Write the letter A 100 times and line them up. One leans slightly left. Another has a longer leg. Same letter, same hand, never quite the same result.

“We’re not computers, right?” said Woon Ju Park, a Georgia Tech neuroscientist and assistant professor of psychology.

Park uses the handwriting analogy to explain something about hearing. Listen to the same note again and again, and you may hear or judge it a little differently each time. The sound hasn't changed. Your response to it has.

In a study published in The Journal of Neuroscience, Park and her colleagues Jasmine Awad and Ione Fine tested nine adults who had been blind since birth or before age 2 and 18 sighted adults of similar ages and musical backgrounds.

Volunteers listened to pairs of notes and pressed a button to indicate whether the second was higher or lower. The adults who grew up blind were better at telling the notes apart. But in another test, the groups performed about the same when asked to identify which of two bursts of noise contained a faint tone.

Park suspected the pitch advantage came from more finely tuned hearing, allowing people to pick up smaller differences between notes. But when the researchers used computer models to test possible explanations, they found stronger support for something else: internal noise.

It isn't static or background sound. Internal noise refers to the small variations Park described with her handwritten A's. The models suggested that adults who grew up blind had less of that variation when hearing or judging pitch, which could explain why they were better at telling similar notes apart.

“We really didn't expect internal noise to play a role,” Park said. “We were surprised and wanted to make sure we weren't seeing something that wasn't really there.” The researchers drew that conclusion from the volunteers' answers, not measurements of their brains. They still don't know exactly where the variation arises.

Musical experience showed a similar pattern. Among both blind and sighted volunteers, people with more years of musical training tended to have lower internal noise in the models. But musical training alone couldn't explain the difference between the groups.

Park wants to know whether the same advantage appears in everyday listening, from following conversations in noisy places to recognizing what's happening through sound. She also wants to understand whether experience and practice can change how reliably people distinguish sounds.

“The brain is working really hard to tune its responses to the world that matters to us,” she said.

Park is affiliated with the Institute for Neuroscience, Neurotechnology, and Society. Her research was supported by the National Institutes of Health and Georgia Tech's Smithgall-Watts Early Career Award.

News Contact

Trillions of microorganisms make their home in the human gastrointestinal tract, where they influence everything from digestion and vitamin synthesis to immune response. This microbial world is remarkably diverse — a complex ecosystem of bacteria, archaea, fungi, viruses, and single-celled organisms called protists. Yet research has largely focused on just one group: bacteria.

Now, new work from Georgia Tech School of Biological Sciences Assistant Professor Abigail Lind is looking beyond bacteria by investigating a common gut protist called Blastocystis.

In support of her research, Lind has received a prestigious Maximizing Investigators’ Research Award (MIRA) from the National Institutes of Health (NIH). The MIRA program provides investigators with five years of flexible funding, allowing them to pursue ambitious research in promising new directions.  

Protists have long been understudied, Lind says, because until recently, these organisms were considered possible pathogens. “Recently, studies have found that they are actually correlated with many different positive health markers, like reduced inflammation and positive cardiovascular outcomes,” she explains. “But we still know very little about these organisms.”

“With this research, we want to lay a foundation to accelerate future work,” Lind adds. “We want to understand what these protists do, how they interact with microbial partners, and how that might ultimately impact their human hosts.”

Understanding the Microbiome Community

The protists are notoriously tricky to work with. They swell up and “pop” if they’re exposed to oxygen, and nobody is quite sure what they eat. “We’ve developed new techniques to work with them in the lab, which is why we’re able to do this research now,” Lind says. 

While decades of work have gone into cultivating gut bacteria in the lab, cultivating Blastocystis is a different story. “Many of the techniques developed to cultivate bacteria kill this protist,” Lind says, “so we had to be creative.” 

Postdoctoral scholar Fredrick Leon tested combinations of nutrients ranging from store-bought protein powders to boiled wheatgrass solutions, eventually settling on the formula they use today. But it still is unclear what exactly the organism is eating — and how.

“We know the strategies that bacteria use to break down and utilize food, but these protists don't have the same genes,” she says. “Do they eat something completely different? Do they eat the same things but with a different strategy? And how does that impact the microbial community and human host? These are the questions that we're answering.” 

