What Can Artificial Intelligence Teach Us about Learning?

If you really want to learn something, the first thing you need to learn is how to fail. That might sound odd, but it's the result of some serious research by a team from the University of Arizona. The group ran a series of machine learning experiments intended to teach computers some basic tasks, such as identifying odd and even numbers and categorizing patterns.

They found that the computers learned at an optimal rate when they failed about 15 percent of the time. According to one member of the team, psychologist Robert Wilson, the studies support previous animal testing that ascertained a certain difficulty level is required for learning.

"If you are taking classes that are too easy and acing them all the time, then you probably aren't getting as much out of a class as someone who's struggling but managing to keep up," he said. According to the research, succeeding at a rate of 85 percent signifies not only the retention of information but also the prompt to keep trying. While the researchers stressed that the tests only covered the basics of learning, not higher-level abstract thought, they also suggested that the results could help point the way forward for many areas of learning, from classrooms to training labs.

Learning about learning:

• Studies have shown that children develop their language, motor, and sensory skills by engaging with nature.
• On average, Chinese students do 14 hours of homework a week -- the most of any students in the world.
• Finnish children don't start school until the age of 7, while children in Holland start on the day they turn 4.