Harvard will take a close look at the brain to build better AIs
Three departments are getting $28 million for the project.
There's no AI that can learn as fast as the human brain -- at least not yet -- but Intelligence Advanced Research Projects Activity (IARPA) wants to change that. The government organization has granted three departments within Harvard University a total of $28 million dollars to find out why our brains are so darn good at learning things compared to artificial systems. For instance, we only have to see a car once or a few times to recognize one, but even the most advanced AI has to look at thousands of samples before it can say what it's seeing is a car.
The researchers from Harvard's John A. Paulson School of Engineering and Applied Sciences (SEAS), Center for Brain Science (CBS) and the Department of Molecular and Cellular Biology have been tasked to record the activities going on inside the brain's visual cortex. They have to find out how the neurons there are connected to each other. Then, they have to use the data they gather to find a way to build better artificial intelligence systems.
Harvard assistant professor David Cox said:
This is a moonshot challenge, akin to the Human Genome Project in scope. The scientific value of recording the activity of so many neurons and mapping their connections alone is enormous, but that is only the first half of the project. As we figure out the fundamental principles governing how the brain learns, it's not hard to imagine that we'll eventually be able to design computer systems that can match, or even outperform, humans.
Clearly, this is a very ambitious project and the team members "have no illusions that this will be easy." Since the whole process is expected to generate over a petabyte (or 1.6 million CDs) worth of data, the researchers believe it could lead to other advances in computing, as well. After all, as the research progresses, they will have to think up new ways to manage data and to speed up its processing.
