educators, and community members contributed perspectives on the country’s opportunities and challenges.

“The strategy was shaped by what we heard from people across Panama,” Brown said. “That input helped identify where AI can have the greatest impact and what Panama needs to build the capacity to use it effectively.”

The resulting framework is organized around five foundational pillars:

1. Governance and regulatory framework
2. Human talent and institutional capacity
3. Data and infrastructure
4. Ethics, inclusion, and sustainability
5. International cooperation and global standards

Together, the priorities are intended to support AI adoption across sectors including logistics, financial services, healthcare, agriculture, education, manufacturing, government services, and energy.

At the center of the strategy is a vision for Panama to become “Latin America’s Trusted Hub for AI Innovation.” The plan builds on the country’s established role in global logistics and finance, envisioning a “Digital Canal” that supports the secure movement of data, technology, talent, and innovation throughout the region.

Eduardo Ortega-Barría, national secretary of science, technology, and innovation at SENACYT, said the strategy provides a foundation for Panama to participate more fully in the global AI economy.

“AI is transforming economies, societies, and the way we work,” Ortega-Barría said. “For Panama, this strategy is about preparing our country to participate in that transformation while ensuring that innovation creates opportunities for our people and strengthens our competitiveness.”

Connecting research and policy

The strategy reflects contributions from Georgia Tech researchers and experts, SENACYT leadership, policymakers, international advisers, and private-sector stakeholders. AMD also participated in the initiative.

Keith Strier, senior vice president of Global AI Markets at AMD, said international partnerships will be important as Panama develops its technology ecosystem.

“Panama is building its domestic AI capabilities and infrastructure together with the United States as a strategic partner,” Strier said. “This is an important moment as Panama establishes its role in the global AI ecosystem.”

For Georgia Tech, the project illustrates how the Institute’s research and technical expertise can inform national-level technology planning.

Rather than focusing solely on AI development, the strategy connects technical capabilities with workforce preparation, research capacity, infrastructure, entrepreneurship, cybersecurity, and responsible governance. It also calls for efforts to expand access to AI opportunities in rural and underserved communities.

Van Hentenryck said developing the strategy required understanding Panama’s priorities alongside the opportunities presented by emerging technologies.

“Working on a national strategy requires bringing together different perspectives,” he said. “Panama has its own strengths, challenges, and ambitions, and those were essential to shaping a strategy that can work for the country.”

From strategy to implementation

The launch marks the beginning of the next phase: putting the strategy into action.

The road map includes recommendations for expanding AI education and workforce training, strengthening research capacity, improving digital infrastructure and cybersecurity, establishing responsible AI governance, and creating mechanisms to support entrepreneurship and innovation.

Franklin Morales, head of the International Technical Cooperation Office at SENACYT and project lead for the AI initiative, said the strategy provides a starting point for that work.

“Now this is not the end. It is not even the beginning of the end. But it is, perhaps, the end of the beginning,” Morales said, quoting Winston Churchill. “After months of work, Panama launched its National Artificial Intelligence Strategy, a vision of what this country needs to build for a new technological era.”

“The launch of the strategy is an important milestone, but the real work begins with implementation,” Brown said. “The focus now is on turning these priorities into lasting capacity, developing talent, and infrastructure while supporting research, entrepreneurship, and responsible AI adoption.”

For Panama, the strategy establishes a framework for using artificial intelligence to strengthen economic competitiveness, expand opportunity, and support the country’s role in an increasingly technology-driven global economy.

News Contact

Georgia Tech’s artificial intelligence (AI) research and education efforts are entering a new phase with $38.4 million in renewed support from the U.S. National Science Foundation (NSF) for two Georgia Tech-led National AI Institutes.

NSF has awarded $20 million to renew the National AI Institute for Advances in Optimization (NSF AI4OPT) through 2031 and $18.4 million to renew the National AI Institute for Adult Learning and Online Education (NSF AI-ALOE) for a new five-year period. The awards will support research at the intersection of artificial intelligence and optimization, as well as the development of AI technologies for adult learning, higher education, and workforce development. A third Georgia Tech-led National AI Institute, the AI Institute for Collaborative Assistance and Responsive Interaction for Networked Groups (AI-CARING), will continue its research under a one-year no-cost extension.

Together, the institutes represent a broad range of AI research at Georgia Tech, with work spanning energy systems, advanced manufacturing, education, reskilling, and workforce development.

The renewed investments come as NSF continues to prioritize artificial intelligence, advanced research infrastructure, and development of the nation’s STEM workforce.

“Georgia Tech has built a nationally recognized ecosystem for artificial intelligence research, education, and application,” said Tim Lieuwen, interim president and executive vice president for Research at Georgia Tech. “These NSF investments recognize the strength of our researchers and the importance of advancing AI in ways that address real-world challenges. The continued support for NSF AI4OPT and NSF AI-ALOE will allow Georgia Tech researchers to push the boundaries of what AI can accomplish while preparing students, educators, and the workforce for an increasingly AI-enabled future.”

