Student research opportunities
Useful Games
Project Code: CECS_197
This project is available at the following levels:
Honours, Summer Scholar, Masters, PhD
Supervisors:
Dr Mark ReidAssoc Professor Jochen Renz
Outline:
There are many AI tasks that can be easily solved by humans, but that are still very difficult to solve by AI systems. Examples are tasks such a translating documents between different languages or recognizing objects in images or video.
One straightforward way of solving such AI tasks is to let other people solve them for us and at the same time to improve the AI system by observing and learning how humans solve these tasks. Instead of paying humans for helping us, we want them to help us for free and even to enjoy helping us. Since many internet users spend a lot of their time playing games, a good way of getting humans to help us is to embed the task into a game.
In this open-ended project we want to select an AI tasks and represent it as a game that humans can play on the internet, and by playing the game they solve the task for us.
There are a number of interesting open problems that have to be solved in order to reach our goal of improving an AI system, such as finding an adequate representation of the task, developing an interesting game that solves the task, identifying the correct solution out of a number of different solutions, identifying trusted users, learning how users solve problems, etc.
The project is suitable for students at all levels and offers exciting problems to students interested in practical applications or in theoretical research.
