A shift toward greener AI and richer virtual worlds took center stage as Cornell Tech researchers unveiled 23 papers at NeurIPS 2025. The conference, held December 2–7 in San Diego, remains one of the field’s premier gatherings for artificial intelligence, machine learning, and related disciplines. Each year NeurIPS attracts thousands of scholars and industry leaders, serving as a spotlight for advances in AI, computational neuroscience, statistics, and large-scale modeling.
At this year’s event, Cornell Tech pushed the envelope across several domains. Highlights include efforts to protect data privacy, raise the bar for AI evaluation, and improve the speed and efficiency of expansive language models. Beyond these, the team introduced tools for evaluating environmental and health interventions, methods to align visual imagery with architectural designs, and techniques for generating highly realistic 3D scenes with remarkable efficiency. Collectively, these developments hold significant promise for fields ranging from public health and robotics to urban planning and immersive media.
For more details, see the Cornell Tech news release about NeurIPS coverage.
