New interdisciplinary center will bring together data science, computing, arts and humanities to study alignment, evaluation and human oversight of <a href="https://myappsplus.com/trump-says-he-will-ban-everyone-from-using-the-phrase-artificial-intelligence/" title="Trump says he will ban everyone from using the phrase ‘<a href="https://myappsplus.com/artificial-intelligence-transforms-the-workforce-regulations/" title="Artificial intelligence transforms the workforce, regulations”>artificial intelligence’”>artificial intelligence
Story by:
Media contact:
- Tiffany Fox
–
tmfox@ucsd.edu - Erika Johnson
–
erjohnson@ucsd.edu
Published Date
Story by:
Media contact:
- Tiffany Fox
–
tmfox@ucsd.edu - Erika Johnson
–
erjohnson@ucsd.edu
Article Content
- UC San Diego is launching the Center for Principled Agentic Intelligence (CPAi), an interdisciplinary partnership between the Halıcıoğlu School of Data Science and Computing and the School of Arts and Humanities.
- The center will study AI alignment, human oversight and independent evaluation, including development of an open-source testbed for assessing agentic AI systems.
- CPAi launches with $1 million in support over three years, plus $250,000 in shared computing resources
Oct. 28, 2026 — As artificial intelligence (AI) systems learn to reason, plan and carry out complex tasks with limited human direction, the University of California San Diego is launching a new interdisciplinary center focused on how humans can understand, evaluate and trust these AI ‘agents.’
The Center for Principled Agentic Intelligence (CPAi) is a partnership between the UC San Diego Halıcıoğlu School of Data Science and Computing and the School of Arts and Humanities. Bringing together researchers across data and computer science, high-performance computing, arts and humanities, the center will study the broader questions advanced AI systems raise about human values, oversight and accountability — and determine when trust in them is warranted.
The center will organize its work around three principles: Alignment, trust and responsibility. Alignment asks whether an AI system behaves as its developers and users intend. Trust requires verifiable evidence that systems perform reliably. Responsibility addresses the accountability of the people and organizations that build and deploy AI systems and the effects those systems have on society.
“As artificial intelligence becomes increasingly capable and autonomous, universities have an important role to play in ensuring that its development serves people and society,” said UC San Diego Chancellor Pradeep K. Khosla. “By bringing together technical expertise with deep inquiry into human values and responsibility, this new center reflects UC San Diego’s commitment to helping shape the future of AI — not simply responding to it.”
“UC San Diego helped shape the intellectual and technical foundations of modern artificial intelligence,” said Rajesh K. Gupta, dean of the Halıcıoğlu School of Data Science and Computing. “The Center for Principled Agentic Intelligence builds on that legacy by bringing together researchers who can examine not only what increasingly capable AI systems can do, but how we understand them, evaluate them, and ensure they remain responsive to human goals.”
Unlike conventional AI tools designed primarily to respond to individual prompts, agentic AI systems can plan and execute a series of actions to accomplish a larger goal. As these systems become more capable and autonomous, researchers face increasingly complex questions about how to test their performance, understand their internal processes and maintain meaningful human oversight.
The Center will initially focus on two major initiatives:
AI Alignment and Society
Co-led by Mikhail Belkin of the Halıcıoğlu Data Science Institute (HDSI) and Eddy Keming Chen of the Department of Philosophy and HDSI, the AI Alignment and Society initiative will examine the ways AI systems are trained and evaluated affect their behavior — and how researchers can determine whether increasingly capable systems continue to act in accordance with human intentions.
Researchers will investigate questions including how alignment can be evaluated when an AI system may outperform those evaluating it, how claims about system safety and alignment can be rigorously tested and which training practices may introduce the greatest risks.
“Questions about artificial intelligence are also questions about knowledge, agency, creativity, meaning and the future of human society,” added Dean of the School of Arts and Humanities Cristina Della Coletta. “The humanities are essential to answering these questions and to ensuring that AI evolves in ways that reflect our values and serve the common good. This partnership brings humanistic and technical expertise together to help shape a future in which innovation is not only powerful, but responsible and worthy of our trust.”
The initiative will also explore the societal implications of increasingly powerful AI systems, including how meaningful human oversight can be maintained as systems grow more capable and what forms of independent evaluation may be needed as AI becomes more deeply embedded in society.
Agentic AI System Evaluation
A second initiative, co-led by Tauhidur Rahman, Lily Weng and Haojian Jin of the Halıcıoğlu Data Science Institute and Rose Yu of the Department of Computer Science and Engineering at the UC San Diego Jacobs School of Engineering, will develop new methods to independently evaluate the reliability and performance of agentic AI systems.
The team plans to develop an open-nd why an AI agent fails at a task, operates within defined boundaries, produces behavior that can be audited or performs reliably across complex, multi-step tasks
The initiative ultimately aims to make rigorous AI evaluation tools broadly available to researchers and other organizations rather than limiting such capabilities to the companies developing the systems themselves.
Building research and education around responsible AI
At launch, CPAi is supported by $1 million over three years across its two research initiatives, and will bring together UC San Diego faculty, postdoctoral researchers, visiting scholars and collaborators from other universities and industry. In addition, the Halıcıoğlu School will provide participating researchers with computing resources on a shared cluster valued at $250,000.
About the Halıcıoğlu School of Data Science and Computing
UC San Diego’s Halıcıoğlu School of Data Science and Computing brings together computing, data and artificial intelligence to address some of today’s most complex challenges. The Halıcıoğlu School builds on the combined strengths of the Halıcıoğlu Data Science Institute, and San Diego Supercomputer Center, creating an interdisciplinary environment for research and education spanning data science, artificial intelligence, advanced computing and neural computation.
About the School of Arts and Humanities
The UC San Diego School of Arts and Humanities brings together scholars, artists and students who explore the ideas, cultures and creative practices that shape our world. Encompassing the departments of History, Literature, Music, Philosophy, Theatre and Dance, and Visual Arts, the school fosters a culture of experimentation and interdisciplinary collaboration where traditional forms of inquiry meet emerging technologies and new ways of thinking. Through scholarship, creative practice and public engagement, students are encouraged to redefine conventional ways of knowing, explore how technology is transforming our world and imagine new possibilities for the future.
Learn more about research and education at UC San Diego in: Artificial Intelligence
Read more news about: School of Arts and Humanities,
Artificial Intelligence