Building a Genetic Blueprint

Lind also aims to develop genetic tools for studying Blastocystis. Those tools could help researchers uncover which genes are responsible for specific functions, providing a clear map of what the protist does and how it does it. 

By establishing new ways to grow, manipulate, and study Blastocystis, Lind hopes the project will provide answers about this organism while creating new tools to advance protist research more broadly.

“Our goal is to make Blastocystis a system that researchers can really begin to understand,” Lind says. “What we learn from this one organism could give us a foundation for studying many other protists and, ultimately, a more complete picture of the human gut microbiome.”

News Contact

Just over an hour outside Atlanta, Dalton – Georgia’s so-called “Carpet Capital of the World” – has become a focal point for concerns about PFAS, a group of long-lasting industrial chemicals increasingly found in waterways and drinking water. 
 

While PFAS contamination has drawn widespread attention, researchers at the Georgia Tech Research Institute (GTRI) are working on practical solutions to improve how these chemicals are detected, removed from water, and ultimately destroyed. Their work focuses on three key areas: detection, removal, and destruction technologies that aim to give utilities, businesses, and communities clearer, faster insight into where PFAS is — and what to do about it.  
 

PFAS, or per- and polyfluoroalkyl substances, are a large family of human-made chemicals that includes thousands of different compounds. They have been used for around nine decades to make products resistant to water, stains, and heat. They show up in many items, ranging from firefighting foams, dental floss, and carpets to nonstick cookware and outdoor clothing. PFAS materials don’t break down easily in the environment, possibly persisting for hundreds or even thousands of years.
 

Studies have linked PFAS exposure to serious health risks, including some cancers, immune system effects, and heart disease. But unlike many contaminants, the primary concern about these chemicals is not a single, high-level exposure. Instead, it’s long-term, low-level exposure.  

Read more in the GTRI Newsroom
 

News Contact

What’s a lizard’s favorite tree? Where do monarch butterflies stop as they migrate through Atlanta? And how can urban areas make room for wildlife?

Soon, Georgia Tech students will be investigating these questions and more. A new initiative from the School of Biological Sciences is transforming Georgia Tech’s campus into a living laboratory for urban biodiversity research, with students at its center.

Using tiny Bluetooth-enabled trackers, students will tag and track butterflies, lizards, and birds. They’ll use the same groundbreaking new technology being employed by conservation biologists worldwide, including Benjamin Freeman and James Stroud, who are bringing their experience with large-scale wildlife tracking to the new campus initiative. Both Stroud and Freeman are Elizabeth Smithgall Watts Early Career Assistant Professors in the School of Biological Sciences and co-leads of the new project.

“Recording animal behavior in the wild is one of the final frontiers in ecology and evolution research,” says Stroud. “This new tracking technology will allow students to design and conduct research projects and learn data collection, spatial analysis, statistical modeling, and scientific communication — all while investigating real conservation questions about urban wildlife corridors, migration patterns, and habitat use.”

The program aims to involve more than 800 students each year across multiple programs and degree paths. “The big idea is to connect Georgia Tech undergraduates with cutting-edge science on campus, and to get them involved with all facets of this research,” Freeman says. Beyond ecology and biology, students will learn to engineer sensor technology, build data visualization tools using computer science, and analyze urban planning decisions through a public policy lens.

“Over the next year, undergraduates will begin deploying these tracking systems and conducting preliminary data analysis, but what’s really exciting is how this project can build on itself,” Freeman adds. “In five years, the Georgia Tech community will have built a dynamic campus biodiversity map — one that anyone can access to see hundreds of tagged birds, lizards, and butterflies moving across campus.”

Tracking Wildlife at New Scales

There’s no shortage of animals on Georgia Tech’s campus — the Stroud Lab recently surveyed a population of green anole lizards in the Kendeda Building EcoCommons area, counting about 50 individuals. That presents a unique opportunity to tag and track each individual, he says, allowing students to study behavioral and ecological questions at the population level.

Freeman will help students tag northern mockingbirds, northern cardinals, eastern bluebirds, and eastern towhees, in addition to migratory birds that pass through campus in the spring and fall. “We’ll also tag monarch butterflies and gulf fritillary butterflies as they migrate through the EcoCommons in fall,” he says. 

