Artificial Intelligence: Application of the Prisoner’s Dilemma
Link to source: http://sociable.co/technology/why-we-need-to-develop-ai-to-explain-ai/
Artificial intelligence (AI) is a topic that is widely discussed among technology elites as well as in everyday life. The discovery of artificial intelligence has led to further creations and innovations of applications of AI to perform tasks once only completed manually. Google’s subsidiary, DeepMind, conducted an experiment by having AI go against itself through a game called “prisoner’s dilemma.” In this case, both of the AI players could potentially betray the other in order to win the game; however, if they both decided to betray each other, both players would lose. The experiment resulted in interesting results; scientists found that based on the context the agents adjusted their behavior. With a common enemy, the players would work together and cooperate, but if there were minimal resources, the players went against each other.
This brings about the question of how AI came to its conclusion to either cooperate or betray its fellow player. The article continues to discuss further exploration of these results through Ulsan National Institute of Science and Technology’s (UNIST) new program to use AI in order to increase understanding of it. The program would allow AI to adjust to different fields and make different decisions based on different contexts.
The application of the prisoner’s dilemma theory to AI is interesting and a milestone in the field of technology as it tests the social conflicts that people encounter on a daily basis for manmade intelligence. One of the examples that the article mentioned that AI could assist with throughout the development of the program is with estimating future stock prices. Putting a crucial part of the world’s economy into the decision-making of AI would be a big step; would AI be able to consider the qualitative aspects that could impact the prices of stocks? Would it be able to make decisions that go against one player in order for personal good of a certain group? The prisoner’s dilemma plays a big part in this development of AI; and while it is exciting to see technological developments, I feel that it is also a controversial move that could also call for a lot of criticism.