From WEF19 to AAAI19: Reflections on the Way

While on my way from the beautiful snows of Davos to the warmth at AAAI19 in Honolulu, I’ve been reflecting about events and discussions at the 2019 World Economic Forum (WEF19). I found that energy and passion were high at WEF19. I especially enjoyed my 1:1 conversations with leaders from industry, government, academia, and civil society.

Advances in AI and their influences seemed to pervade conversations and presentations at WEF19. It was inspiring to see the great enthusiasm about advances in AI. However, it seemed that many folks' understandings and expectations about AI technologies have been rooted in tech press writings, popular books, and science fiction. I found quite a contrast between expectations about AI capabilities voiced at sessions and discussions at WEF and the sensibilities that permeate technical AI meetings like AAAI19. AI scientists and engineers have continued to slog through hard and important AI challenge problems. Numerous fundamental challenges have been resistant to solution, even after decades of hard and insightful research.

 On expectations, regardless of the overheating around the capabilities and rate of progress of AI, we’re clearly at a promising inflection point with pattern recognition, prediction, and natural language understanding. I see the biggest wins for organizations, and society more broadly, as hinging on making effective investments to integrate available AI competencies into rich, complex workflows, processes, and decision making. The latter includes efforts (1) to identify good matches between available technologies and the operations and goals of enterprises and (2) work to broaden target applications beyond relatively simple, “modular” uses of classification and prediction so as to support richer decision making. In many cases, this will require careful consideration of multiple uncertainties, the cost-benefit structure of outcomes, representations of human preferences, and designs that enable effective human-AI collaboration. It’s also critically important to continue to be vigilant and proactive about potential rough edges, inadvertent influences, and potential for abuses rising with advances in perception, reasoning, and new forms of automation enabled by AI. It’s been a privilege to work closely with colleagues at Microsoft, the Partnership on AI, and other organizations on responsibilities with the development and fielding of AI technologies. Work on the responsible fielding of AI spans the technical and policy realms—and their admixture.

People are talking quite a bit these days about competition among nation states for AI supremacy. I believe that leadership in AI, whether for organizations or countries, will not be based on competition around the core sciences of AI technologies. Scientific advances are shared broadly and quickly throughout the world and there's very little chance that a big breakthrough will happen in a secret lab. Leadership and competitive edge will hinge on becoming great at mastering the challenges of identifying valuable applications and working to deeply integrate the technologies into workflows and processes, and identifying how to best use the capabilities to support human decision making.

 It was good to have a chance to share some of these thoughts on harnessing AI advances at several events at WEF19, including at the New York Times Roundtable on the Future of Work and the ETH panel on Man and Machine. Video of the ETH event will be available.

 This post will sync with the LinkedIn graph when I land in Oahu and head to the AAAI meeting, where our team from Microsoft Research will be sharing exciting new directions and results, with deep relevance to fielding #AI in the open world. When I made “AI in the Open World” the theme of my 2008-9 presidency of the Association for the Advancement of AI (AAAI, the host of AAAI19), and presented on the technical and societal challenges and opportunities ahead, I didn’t foresee the rise of such a rich, energetic scholarly field around that statement. I’m truly excited about the multiple trajectories forward for the science, and for the possibilities that the advances will open up to everyone.