Both Freeman and Stroud have deep experience with the technology. For more than a decade, Stroud has pioneered one of the longest-running evolutionary studies of its kind: catching, documenting, and releasing each of the roughly 1,000 lizards living on “Lizard Island.” In 2025, he was awarded a Packard Fellowship to equip each lizard on the island with tiny sensor backpacks to document their behaviors and movements in real time.

In Ecuador, Freeman is building a similar project with support from a Packard Fellowship — a living laboratory spanning thousands of meters along the flanks of “Tech Mountain,” from lowland rainforest to cloud forest. Leveraging the same technology Georgia Tech students will use on campus, he is tagging and monitoring the daily movements of tropical birds and mapping their movements against microclimate sensors placed on the mountain’s slopes. It’s the first project of its kind.

“One of the most exciting things is how accessible this new technology is,” he says. “When it’s deployed on Georgia Tech’s campus, students will get to play a role in every single stage of the research.”

Connecting Campus and Community

At Georgia Tech, that research will build on the living laboratory already centered in the Kendeda Building EcoCommons, where environmental conditions like wind, temperature, humidity, and soil quality are monitored by a variety of sensors. Students are already monitoring environmental data through the popular Living Building Science Vertically Integrated Program (VIP) course. 

Considering climate information alongside animal movement patterns will offer a unique window into how animals respond to changing conditions, Stroud explains. “How do these environmental factors affect animals, and how can we use that information to support wildlife as our campus continues to develop?” he asks. “With this new data, students can lead projects to answer these questions and more.”

Freeman and Stroud also hope that the Kendeda Building’s accessible location will make the research a visible part of campus life, giving community members and visitors the opportunity to encounter the project firsthand.

“This work can have a meaningful impact far beyond the classroom,” Freeman says. “By understanding how wildlife uses the spaces around us, we can better support it in a changing world, all while training and empowering the next generation of scientists and leaders.”

News Contact

As artificial intelligence (AI) becomes part of everyday life, ensuring it provides ethical, reliable information is paramount. New research from Georgia Tech, in partnership with OpenAI, explains how and why large language models (LLMs) like ChatGPT hallucinate, and what developers can do to mitigate the behavior.

The study, Evaluating Large Language Models for Accuracy Incentivizes Hallucinations, was recently published in Nature. 

Understanding why hallucinations arise is a key factor in mitigating them, says Georgia Tech Regents’ Professor Santosh Vempala, who co-authored the paper. “We’ve seen models hallucinate everything from references to simple facts, and this can be a real problem when users take that information as true,” he adds. “It’s important to determine to what extent this is unavoidable and to what extent it’s an issue of what we're focusing on and not prioritizing.”

Vempala is the Frederick Storey II Chair of Computing in the College of Computing, an adjunct professor in the H. Milton Stewart School of Industrial and Systems Engineering, and an adjunct professor in the School of Mathematics, where he specializes in studying algorithms and high-dimensional geometry. He is also director of the Algorithms, Combinatorics and Optimization Ph.D. program.

As LLM usage becomes increasingly prevalent, Vempala aims to use his expertise to increase AI safety. “We need to identify and understand the dangers, but we also need to form principles around them,” he explains. “That’s a key goal of my work now.”

“Intuition is helpful and important, but rigorous principles emerge only when you write them down mathematically,” he adds. “We need a model that's precise and scientific, that mathematically quantifies principles that could increase AI performance while keeping risk under control.”

Understanding AI Hallucinations

Vempala sees this work on hallucinations as a small first step toward increased AI safety. “Hallucinations are inevitable because LLMs are calibrated to a limited dataset,” he explains. “Until a model sees every piece of information in the world, and assuming every piece of information that it sees is correct, there is a chance it will have to guess the answer to some questions — and that could lead to hallucinations.”

This idea builds on previous research published in 2024 by Vempala and Adam Kalai, then a research scientist at OpenAI and now the cofounder of the Institute for Responsible Superintelligence (RESI), an AI safety nonprofit. The work showed that hallucinations are statistically unavoidable in trained AI models for certain types of data. 

Instead of changing the training dataset that an LLM absorbs, the new research focuses on post-training, when the LLM is tested on how well it can answer questions and align with the developer’s goals. During these tests, almost always, the AI model is penalized for saying “I don’t know.” That penalty is the same as if the AI gave a wrong answer.

Just like a student taking a multiple choice test, the model is incentivized to guess when it doesn’t know the correct answer. “This issue is industry-wide,” Vempala says. “The benchmarks and methods being used for evaluation encourage this type of guessing behavior.”

