Arjun Balaji's Recent LinkedIn Posts

Thrilled to share that our paper “Encoding Values: Injecting Morality into Machines via Prompt-Conditioned Moral Frames” has been accepted to NeurIPS 2025 Creative AI Track! 🎉 Our work explores how diverse moral frameworks can steer large language models, benchmarking their behavior under varied ethical “constitutions” across moral dilemmas, everyday interactions, policy questions, and rewriting tasks to reveal how prompt-level guidance shapes their reasoning and responses. An enormous thank you to my brilliant collaborators Jyothika K Raju, Neha Kamath, and Aarushi Nema - their insight, dedication, and creativity shaped every stage of this project, and this work simply would not exist without their tireless contributions and shared vision. We’ll soon open-source the full framework, datasets, and paper so others can explore and build on this work.

I’m thrilled to share that I’ve joined Columbia | SIPA to pursue my Master of Public Administration (MPA)! At SIPA, I’m focusing on how AI, data, and public policy can come together to create equitable and sustainable impact at scale. It’s an incredible opportunity to learn from world-class professors, engage with global policy leaders, and collaborate with peers who are passionate about solving real-world challenges. I’m also deeply grateful to have received offers from the University of Chicago and Carnegie Mellon University. Excited for this next chapter in New York City and all that lies ahead!