Duality

Washington State University Assistant Professor Matthew E. Taylor joins us and discusses the resourcing of carbon-based and digital assets, "right now there's lots of well-paid, well trained technicians who can go and set up your data center for you, but they're humans. They're only human. They're going to be suboptimal. Instead, you can have a reinforcement learning algorithm go and tweak things so that you're more efficient and you can save lots of money. You can also save lots of energy, which is better for the environment."

Ben-Gurion University of the Negev Professor Kobi Gal joins us and shares that to fully realize the potential of artificial intelligence, we first have to truly understand human intelligence and emotion, "people exhibit varying degrees of motivation. And a big part of my work is understanding what makes each student tick, and personalize. So the sequencing problem, what's important is to be able to personalize the question to choose for the student. And of course, what's the right question for me might not be the right question for you."

University of Massachussetts Professor Shlomo Zilberstein joins us and takes us through a history of artificial intelligence. "The quest for AI is started very early with the first computer was being built. And actually with Touring, where you were just thinking about the possibility of electronic computing– people immediately felt this machine could be a thinking machine. Asking, 'What could it do that is better than people? What can you do that is as good as people?' That foundational work actually benefited computer science as a whole."

Georgia Tech University Professor Charles Isbell, Brown University Professor Michael Littman and University of Texas Professor Peter Stone discuss and debate the value of human/AI interaction. "A lot of people have been working for many, many, many, many, many years on the problem of leveraging humans and leveraging human behavior to make machine learning better. We talk about it all the time- humans in the loop. But it's always fun to take a moment, step back and ask whether all of your assumptions are actually valid and real- and here the assumption is that humans actually can help us but it's not clear that that's true."

Professor at UC Berkeley, Peiter Abeel joins us. Pieter grew up in Belgium, came to the US and got his PhD in Robotics and Machine Learning from Stanford. He notes that he and Andrew Ng pushed the envelop at the time on how robots learn from humans demonstrations as well as their own trial and error. Peter graduated and came to Berkeley to continue to work on the junction of robotics and learning, machine learning. He’s been focused on end-to-end reinforcement learning, end-to-end imitation learning. Training the neural net end-to-end without specific structure. Singularity is the notion that a system you build is smart enough to self improve...and things accelerate out of control. How far are we away from this? Pieter notes that 10 years ago computer vision it was difficult to conceive of a solution. Enabling factors and breakthroughs are the keys. Data is an enabling factor. Neural nets are now data driven as opposed to algorithm designed. Will we continue to have more data and can you do things with unl