A new study by a team of researchers at UK-based AI company DeepMind suggests that AI can make better public money decisions than humans. For this purpose, DeepMind has developed more capable problem-solving systems known as artificial general intelligence.
According to the research, AI can devise methods of wealth distribution that are more popular than systems designed by people. It also shows that machine learning systems are good at delivering more open-ended social objectives, such as the goal of realizing a prosperous and fair society.
The team trained an AI system to find a popular public fund distribution policy in a four player online game.The AI learned from more than 4000 people and computer simulations . Players voted on their favorite policies for distributing public money. The AI policy won more votes from human players.
Creating a machine that delivers beneficial results humans actually want is defined as value alignment. One problem with value alignment is that human society admits a plurality of views which makes it unclear to whose preferences artificial intelligence should align. For this, the AI discovered a mechanism that redresses the initial wealth imbalance and sanctions-free riders, thereby successfully winning the majority vote.
This new approach by DeepMind researchers combines artificial intelligence with human democratic deliberation to come up with better solutions to social dilemmas, such as public money distribution.