GT Sophy wasn't a winner without speed. The program was able to set lap times on three different virtual courses. GT Sophy lost when Sony tested it against multiple human drivers. At times the program was too aggressive, but at other times it was too timid.
Sony regrouped and prepared for a second match in October. GT Sophy won easily. The difference was made by something. Sony came back with a bigger neural network, which gave it more capabilities to draw from on the fly. GT Sophy was given the ability to balance aggression and timidity in order to pick the most appropriate behavior.
This is the reason why GT Sophy is relevant. The kind of dynamic, context aware behavior that robots will be expected to have when interacting with people is an example of howiquette between drivers on a track is a specific example.
An awareness of when to take risks and when to play it safe would be helpful for artificial intelligence that is better at interacting with people.
"I don't think we've learned general principles yet about how to deal with human norm that you have to respect." It is a start, and hopefully will give us some insight into the problem.
GT Sophy is the latest in a line of artificial intelligence systems that have defeated the world's best human players at a number of games. Sony had a new type of challenge. Gran Turismo requires its best players to control a vehicle at the limits of what is physically possible, in real time, and in close proximity with other players all trying to do the same.
All Rights Reserved © 2024