Advancing AI-Powered Decision-Making

The $20 million renewal of NSF AI4OPT builds on five years of research at the intersection of artificial intelligence and mathematical optimization.

Led by Pascal Van Hentenryck, the A. Russell Chandler III Chair and professor in the H. Milton Stewart School of Industrial and Systems Engineering at Georgia Tech, NSF AI4OPT develops AI and optimization technologies for complex systems where decisions must be made quickly, efficiently, and under uncertainty.

In its next phase, the institute will place greater emphasis on real-time and interactive optimization, advancing scalable, efficient, and trustworthy approaches to decision-making. Research will continue to address challenges in areas such as energy, supply chains, transportation, and advanced manufacturing.

“NSF AI4OPT’s first five years demonstrated the power of bringing AI and optimization together to address some of the most challenging decision-making problems facing society,” said Van Hentenryck. “This renewal gives us the opportunity to build on that foundation and develop AI systems that can reason, optimize, and make decisions in real time. We are especially excited about advancing trustworthy AI for complex systems in energy, supply chains, and manufacturing, while continuing to educate the next generation of AI researchers and practitioners, starting with middle and high schoolers.”

NSF AI4OPT combines artificial intelligence, mathematical optimization, and computing to create systems that support decisions in complex and dynamic environments. Its research includes applications in critical infrastructure and industry, where decisions often involve competing objectives, uncertainty, and rapidly changing conditions.

The institute also has an education and workforce development mission, introducing AI and optimization concepts to students and educators while preparing learners for an AI-enabled workforce.

“NSF AI4OPT represents exactly the kind of interdisciplinary research that Georgia Tech is positioned to lead,” said Raheem Beyah, provost and executive vice president of Academic Affairs at Georgia Tech and former dean of the College of Engineering and Southern Company Chair. “The institute connects foundational advances in artificial intelligence and optimization with engineering challenges that have direct implications for our economy and society. Its continued investment in research and education strengthens Georgia Tech’s ability to prepare students to lead in a rapidly changing technological landscape.”

Advancing Adaptive Expertise Through AI

NSF AI-ALOE’s renewed $18.4 million award launches the institute’s next five years of research into how artificial intelligence can transform adult learning and online education.

The renewal builds on NSF AI-ALOE’s previous work, including the development of seven AI agents and a data architecture for AI-augmented learning. Its next phase will focus on adaptive expertise, or the ability to apply and transfer learned skills to new situations.

AI-ALOE’s research in Phase 2 will expand beyond traditional educational settings into workplace and vocational training, with a focus on occupational skills in areas including advanced manufacturing and data science.

NSF AI-ALOE will also build on advances in generative and agentic AI to develop AI agents that support the acquisition of complex skills across higher education and workplace environments.

“NSF AI-ALOE’s renewal reflects an important shift in how we think about AI and learning,” said Ashok Goel, director of NSF AI-ALOE and professor of computer science and human-centered computing in the School of Interactive Computing at Georgia Tech. “Our goal is not simply to use AI to deliver more content or automate instruction. We are exploring how AI can help people develop adaptive expertise, so they can transfer what they learn to new and changing situations. That has enormous implications for higher education, workforce development, and lifelong learning.”

The institute’s research will examine four dimensions of AI-supported learning: scalability, personalization, transfer of learned skills, and whole-person learning. It will also advance foundational AI research in machine teaching, metacognition in AI agents, and mutual theory of mind between humans and AI agents.

NSF AI-ALOE will use the data architecture developed in Phase 1 that is capable of ingesting, standardizing, analyzing, and visualizing information generated through interactions with AI tools, learning management systems, and student information systems. Researchers will use that data to better understand how learners and instructors interact with AI and improve AI-powered learning technologies.

The institute is also expanding partnerships with higher education institutions and industry. Its research network includes five colleges and universities and six industry partners in advanced manufacturing and related fields, creating opportunities to reach as many as 370,000 learners nationally through skilling, reskilling, and upskilling efforts.

“AI is changing both what people need to learn and how they can learn it,” said David Joyner, Zvi Galil PEACE Chair in the College of Computing and interim vice provost for AI in Education at Georgia Tech. “NSF AI-ALOE is uniquely positioned to study that transformation at scale. Its work can help us understand how AI can support learners without replacing the human elements of education, while giving people new ways to build and apply skills throughout their lives and careers.”

Continuing AI-CARING’s Research

The renewed support for NSF AI4OPT and NSF AI-ALOE builds on Georgia Tech’s broader participation in the National AI Institutes program.

Georgia Tech’s AI-CARING Institute will continue its research under a one-year no-cost extension, allowing the institute to carry forward its research objectives and partnerships.

The continued work of AI-CARING, alongside the renewed investments in NSF AI4OPT and NSF AI-ALOE, reflects the breadth of Georgia Tech’s AI research community and its work across disciplines.

From AI Research to Real-World Impact

The renewed NSF investments build on Georgia Tech’s broader efforts to advance AI research, education, and application through Georgia Tech AI and Tech AI.

