Artificial intelligence has moved into the commercial mainstream thanks to the growing prowess of machine learning algorithms that enable computers to train themselves to do things like drive cars, control robots or automate decision-making. But as AI starts handling sensitive tasks, such as helping pick which prisoners get bail, policy makers are insisting that computer scientists offer assurances that automated systems have been designed to minimize, if not completely avoid, unwanted outcomes such as excessive risk or racial and gender bias. A team led by researchers at Stanford and the University of Massachusetts Amherst published a paper Nov. 22 in Science suggesting how to provide such assurances. The paper outlines a new technique that translates a fuzzy goal, such as avoiding gender bias, into the precise mathematical criteria that would allow a machine-learning algorithm to train an AI application to avoid that behavior. "We want to advance AI that respects the values of its human users and justifies the trust we place in autonomous systems," said Emma Brunskill, an assistant professor of computer science at Stanford and senior author of the paper.

Artificial intelligence has moved into the commercial mainstream thanks to the growing prowess of machine learning algorithms that enable computers to train themselves to do things like drive cars, control robots or automate decision-making. Go to the web site to view the video. As robots, self-driving cars and other intelligent machines weave AI into everyday life, a new way of designing algorithms can help machine-learning developers build in safeguards against specific, undesirable outcomes like racial and gender bias, to help earn societal trust. But as AI starts handling sensitive tasks, such as helping pick which prisoners get bail, policy makers are insisting that computer scientists offer assurances that automated systems have been designed to minimize, if not completely avoid, unwanted outcomes such as excessive risk or racial and gender bias. A team led by researchers at Stanford and the University of Massachusetts Amherst published a paper Nov. 22 in Science suggesting how to provide such assurances.

In recent years, artificial intelligence has struggled with a major PR problem: whether or not it's intentional, developers keep programming biases into their systems, creating algorithms that reflect the same prejudiced perspectives common in society. That's why it's intriguing that engineers from MIT and Harvard University say they've developed an algorithm that can scrub the bias from AI -- like sensitivity training for algorithms. The tool audits algorithms for biases and helps re-train them to behave more equitably, according to new research presented this week at the Conference on Artificial Intelligence, Ethics and Society. And even then, once complex AI systems deploy in the real world, it becomes very difficult to evaluate how exactly they're making their decisions. That's why automating the process is so important -- the new tool can go in and reconfigure how much value the AI system gives to each aspect of its training data, according to the research.