Wow. What a fun evening! I can’t think of a better way to spend a Wednesday night than sitting with some of the top Data Science minds in Seattle & listening to Oren Etzioni do his part lecture, part stand-up routine in a gorgeous setting, replete with Northlake Pizza and Talking Rain. Was I dreaming? Was this for real? Yes! In this case, Oren was the featured speaker at the first ever SIGKDD Seattle Chapter Meetup last Wednesday, October 29th. My Velocity co-founder, Jennifer VeVea and I accepted Oren’s invitation to join him at Ai2’s breathtaking offices on South Lake Union in Seattle as his special guests. Oren gave a fantastic talk about the world-changing work going on at Ai2. Here are my layperson’s impressions; keep in mind I’m a technology enthusiast, but I am not a technologist.
- Ai2’s tagline is “Artificial Intelligence for the Common Good” and I really get what this means. This is an intention to improve AI research, development, resources, encourage openness in the community, and enable individual and collective exponential leaps forward.
- The culture and approach at Ai2 is completely open and open source. No protection of IP, no focus being given to hiding or protecting information, learnings or data. (This info was generated in response to a direct question from the audience). In fact, quite the opposite is true. Ai2 is using Scala exclusively to enable open development environment. Ai2 built their own retargetable parser – this will actually be the first Ai2 release of an open source share (ETA about 6 months). Ai2 is on monthly release cycles, and dedicated to creating real value in near term – no 30 year dev cycles for this crew! Ai2 is an entrepreneurial environment, in fact there are some incubator rooms at Ai2 for exploration of ideas that might lead to the next Ai commercialization such as the next Siri, etc.
- ARISTO: Oren’s opening gambit was “Deep learning is successful but shallow”. I took this to mean that Deep Learning is about to be turned on its head by the work at Ai2. Oren invited hecklers to comment, and got one or two which he handily deflected with some levity. Apparently, according to Oren, “Big Data” does not deliver reasoned explanations; Project Aristo is solving this through extraction and inference (an under-representation of the distinctions, but this is what I could comprehend). I think I heard Oren predict that Aristo will achieve 80% accuracy on 4th grade texts within the next year. That seems stunning to me.
- LEVAN means “Learning Everything About Anything” – it’s a visual concept learning program using machine vision – highly cutting edge and an area of intense focus and 2015. Take note: talent acquisition will be a focus in this discipline at Ai2 over the coming year.
- SEMANTIC SCHOLAR (S2): is a high powered scholarly research service that delivers highly relevant results that can be parsed & filtered by topics, dates, citations (and visually represented). Oren gave a demo: I thought “this is Google Scholar on steroids”; I’m sure that’s an understatement.
Logistically, the Meetup was perfectly executed, thanks in large part to the amazing staff at Ai2, ACM and SIGKDD. Kudos to the Seattle Chapter organizers: Ying Li, Johannes Gehrke, Raghu Ramakrishnan, Bing Liu, and Zarina Strakhan. http://sigkddseattle.acm.org/. If you are looking to join a community of Data Scientists, Statisticians, Machine Learning experts/researchers and practitioners from both industry and academia, these are your people! Upcoming speakers, including Carlos Guestrin, CEO, GraphLab and Prof. @ UW, Josepha Sirosh, CVP of Machine Learning, Microsoft;Ai2 RecruitingFlier PDFAi2 RecruitingFlier PDF Roger Barga, Director, Amazon, and several other highly distinguished experts from various Data Science fields. If you are interested in sponsoring one of these talks, the chapter would love to hear from you.