Georgia Tech AI connects Tech’s AI community across research, education, and impact. Tech AI complements this work by serving as Georgia Tech’s engineering and delivery arm, bringing together researchers, engineers, research software engineers, students, and technical expertise to move AI projects from early-stage concepts toward operational systems.

This connection helps create a pathway from foundational research to technologies that can be developed, tested, and deployed in real-world settings.

For NSF AI4OPT, that work includes AI systems that support complex decisions in critical systems and industries, from energy and manufacturing to supply chains and transportation. For NSF AI-ALOE, it includes AI technologies that help people acquire, apply, and transfer skills across classrooms, workplaces, and lifelong learning environments.

Tech AI works at the application and deployment stage, partnering with researchers, industry, and public-sector organizations to develop and implement scalable AI solutions. Its work spans AI engineering and solution delivery, applied research, industry and government partnerships, workforce development, and responsible and trustworthy AI.

“Georgia Tech’s strength in AI comes from the connections between research, talent, and application,” said Timothy M. Brown, managing director of Tech AI. “Our NSF AI Institutes advance research across critical areas, while Tech AI serves as Georgia Tech’s engineering and delivery arm, working with industry and public-sector partners to turn AI expertise and research into deployable solutions. Together, these efforts create a pathway from AI innovation to real-world impact.”

Together with AI-CARING and Georgia Tech’s broader AI research community, the institutes contribute to a universitywide effort to advance AI that is technically innovative, human-centered, and designed for meaningful impact.

“AI is transforming virtually every sector, and Georgia Tech has a responsibility to help shape that transformation,” Lieuwen said. “These institutes bring together some of Tech’s strongest capabilities in research, education, and interdisciplinary collaboration. Their renewed support gives us an opportunity not only to advance the science of AI, but also to ensure that the benefits of these technologies reach the people, organizations, and communities that will rely on them.”

About the Georgia Institute of Technology

The Georgia Institute of Technology, or Georgia Tech, is one of the top public research universities in the U.S., developing leaders who advance technology and improve the human condition.

The Institute offers business, computing, design, engineering, liberal arts, and sciences degrees, as well as professional development and K-12 programs for fostering success at every stage of life. Its more than 56,000 undergraduate and graduate students represent 54 U.S. states and territories and more than 146 countries. They study at the main campus in Atlanta, at instructional sites around the world, and through distance and online learning.

As a leading technological university, Georgia Tech is an engine of economic development for Georgia, the Southeast and the nation, conducting more than $1 billion in research annually for government, industry and society.

About the National Science Foundation

The U.S. National Science Foundation (NSF) is an independent federal agency that supports science and technology in all 50 states, the District of Columbia, and U.S. territories. For more than 75 years, NSF has advanced the frontiers of STEM research and innovation. To keep the United States at the leading edge of discovery, NSF funds basic and solutions-oriented research that generates new knowledge to fuel industries of the future, create world-leading technologies, and improve the lives of the American people.

News Contact

Artificial intelligence (AI) chatbots can give people a comfortable place to discuss personal thoughts and feelings. However, when a chatbot constantly agrees with a user, it can reinforce false beliefs and potentially cause harm.

New research from Georgia Tech suggests that a simple change to a chatbot’s instructions could reduce that risk. The work was accepted to the ACM AI Letters publication. 

Researchers Lorenzo de la Loza and Vijay Madisetti tested eight AI models in 1,280 experiments involving simulated conversations about delusional beliefs. They found that instructions designed to discourage excessive agreement reduced the models’ confirmation of those beliefs by 73.8%. Adding a self-reflection step increased the reduction to 77%.

“We think that this paper will help AI providers and solution creators better understand how best to protect and safeguard their users from harming themselves,” Madisetti said. “This may make AI safer for all of us.”

AI chatbots are designed to be helpful and supportive. But that behavior can become risky when a chatbot agrees with a user’s beliefs without questioning whether they reflect reality. Researchers call this behavior sycophancy.

Courtney Crooks, a principal research scientist at the Georgia Tech Research Institute and adjunct faculty member in Georgia Tech’s School of Cybersecurity and Privacy, said chatbot conversations can have both positive and negative effects.

People may feel more comfortable discussing difficult thoughts with a chatbot than with another person. But if the chatbot always agrees, users may miss opportunities to question their beliefs or consider other points of view.

The same behavior that makes a chatbot feel supportive, Crooks explained, can also encourage harmful patterns of thinking.

The researchers tested instructions that told AI models not to reinforce users’ beliefs simply to be agreeable. The team evaluated the models using 16 clinically derived scenarios involving delusional beliefs.

The team also tested other safety tools, including systems designed to flag potentially harmful content. The results suggest that changing a chatbot’s instructions can be an effective way to reduce excessive agreement in the scenarios tested.

Crooks called the findings a promising direction for future research combining psychology and AI safety.

The study did not test people in clinical settings or prove that the approach prevents real-world harm. However, it offers developers a potential way to make chatbots more careful when responding to vulnerable users without building an entirely new AI model.

News Contact