Solving Hard Problems in Science and Education with Game-based Symbiosis of Humans and Computers

Solving Hard Problems in Science and Education with Game-based Symbiosis of Humans and Computers

This talk will demonstrate the surprising problem-solving power of the game-based, human-computer machine. To be successful, this new problem solving architecture needs to achieve two goals: 1) the game players need to be transformed into domain experts through the process of game play, and 2) the game framework needs to automatically adapt to become the most effective tool for human problem solving. Both of these goals present formidable computer science challenges ranging from adaptive engagement maximization for all people, to analytics based self-adaptation of the game experience. We will describe our first such platform, Foldit which has already resulted in two Nature papers, demonstrated outperformance of this approach to purely computational methods as well as a newly designed synthetic protein confirmed in the laboratory experiments. This new genre of scientific discovery games, presents a unique set of challenges orthogonal to the standard algorithmic computer science challenges as well as the standard game design challenges.

The second part of the talk will expand on perhaps the most powerful lesson learned from this framework is that it is possible to achieve world-class expertise in a specific domain through deliberate prolonged game play. Our more recent efforts try to generalize this finding and apply it to education in general. Specifically, we focus on turning education enterprise into a data driven science through use of interactive games that optimally evolve towards each individual student. I will describe some of the great machine learning problems that arise from this framework and the impact this technology could have on millions of student worldwide.