CYC is a long running project to capture and reason about general knowledge. The CYC project was founded in 1984 by Doug Lenat. It uses symbolic AI explicitly representing knowledge and reasoning and it is the largest endeavour of its kind: OpenCYC 4.0, released in 2012, contains around a quarter of a million concepts and over two million facts about them ins a large semantic network. Although it is a general knowledge base, it is divided into microtheories about different topic areas, this structure has enabled some of its successful applications in specific sub-areas. In some ways large-language models capture elements of general knowledge due to the vast text sources they process, but lack the explicit semantics and explainable AI of CYC. In his last paper before his death, Lenat and long-term colleague Gary Marcus, wrote about the ptential of LLMs to learn frm CYC, in a form of neurosymbolic AI AI.
Used on Chap. 2: page 31
Also known as CYC
Links:
Encyclopedia Britanica: CYC- Wikipedia: Cyc
- ojs.aaai.org: CYC: Using Common Sense Knowledge to Overcome Brittleness and Knowledge Acquisition Bottlenecks
- arXiv: Getting from Generative AI to Trustworthy AI: What LLMs might learn from Cyc