In case you haven't heard, Doug Lenat died yesterday. He was a hero of mine (and to many others in the field), whose dedication and perseverance was laudable. He was also a great friend, whose warmth and amazing sense of humor I treasured. People in AI often don't give the Cyc project the respect it deserves. Whether or not you agree with an approach, understanding what has happened in different lines of work is important. The Cyc project was the first demonstration that symbolic representations and reasoning could scale to capture significant portions of commonsense. While today billion-fact knowledge bases are common in industry, there were claims in the 1980s that symbolic techniques could never scale to even a million facts. The same changes in scale that have helped statistical and neural methods have also helped symbolic methods (And hybrids too of course). There are still things the AI community can learn from the Cyc project. Cyc remains the most advanced in terms of expressiveness, capturing more of the range of thoughts that humans are capable of. My group has been using Cyc’s representations in our research for decades, sometimes within ResearchCyc, but more often with knowledge extracted from their system and used with our own reasoner. Not that their reasoner was bad: Far from it, it had capabilities that I still envy. But we are exploring a different path (analogy-centric) and need to be able to distribute source code and knowledge bases, which is why we built our own reasoning engine and NextKB. NextKB uses OpenCyc derived knowledge, combined with FrameNet material, a large open-license lexicon, and of course our own work on qualitative and analogical reasoning and learning. Starting with OpenCyc was a fantastic accelerator: Many subtleties of representation and reasoning had already been worked out saving us from reinventing many, many wheels. There are people who are using Cyc-derived knowledge every day and have no idea that they are doing so. Sketch worksheets, one of our fielded educational software systems, uses NextKB as its knowledge base. It provides the representational vocabulary used in analogical matching to provide feedback, based on qualitative visual relationships CogSketch automatically computes. I am sure there are many other projects who could benefit from building on the Cyc ontology. Our field would do well to learn more from the Cyc project. #ai #knowledgegraphs #KRR #logic #logicalreasoning #analogy
Very sad indeed. An underappreciated scientist.
Very sad news. Unfortunately many students, even those in the AI/Data Science stream, are unaware of Doug Lenat's contribution to knowledge representation and the underlying complexities. Lex Friedman did a podcast with Doug (https://youtu.be/3wMKoSRbGVs?si=NNre-N6OPL5yKyax) that explains Doug's passion for the field. Well worth listening to.
So sad to hear.I've been saying for a long time that if I could have lunch with any person in the world, it would be Doug Lenat! I once watched a documentary about Doug and Cyc and the Cyc team and I was so excited and inspired by the story - he also seemed to be a lovely person! Ever since I've often told people about Doug and Cyc, including students and professionals in AI and computer science, and it was always new information for them (how can you lack curiosity about the development of your own field?!). It's really a loss in many ways.
I'm so sorry to hear of his death. Reading his work certainly changed my life. I decided to start out in AI in 1991 specifically after reading about AM and Eurisko. I can still recall how enchanted I was by several of his ideas, especially that of using general heuristics to drive exploration of a design space.
I totally agree with you. He was a great person. CyC was, is and will be one of the greatest achievements in AI. Far apart from all marketing and the actual noise.
I remember speaking with him years ago while riding an airport shuttle together from a conference. He was so passionate about his work and there are so many wonderfully useful applications. At the time, he was using Cyc to reason about changes for regulatory compliance in the financial industry. Thank you, Dr. Lenat, for your important contributions.
Cyc was a visionary project. Tackled "common sense" head-on. We'll be toasting the harvest from Lenat's field for generations :-D
Glad you posted something really nice about his work. After you told me about Lenat, I read some pretty long articles about his work. Very sad news about him, but I am delighted to see that his contributions were being recognized.
Thanks for writing about Doug Lenat and cyc. Early in my career your approach to education with symbolic reasoning has also influenced me immensely to think about non vector approaches to intelligence.
Ken Forbus so what is the present state of the Cyc project and yours ? is there a way to do some evaluations ?