 These are virtual humans or embodied conversational agents and they kind of look like video game characters or maybe bots of some kind, but they're very different than your usual video game character because they're based on hundreds of thousands of hours of human-human interaction data analyzed at one-thirtieth of a second granularity, a lot of undergraduates tied to computers analyzing these things. Because to build a virtual human is like building a human from the bottom up. You need to know everything about every eyebrow raise, every chin movement, every eye, every movement of the torso, everything is said and the interaction amongst all of those modalities. It's something that teaches you what you don't know and as Steve said, the best way to look at human-computer interaction or human-machine interaction is iteratively. We start by looking at human-human data and then we build a model or a theory of that interaction. That model is the basis for algorithms that make up a virtual human because you can program a virtual human to do things that at least Carnegie Mellon won't let me do with real humans. So we can study things about human behavior with virtual humans that we can't study with real humans. And the insights we gain are really extraordinary. For example, that bodies entrained to one another over a period of 20 minutes in a conversation, you'll start to move the way I do. And these are insights that are difficult for psychologists to make but easy when you have a model of a human. But as I said before, they also teach you what you don't know about humans. And so the theory building is iterative. Sarah is not a copy of a person. She's a copy of the commonalities of hundreds of people who've spent up to five weeks working with one another once a week. And that theory, we hope, is an increasingly powerful theory to change artificial intelligence. One of the things we learned was that people become less polite with others over time and that that impoliteness increases the bond between people if a smile follows it within a 30th of a second. If no smile follows the insult, rapport or relationship goes down. We've learned other things with our virtual children. We've learned that children's gains in science are greater when they work with someone or something that speaks their own native dialect, even if that's a dialect that teachers don't value. And this is something that was never so clear as when it was studied that way. But let's come back to artificial intelligence because the core goal of Sarah is to change the face of artificial intelligence. My virtual humans are never going to go to war. They're never going to replace workers on an assembly line because they make no sense without people. They are bodies that depend on other bodies. And I would argue that bodies, of course, allow us to interact with our physical surroundings, but that they also are the basis of social interaction and that that's what makes us human. And therefore, paradoxically, it's in developing a virtual human like this one that we better understand what it means to be a real human and we are more likely to retain what it means to be a real human in this increasingly digital world. And that's very important to me. So Sarah is socially aware and she allows us to learn things about social awareness in people and in artificial intelligence. Other people have tried this with virtual reality. This is an experiment that found that when your virtual reality body is shorter than you, your self-esteem goes down. It's amazing, isn't it? It's truly amazing work. But not enough artificial intelligence researchers are using these kinds of social embodied agents and therefore the future could have robots that take our jobs, that don't cooperate with us. And so I would push you to think about the negatives and the positives of a future with virtual bodies. That if those virtual bodies, those virtual humans are social, collaborative and only work in collaboration with us, then those hundreds of thousands of hours of study that result in systems that can make a future where social interaction is maintained and where we retain what it is to be human can be ours. Thank you.