Harvard Wants To Build Artificial Intelligence As Smart As You

For now, you're still better at recognition and interpretation.

In many ways, artificial intelligence is already very advanced. There are machines that can take more than 90 percent of the work humans used to do in certain factories. Facebook CEO Mark Zuckerberg's goal for 2016 is to build his own personal AI assistant. Fellow tech billionaire Elon Musk is even so worried about the destructive potential of self-aware machines that he signed a letter calling for a ban on autonomous weapons and military robots. However, there's still plenty of ground AI systems need to cover to match the human brain in key areas of analysis and learning. That's a gap researchers at Harvard aim to close soon.

The university has received a $28 million award from the Intelligence Advanced Projects Activity (IARPA), which it will use to understand how AI can be faster, smarter, and one day as complex as the human brain. While some computers are comparable to our brains in storage capacity — neuroscientists estimate we can hold somewhere between 10 to 100 terabytes of information — they're all far behind in mimicking our ability to quickly recognize patterns, analyze data, and retain specific knowledge.

With this in mind, Harvard's John A. Paulson School of Engineering and Applied Sciences (SEAS), Center for Brain Studies (CBS), and Department of Molecular and Cellular Biology will work in conjunction to study and record the activity of the brain's visual cortex. This should give them a better understanding of how neurons connect and interact with each other, ideally mapping a road to building AI that is equally complex.

"This is a moonshot challenge," project leader David Cox, assistant professor of molecular and cellular biology and computer science, told Wired. "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."

The second half would be replicating their findings into a functioning AI system that can learn and interpret information as good or better than humans. Naturally, the word "better" may sound somewhat alarming given the unknown ramifications of reaching the technological singularity, the point at which AI's computing capabilities surpasses that of the human brain. However, the point isn't to build a robot that's smarter than humans and set it loose to do whatever it wants. This $28 million is supposed to go towards the study of the brain at its most fundamental level, and applying the findings to creating technologies that can advance our race, not hinder it. If everything goes according to plan, it should be a win for all.

Cover image: Shutterstock


Subscribe to our newsletter and get the latest news and exclusive updates.