Georgia Tech Student Group Tackles AI Safety Through Research and Policy
A student-run group at Georgia Tech is positioning itself at the center of AI safety research, moving beyond theoretical discussions to conduct projects that inform policy and practice. The AI Safety Initiative (AISI), founded in 2022, has grown to include more than 70 students in its fellowship program each semester, with members publishing 13 papers at top conferences over the past two years.
The group focuses on three core activities: hosting talks from AI researchers, running fellowship programs that immerse students in safety research through reading and discussion, and supporting independent projects that contribute directly to the field. Past work includes demonstrating large language model security risks on Capitol Hill, responding to federal requests for information, and conducting war games for Georgia Tech Research Institute faculty.
From Students to Researchers
AISI's structure-entirely student-run with faculty advisors-allows the group to pursue research that traditional academic labs struggle to fund. This arrangement has attracted faculty collaboration. Glenn Matlin, a computer science Ph.D. student, credits AISI with advancing his own AI safety work and calls it "like a third lab" for recruiting and collaborating with undergraduate researchers.
Rocio Perales Valdes, AISI co-director and second-year computer science student, frames the challenge directly: "AI introduces new kinds of challenges into our legal and societal frameworks. Its capabilities emerge fast and on a jagged, hard-to-predict edge, which leaves AI governance like chasing a moving target."
Yixiong Hao, the other co-director, sees the group as a lever for moving the field forward. "There is a rapidly growing gap between what AI systems can do and what we understand about them; yet mitigating AI risks is systematically neglected by current market incentives," Hao said. "I have higher leverage in convincing smart people to work on neglected problems in AI safety."
Pipeline to Industry and Policy
AISI alumni have moved into roles at organizations including Anthropic, RAND, the UK AI Security Institute, and the Institute for Progress. The group's co-founder, Gaurav Sett, is now a Ph.D. student at the RAND School of Public Policy.
For students like Ishan Khire, who joined AISI as a first-year, the group provided both community and research opportunity. After attending the fellowship program, Khire began conducting AI research with faculty member Giri Krishnan on protein structure prediction. "Finding a community that cares about AI safety was a big part of joining," Khire said.
Tom Conte, associate dean for Research in Georgia Tech's College of Computing, noted that AISI addresses a real institutional gap. "Any cursory look at the news today will show there is significant angst about AI and whether it is being developed responsibly and with sufficient guardrails in place. AISI has Georgia Tech at the forefront of that conversation."
For researchers seeking to understand how AI systems work and the policy frameworks governing them, Generative AI and LLM Courses and AI Research Courses provide structured entry points into the field.
Your membership also unlocks:
