Upcoming Events:30 November 2025 - 7 December 2025
About the Event
Clear your calendars and prepare your academic posters, because the annual pilgrimage for AI researchers is upon us. The Thirty-Ninth Annual Conference on Neural Information Processing Systems, or NeurIPS 2025, is taking a novel approach this year, splitting its considerable intellectual mass between two vibrant locations. The main event unfolds at the San Diego Convention Center from December 2nd to December 7th, while a significant, overlapping programme kicks off earlier at the Hilton Mexico City Reforma from November 30th to December 5th.
This dual-hub format promises a sprawling, week-long immersion into the absolute bleeding edge of machine learning. If you’re an academic, a researcher at a tech behemoth, a grad student hoping to get noticed, or simply a professional who needs to know what’s coming next in the world of AI, this is your intellectual Glastonbury. NeurIPS is where foundational research gets its first major airing before trickling down into the products and platforms that will dominate the next decade.
Attendees can expect a formidable agenda packed with tutorials, workshops, oral presentations, and poster sessions covering every conceivable niche of neural information processing. From the theoretical underpinnings of deep learning to the ethical quandaries of generative models, NeurIPS is the definitive forum for the global AI community to converge, debate, and define the future.
Key Themes
While the full schedule could make a supercomputer blush, several key themes emerge from the organised chaos. Prepare for deep dives into:
- Large Language Models: Naturally. The agenda is saturated with sessions on everything from model merging and mechanistic interpretability to the endless quest for more robust reasoning and less catastrophic nonsense.
- Generative & Multimodal AI: The industry is moving far beyond text. Expect to see the latest in generating and understanding images, video, music, and complex real-world simulations, along with a healthy dose of creative AI applications.
- Reinforcement Learning & Embodied AI: This is where the silicon meets the road. Sessions will explore the path to physical intelligence, with insights from leaders in scalable self-driving, robotics, and multi-agent systems.
- AI Theory & Foundations: For those who prefer their intelligence with a side of mathematical proof. Dive into optimisation theory, the principles of deep neural networks, and the science of what’s measured versus what’s missed in benchmarking.
- AI for Science & Society: Machine learning is being wielded to solve tangible problems, with workshops dedicated to climate change, healthcare, materials science, and ensuring the development of socially responsible and trustworthy foundation models.
Highlighted Speakers
This year’s lineup features some of the heaviest hitters in the field, poised to deliver keynote insights on the state and future of artificial intelligence. Confirmed invited speakers for the main conference include:
- Rich Sutton - The Oak Architecture: A Vision of SuperIntelligence from Experience
- Zeynep Tufekci - Are We Having the Wrong Nightmares About AI?
- Yejin Choi - The Art of (Artificial) Reasoning
- Melanie Mitchell - On the Science of “Alien Intelligences”: Evaluating Cognitive Capabilities in Babies, Animals, and AI
- Kyunghyun Cho - From Benchmarks to Problems - A Perspective on Problem Finding in AI
- Andrew Saxe - Demystifying depth: Principles of learning in deep neural networks
Programme Highlights
The schedule is dense enough to be its own dataset. Here’s a brutally condensed look at what to expect.
Mexico City (Nov 30 - Dec 5)
The Mexico City leg kicks off the festivities with a focus on workshops and community building. Highlights include workshops on “Socially Responsible and Trustworthy Foundation Models,” “Embodied and Safe-Assured Robotic Systems,” and centering low-resource languages in the age of LLMs. Affinity groups like Women in ML, LatinX in AI, and Queer in AI will have a strong presence from the start.
San Diego (Dec 2 - 7)
The main event in San Diego is an absolute torrent of cutting-edge content.
- Tuesday, Dec 2: The conference opens with a full day of deep-dive tutorials on topics like “Human-AI Alignment” and “Planning in the Era of Language Models.” The Expo hall also comes alive with panels on everything from generative e-commerce recommendations to Tesla’s foundational models for robotics.
- Wednesday, Dec 3: The first oral sessions begin, covering language models, generation, and theory. The prestigious Test of Time Award will be presented, followed by poster sessions and creative AI panels.
- Thursday, Dec 4: More oral sessions on reinforcement learning and deep learning, alongside provocative Position Paper Panels tackling “Responsible AI Research” and the “Ecosystem of AI Research.”
- Friday, Dec 5: The week’s main programme wraps up with sessions on multimodal AI, neuroscience, and graph-based models, culminating in the Sejnowski-Hinton Award and a closing reception.
- Saturday, Dec 6 & Sunday, Dec 7: The conference concludes with two full days of highly specific workshops and competitions. Topics range from “AI for non-human animal communication” and “Mechanistic Interpretability” to the “Google Code Golf Championship.”
Registration
Ready to join the world’s top AI minds? Registration is open, and given the event’s popularity, you’d be wise not to procrastinate.
Pricing is tiered for students, academics, and industry professionals, with different packages available for the San Diego conference, the Mexico City workshops, and virtual-only access. Early registration discounts are available until October 12, 2025. Be sure to check the official website for detailed pricing and package options, especially for the distinct Mexico City programme.
