Artificial intelligence researchers in Washington State University's School of Electrical Engineering and Computer Science have developed technology that gives a computer the ability to give advice and teach skills to another computer in a way that mimics how a real teacher and student might interact.

Artificial Intelligence Professor  Matthew E. Taylor says the agents – virtual robots – act like true student and teacher pairs: student agents struggled to learn Pac-Man and a version of the StarCraft video game but the student agent learned the games and, in fact, surpassed the teacher.

While it may sound like fun and games, creating robots that can teach each other is not easy but the potential is enormous. If robots can teach each other tasks, then a housecleaning robot can teach its younger, sexier replacement how you prefer the toilet paper placed on the roll.


So will robots teach each other how to take over the world? 

Not these ones. "They're very dumb,'' says Taylor,  and even the most advanced robots can be easily confused. When they get confused, they stop working and it often takes multiple efforts to get a robot to work at all. The easiest way to successfully teach a robot new skills is to remove the "brains" of the old one and put them in the new one. Problems occur, though, when hardware and software don't work in the new model.

Want to give it a try yourself? Here are their instructions for how to download it and get it set up.

Matthew Taylor. Credit: University of Washington

And if we want robots to teach humans - which will bypass the social engineering and other issues of current matriculation - incompetent robots can be fired - we have to rely on something beyond inserting their hard drives.

The artificial intelligence researchers programmed their teaching agent to focus on action advice, or telling a student when to act. As anyone with teenagers knows, the trick is in knowing when the robot should give advice. If it gives no advice, the robot is not teaching. But if it always gives advice, the student gets annoyed and doesn't learn to outperform the teacher.

"We designed algorithms for advice giving, and we are trying to figure out when our advice makes the biggest difference,'' Taylor says.

He aims to develop a curriculum for the agents that starts with simple work and builds to more complex. 


Citation: Reinforcement Learning Agents Providing Advice in Complex Video Games by Matthew E. Taylor, Nicholas Carboni, Anestis Fachantidis, Ioannis Vlahavas and Lisa Torrey, Connection Science.
Source: Washington State University