By uncovering the mechanism driving hallucinations, the study provided developers with an easy-to-implement method to mitigate hallucinations in post-training. Specifically, they propose open rubrics, which clearly specify the scoring mechanism for LLM responses.

Building Better Guardrails

But hallucinations are just the tip of the iceberg. Now, Vempala is collaborating with psychologists, philosophers, and other experts to see what can be done to ensure that AI behaves in a way that aligns with ethical standards.

For example, many people use LLMs with the model in a specific role: teacher, nurse, tax advisor. Vempala is interested in how to enable role-based abstention. “If the AI is playing a role, like a doctor, how do we ensure that it continues to behave in an ethical way within the scope of that role?” he says. “How can we ensure it’s giving accurate, credible, and ethical information to people, and how can we make sure it declines to answer inappropriate questions?”

Another issue Vempala is working on is called reward hacking, which happens when models try to maximize their “score” results through any means necessary. 

In one recent example of this, an OpenAI model was being tested and realized that the answers were available on another website. Rather than do better on the tests, it deployed hundreds of agents to try and steal the test answers.

What do we do to prevent this? How do we understand the possibilities of reward hacking and how do we guard against it? “Some of this comes down to ensuring AIs behave in a way that ethically aligns with our guidelines,” Vempala says. “But it’s also impossible to write down rules to cover every scenario, so it’s vital that we begin to understand what leads to behaviors like reward hacking and how to avoid them.”

The Self-Improving Era

Now, AIs are at an important inflection point: they are developing themselves. “Until just a few months ago, humans were deciding what to do next in AI development,” Vempala explains. “Now the AI is self-improving, which makes it even more important to discover and implement these safety protocols.”

The work is not a traditional math problem, he says — it’s a problem for society as a whole. “We can’t just work individually or in small groups on this issue. Everybody needs to pay attention and contribute,” he adds. “Whether you’re working in theoretical mathematics, or ethics, or psychology, developing the principles of AI safety will involve all of us in different ways. There is a real need for people from all walks of life to be involved in this journey.”

DOI: https://doi.org/10.1038/s41586-026-10549-w

Funding: Santosh Vempala was supported in part by the National Science Foundation (NSF) and the Simons Foundation.

News Contact

At Georgia Tech Research Institute, 40 percent of our workforce are veterans who continue serving the nation through mission-critical research. For these professionals, the transition from military service to GTRI is not a career change but a continuation of their commitment to national security. The skills they developed in uniform (operating complex systems, leading teams under pressure, understanding real-world military operations) directly strengthen the research they conduct today.

This Veterans Month, we're proud to share the stories of six GTRI employees whose service across the military branches exemplifies our core values: integrity, innovation, and dedication to the nation's defense.

Read their stories.

News Contact

While studying health systems around the world during a yearlong fellowship, Abel Sapirstein traveled to northern Rwanda to visit Butaro Cancer Center of Excellence, at the time, the country's only publicly accessible facility to offer comprehensive oncology services.  

The center was intentionally placed in a rural area to bring specialized cancer care closer to patients who might otherwise struggle to reach it.  

While staying in Rwanda’s capital, Kigali, Sapirstein prepared to make the roughly 60-mile trip north to Butaro. Before leaving, he received unexpected advice: Get there before it rains.  

The journey from Kigali to Butaro takes about two hours in good conditions. But heavy rains often wash out roads along the way, making the journey far less predictable, and in many cases, impossible.  

For Sapirstein, the travel advice raised a more fundamental question about healthcare access: What good is care if patients can't reach it?  

Sapirstein, a Ph.D. student in Georgia Tech’s H. Milton Stewart School of Industrial and Systems Engineering (ISyE), would eventually turn that question into an optimization problem. Working with ISyE faculty Lauren Steimle, the Harold R. and Mary Anne Nash Early Career Professor and assistant professor, and Mathieu Dahan, associate professor, he developed a model to help health systems decide how to invest limited resources in cancer facilities and roads while accounting for shortages of oncologists and unpredictable weather.  

The model showed that improving roads can complement investments in new treatment facilities, expanding access even in areas where building additional facilities may not be viable. The work was recently published in Health Care Management Science.  

