Exploring the Impact of Generative AI on Education, Research and More
UC San Diego summit examines frontiers of AI and its impact on society
Story by:
Published Date
Article Content
Better simulations to understand how viruses work. Better ways to model and predict climate change. Better robots that can navigate the real world.
Researchers from all across the University of California San Diego and around the world converged on campus last week to explore the promise of generative artificial intelligence (AI).
These AI models—already well-known for generating text, images, video, and music—stand to play an outsized role in many aspects of society moving forward, including research and education. The UC San Diego summit brought together researchers from many different disciplines to discuss opportunities and challenges that are ahead as generative AI becomes more present in our lives.
“GenAI is changing the landscape of research and education, and we are very proud to be part of this revolution,” said Rose Yu, a UC San Diego computer science professor and one of the two primary organizers of the event. “We are excited for the opportunities it will bring for UC San Diego and for the Department of Computer Science and Engineering.”
The event gave attendees opportunities to get in-depth points of view on various aspects of generative AI, from frontier technologies, to its impact on scientific discovery, to societal and safety implications, to creating a whole new way to make music and interact, said Yu and fellow computer science faculty member Taylor Berg-Kirkpatrick, the other co-lead organizer of the event.
The GenAI Summit 2025 was organized by faculty across several UC San Diego departments and led by the Department of Computer Science and Engineering. The event also showcased speakers from leading AI companies, such as Meta and Google AI, as well as companies that use generative AI for various applications, such as Qualcomm, Service Now, Cygames and Hillbot. Speakers traveled from afar to share their insights. UC San Diego students and faculty also showcased their groundbreaking research across many areas.
A long AI history
UC San Diego played a critical role in the development of artificial intelligence in its early days and is home to the nation's first cognitive science department, UC San Diego Chancellor Pradeep K. Khosla explained. Indeed it is here that Nobel Prize winner Geoffrey Hinton did some of the seminal work that led to his Nobel Prize in physics for discoveries that brought about machine learning with artificial neural networks. Hinton was a postdoctoral researcher and visiting faculty and collaborated with Terrence Sejnowski, now a key figure at UC San Diego and the Salk Institute, and Garrison Cottrell, now a professor in the UC San Diego Department of Computer Science and Engineering.
Today, AI is woven through a large number of research projects across every corner of campus. The fact that UC San Diego helped develop the roots of AI and is today an AI pioneer reflects campus’ longstanding cross-disciplinary research ecosystems. UC San Diego’s computer science ecosystem and its larger campus ecosystem are renowned for bringing together theory and fundamental and applied research–and this is a powerful combination, noted Albert P. Pisano, Dean of the Jacobs School of Engineering and Special Adviser to the Chancellor. Bringing all these elements together is how real collaborations take place, and how real problems get solved, he said.
The summit included a poster session, where Ph.D. students from across campus showcased a broad range of work.
Generative AI and Computer Science at UC San Diego
When it comes to the future of generative AI, one key player at UC San Diego is the Department of Computer Science and Engineering, said department chair Sorin Lerner.
Computer science researchers at UC San Diego recently launched the Laboratory for Engineered Intelligence, which developed an AI tutor that was deployed in a pilot program in programming and nanoengineering classes this academic year. A broader deployment is projected for the 2025–26 academic year.
The department will also launch an AI major in the fall, in cooperation with the Halıcıoğlu Data Science Institute (HDSI). The AI courses in the program are drawn from many departments across campus, including Computer Science and Engineering, the Halıcıoğlu Data Science Institute, Cognitive Science, Mathematics, Electrical and Computer Engineering and Philosophy.
“The core philosophy of this program will be to provide strong and enduring foundations, while allowing students to explore a breadth of AI application domains,” Lerner said.
These undergraduate students, and Ph.D students already working on AI, are the talent that will go on to work at companies focusing on and working with AI. Researchers already collaborate regularly with industry.
“At Qualcomm, we are collaborating with UC San Diego to push the boundary of AI research and to bridge the gap between research and real world applications,” said An Chen, Vice President of Engineering at Qualcomm. “We at Qualcomm are working to bring AI into the real world.”
From left: summit co-organizer Taylor Berg-Kirkpatrick, ServiceNow VP Magaly Drant, Qualcomm VP An Chen and Shuichi Kurabayashi, Director of Cygames Research, during the summit's industry fireside chat.
A wide range of applications
The summit was organized around four themes: Frontier, Science, Society and Creativity. A common thread running through all these themes was the various applications of AI.
“UC San Diego has a strong foundation in bringing together researchers who are serious about using computing to advance science and engineering, and that has led to exciting advances in public health,” said Rommie Amaro, a professor in the Department of Molecular Biology.
She demonstrated how her team was able to develop high-resolution computer simulations to create the first-ever atomic-level computer model of aerosolized SARS-CoV-2, the virus that causes COVID-19. “Simulations allow us to see the unseen,” she said. “They give us new discoveries in biology that otherwise wouldn't be possible.”
AI will be instrumental in making more discoveries in health and biology, Amaro said. Already, machine learning methods have allowed researchers to look more accurately at biological processes happening at the quantum level.
