NEURO-SYMBOLIC AI

Research and Education

“Centaur AI” systems refer to the idea (going back to the 1950s) of AI systems designed to collaborate in a deeply symbiotic way with humans. The Institute strives to narrow the large current gap between today’s AI systems and this goal. Today’s dominant AI systems don’t provide any fine-grained way for humans to understand and modify their innards, contributing to them in an essential way with their knowledge and preferences, rather than just selecting unknowably large datasets, hitting “go”, and hoping for the best. Further, the dominant mechanisms in those innards are alien-seeming and simplistic compared to human psychology and human ways to reason and solve sophisticated problems. These mismatches between AI and human stand in the way of the potential for AI to reliably help humans solve their problems, great and small. We help to lead a growing movement around neuro-symbolic AI to develop the next generation of AI concepts and tools.

Can we make

machines that serve

humans

b e t t e r ?

HIGHLIGHTS AND NEWS

  • Recent 3rd Neuro-Symbolic AI Summer School (NSSS ‘24), Sep 4-6, 2024. Featured a curation of star speakers and participation by thousands of researchers, students, and data scientists. Register to access the videos (to be made available shortly).

Graduate students

researchers

and AI practitioners:

Strive for more

Are you…

  • Curious about why “neuro-symbolic AI” seems to be appearing increasingly in paper titles, funding calls for proposals, academic job ads, and startup announcements?

  • Someone fundamentally not only interested in AGI (Artificial General Intelligence) but in a serious set of paths to get there?

  • Looking for an alternative to the mainstream set of ideas in AI?

Develop your knowledge and skills. Join us as we help each other through the struggle of the cross-disciplinary learning needed to be part of the next wave of AI.

Are you…

  • A PhD student interested in neuro-symbolic AI or AGI, particularly something ambitious, fundamental, and/or multidisciplinary, but need a co-advisor or mentor?

  • A researcher coming from outside AI (e.g. a cognitive scientist, linguist, physicist, information theorist, logician, category theorist, other mathematician, neuroscientist) and looking for collaboration on a multi-disciplinary or foundational topic related to our research?

  • Someone who has been called a 10x programmer? Winner of programming competitions?

Help us create the path forward. Join our research efforts to create the new AI foundations and techniques needed to:

  • Reason reliably and transparently

  • Safeguard against hallucinations and dangerous behavior

  • Learn to generalize to novel situations efficiently