“Geographic access is a very neat concept, because there’s a whole lot of other dimensions to access, such as, can people afford treatment or do they even want treatment?” Sapirstein said. “But if it’s not geographically accessible, those other dimensions are kind of moot.”  

After graduating from Harvey Mudd College in 2021, Sapirstein received a Thomas J. Watson Fellowship, which supports a year of independent exploration abroad for graduating seniors from participating liberal arts colleges. He centered his fellowship on health equity, examining how different countries structure and deliver healthcare.  

Rwanda wasn’t originally part of his plan. After a visa issue disrupted his itinerary, Sapirstein reached out to people working in global health and eventually connected with Partners In Health, an international nonprofit that works to expand healthcare access in underserved communities.  

He spent about four months in Rwanda learning about the country’s health system and the challenges of expanding access to care.  

Those challenges are particularly acute for cancer care.  

Even during Rwanda’s dry seasons, which generally run from June to September and December to February, 70% of Rwandans travel more than two hours for care.  

Only 35% of estimated cancer cases are diagnosed, and 75% of diagnosed cases are caught at severe stages, when prognoses are worse. This is estimated to contribute to 10% of premature deaths.  

Sapirstein boiled the optimization problem down to a simple tradeoff.  

“Basically, you have a fixed pot of money, and that money can either be used to build more cancer hospitals or it can be used to pave your roads,” he said. “In parallel to that fixed pot of money, there’s also a fixed number of oncologists.”  

Building more cancer facilities might seem like the obvious solution, but each new facility also requires trained oncologists to staff it. Improving roads offers another lever, making existing care easier to reach without requiring another medical team.  

To evaluate those tradeoffs, Sapirstein, Steimle and Dahan built a model of Rwanda’s cancer-care network using roughly 7,500 road segments, 1,066 population centers and 48 potential facility locations. The model evaluates investments in both roads and cancer facilities over five years while accounting for limited budgets, oncologist availability, and changing weather conditions.  

The results reinforced what Sapirstein had encountered on the road to Butaro.  

In one scenario with a $5 million budget and four oncologic teams, average travel time during the rainy season fell from six hours to three. The model also showed that road improvements could meaningfully improve access in areas where additional cancer facilities could not be built.  

For Sapirstein, however, the model is not necessarily a prescription for solving cancer access in Rwanda.  

“I don’t want the narrative to be there’s this big problem, we solved it,” he said. “It’s still a big problem.”  

His research is instead intended to add another consideration to discussions about expanding cancer care: healthcare infrastructure doesn’t start at the hospital door, so investments in improving access shouldn’t either.  Roads, staffing and the conditions patients encounter on their way to care play an enormous role. This modeling work could help decision-makers explore potential solutions and their anticipated costs and impacts on access.  

“A big part of what I’m trying to do is show how systems engineering and systems thinking is useful in addressing problems like these and how engineers can approach these solutions responsibly," he said.   

News Contact

David Joyner, Georgia Tech's interim vice provost for AI in Education, has been named to CDO Magazine's 2026 Top 30 Higher Education Data & AI Influencers list. The annual recognition highlights leaders shaping how colleges and universities use artificial intelligence (AI) and data to advance higher education.

“I'm honored to be recognized among CDO Magazine's Top 30 Higher Education Data & AI Influencers for 2026,” Joyner wrote in a LinkedIn post.

Joyner was recognized for his contributions to Georgia Tech's academic AI strategy, including his leadership in advancing the 2026 Academic AI Playbook. The publication brings together 67 faculty contributions representing every College at Georgia Tech, along with 20 practical lessons drawn from classroom experiences. It provides educators with examples and insights for thoughtfully integrating AI into teaching, redesigning assignments and preparing students for an increasingly AI-enabled world.

As interim vice provost, Joyner leads implementation of Georgia Tech's Academic AI Strategy, coordinating efforts across Colleges and administrative units. The strategy centers on three goals: advancing AI-driven teaching and learning, equipping faculty and students to build and integrate AI tools effectively and responsibly, and strengthening Georgia Tech's leadership through responsible AI innovation and engagement.

A central part of this work is ensuring faculty expertise and academic freedom remain at the forefront of how AI is introduced into the classroom. Rather than prescribing a single approach, Georgia Tech's strategy supports faculty-led experimentation, expands access to AI resources and encourages the sharing of effective practices across disciplines.