Meanwhile, researchers led by Yu, the conference co-organizer and faculty member in the UC San Diego Department of Computer Science and Engineering, showed that the algorithms behind generative AI tools like DallE, when combined with physics-based data, can be used to develop better ways to model the Earth’s climate. “We were able to successfully generate fast and accurate 10- to 100-year-long global climate simulations at 6-hourly resolution,” said Salva Ruhling Cachay, a Ph.D. student in Yu’s research group, during his presentation at the GenAI Summit.
Hao Su, also a faculty member in the UC San Diego Department of Computer Science and Engineering, is working on helping generative AI understand the real world. The models he creates could be used to improve the way robots navigate. “The ability to understand, reason about and generate 3D structures is fundamental for intelligent agents operating in the real world. Recent advances in 3D generative AI have opened new frontiers in spatial intelligence, enabling breakthroughs in areas such as scene reconstruction, object synthesis, robotic perception and embodied AI,” Su said.
GenAI can also be a partner in human creativity, such as music and visual arts. Computer science Ph.D. student Zachary Novack showcased Presto!, a model that can generate music from text prompts faster and more efficiently than the state of the art. He is advised by computer science faculty members Julian McAuley and Taylor Berg-Kirkpatrick. Meanwhile, computer science Ph.D. student Sophia Sun presented "Learning to Move, Learning to Play, Learning to Animate" which is a cross-disciplinary multimedia performance she created in collaboration with Ph.D. students from music and visual arts. The performance features robots made from natural materials and uses technologies like AI-generated visuals, real-time biofeedback and electroacoustic sound.
Rommie Amaro, a professor in the Department of Molecular Biology, showed high-resolution computer simulations used to create the first-ever atomic-level computer model of aerosolized SARS-CoV-2.
AI under the hood
UC San Diego researchers also are working toward improving the processes running under the hood of AI and generative AI models. Haiyang Xu, a Ph.D. student in the research group of cognitive science faculty member Zhuowen Tu, showcased a novel predictive algorithm that achieves state-of-the-art results on both synthetic and real-world benchmarks for 3D shape reconstruction.
Other researchers are working toward getting a better understanding of how AI and generative AI models actually work. HDSI Professor Mikhail Belkin argues that a method known as recursive feature machines—essentially an algorithm that selects features within neural networks and their training sets—is the best tool to understand some of the qualities these models display. One example is understanding “grokking,” a phenomenon where a neural network or a large language model experiences a sudden and dramatic improvement in performance.
Meanwhile, Tuomas Oikarinen, a Ph.D. student in the research group of HDSI faculty member Lily Weng presented a method that scales neuron-level understanding up to large models with thousands or millions of individual components, faster or more efficiently. “With deep learning models performing exceptionally well across a wide range of tasks, it is more important than ever to understand how they work. Understanding these models is important to improve safety and reliability, and to place appropriate trust in them,” he said.
Safety and privacy
Chhavi Yadav, a Ph.D. student in the research group of computer science Professor Kamalika Chaudhuri, is tackling a different issue. “In principle, explanations are intended as a way to increase trust in machine learning models and are often obligated by regulations. However, many circumstances where these are demanded are adversarial in nature, meaning the involved parties have misaligned interests and are incentivized to manipulate explanations for their purpose,” she said. She developed a method that seeks to solve this problem.
In addition, like all computer systems, AI models are subject to cybersecurity risks. For example, a team led by Earlence Fernandes, a faculty member in the UC San Diego Department of Computer Science and Engineering, demonstrated that specific prompts could be used to get chatbots to steal and leak a user’s personal information. Instead of focusing on protecting AI models in isolation, a system-level approach would be more effective, Fernandes said. “We need to improve the infrastructure supporting AI systems with a particular focus on authorization protocols that allow AI systems to access external resources,” he said.
HDSI Professor Mikhail Belkin gave a talk during the summit.
Industry and research panels
In addition to talks, the event included two panel-style components. An industry fireside chat took place Thursday, with Qualcomm VP An Chen, Magaly Drant, Vice President Developer Productivity at ServiceNow and Shuichi Kurabayashi, Director of Cygames Research and Visiting Associate Professor, Tokyo University of Science. It was moderated by Berg-Kirkpatrick, event lead coorganizer with Yu.
The Summit ended with a panel discussion moderated by computer science faculty member Raj Ammanabrolu, on how AI would impact society in the future. The three panel speakers were Elad Hazan, Director/Co-founder of Google AI Princeton and Professor at Princeton University, UC San Diego computer science faculty member Su, and Anna Huang, Assistant Professor at the Massachusetts Institute of Technology.
Other event participants included: Russ Salakhutdinov, Vice President of Research, GenAI at Meta and Professor, Carnegie Mellon University; Shakir Mohamed, Research Director at Google DeepMind; and Nao Tokui, Artist and Researcher, and Founder of Qosmo and Neutone.
Learn more about research and education at UC San Diego in: Artificial Intelligence
A summit participant takes a selfie with Jacobs School Dean Albert P. Pisano, Chancellor Pradeep K. Khosla, summit co-organizer Rose Yu and Office of Research Affairs and Innovation Senior Director Miwako Waga.
Share This:
Stay in the Know
Keep up with all the latest from UC San Diego. Subscribe to the newsletter today.