Joyner also serves as executive director of Online Education and the Online Master of Science in Computer Science (OMSCS) program in the College of Computing. His work in online education, educational technology and learning at scale has helped shape how Georgia Tech delivers accessible, technology-enhanced learning experiences to students around the world.

The recognition reflects Joyner's contributions at a time when higher education institutions are working to understand and respond to AI's growing influence on teaching, learning and institutional operations.

CDO Magazine's annual list recognizes 30 leaders across North American higher education for their work advancing AI innovation, institutional strategy, responsible technology adoption and data-informed decision-making.

Learn more: Explore the CDO Magazine 2026 Top 30 Higher Education Data & AI Influencers.

Drones can already provide a valuable view of disaster areas, search difficult terrain, and reach places where sending people may be difficult or dangerous. The next challenge is enabling multiple drones to work together as a swarm, covering larger areas while staying connected throughout complex missions. 

Karthikeyan Sundaresan is tackling that challenge through Project OMNISCIENT, a three-year effort supported by a $4.8 million award through the National Science Foundation’s Verticals-enabling Intelligent Network Systems (VINES) program.  

The project will develop an intelligent wireless network that can adapt as drones move, gather information, and take on different roles. 

“The goal is to move beyond thinking about one pilot and one drone,” Sundaresan, a professor in the School of Electrical and Computer Engineering, said. “We want the focus to be on the mission while the drones and the network figure out how to work together.”

Read more...

News Contact

Tech AI, the applied AI initiative within Georgia Tech AI at Georgia Tech, and Velastegui Ventures are advancing an enterprise AI platform designed to make organizational information easier to find, understand and use.

The collaboration centers on an enterprise AI knowledge platform that allows users to interact with information through natural language queries. The platform brings together information from documents, websites, databases and other organizational sources, giving users a more intuitive way to search for and work with institutional knowledge.

The project has moved from Phase 1, which established the platform’s technical foundation, into Phase 2, which will focus on scalability, operational maturity and the capabilities needed to support broader enterprise use.

Enterprise AI Knowledge Platform

During Phase 1, the Tech AI team developed a working platform that combines retrieval-augmented generation, semantic search and citation-backed responses.

Retrieval-augmented generation, or RAG, allows an AI system to retrieve relevant information from a collection of sources before generating a response. Combined with semantic search, this approach helps the platform identify information based on meaning and context rather than relying solely on exact keywords.

The platform also provides citations with its responses, allowing users to trace information back to its original sources.

Together, these capabilities establish the foundation for the broader Tech AI Knowledge Engine vision: using AI to make organizational knowledge easier to access, understand and apply.

“This collaboration shows how Georgia Tech can work with industry partners to move an AI concept from development toward practical application,” said Ajay Upadhyaya, associate director of research administration at Tech AI. “Phase 1 gave us the opportunity to develop and test the technical foundation, while Phase 2 allows us to focus on the systems and processes needed to support broader enterprise use.”

From Prototype to Practical Application

Phase 2 will build on the capabilities developed during the initial phase, with an emphasis on the requirements for broader enterprise adoption.

The team will focus on scalability, governance, monitoring, analytics and administrative capabilities. These additions are intended to help organizations manage the platform more effectively while providing greater visibility into system performance and user activity.

The transition also reflects a broader challenge in applied AI: moving technologies from working prototypes into systems that can operate reliably in real-world environments.

“Most organizations don’t have a knowledge problem. They have an access problem. The answers are already in their documents and systems, but they’re hard to find and harder to trust,” said Nicholas Velastegui, head of operations at Velastegui Ventures. “Phase 1 showed that AI can give people grounded, cited answers. Phase 2 is about making that dependable enough for an organization to run on.”

Team Behind Project

The project brings together researchers, engineers and program leadership from Tech AI.

Alex Djalali, senior research software engineer, serves as the project’s principal investigator and technical lead. Dave Brownell, head of engineering at Tech AI, provides engineering leadership and project oversight.

Research software engineers Austin Weeks and Jeremy Garcia contribute to the platform’s development, engineering and implementation. Upadhyaya supports project coordination, contract management, sponsor engagement and overall program activities. Tim Brown, managing director of Tech AI, serves as the executive sponsor.

As the collaboration moves into Phase 2, the team will continue strengthening the platform’s technical and operational capabilities while identifying additional opportunities to apply AI to complex organizational knowledge.