Artificial intelligence (AI) researchers are working hard to make computers smarter and more capable, some in hopes of achieving human-level intelligence.
The surest strategy might seem like emulating human intelligence or even mimicking brain structure.
But creating an intelligent machine based on the human mind might be impossible, unnecessary, or even counterproductive.
In fact, many researchers believe what’s inside our heads is just one example of how to achieve intelligence — and not necessarily the best way.
“An aeroplane flies in a way that’s very different from the way a bird flies, but they both fly — it’s the same thing with intelligence,” Peter Stone, an AI researcher and computer scientist at the University of Texas, Austin, told Tech Insider.
Cornell University’s Bart Selman, also an AI researcher, used speech recognition as an example to make a similar point.
“We don’t quite know how the brain does it,” Selman told Tech Insider, but the method “is probably more complicated than the way we’re doing it right now” with machines. “The main progress right now and in the near future will be getting to a performance at a human-level without getting the details of the human brain all figured out.”
Some computer systems do take some inspiration from the human brain, namely the interconnected structure of neurons, to form artificial neural networks. Such systems can be “trained” on a specific task like image recognition with many thousands of examples. Over time, the devices improve at the task.
Machine learning and neural networks have driven a lot of the recent successes in computer science, says Geoffrey Hinton, an AI researcher at Google and the University of Toronto.
But in many cases, it’s best not to mimic human intelligence or brain structures. While humans are better than machines at things like perception, reasoning, and object manipulation, robots already exceed human intelligence along very narrow dimensions. Machines have faster reactions times, perfect memories, and are superior to humans at crunching a lot of numbers.
“In some cases it would be unintelligent to mimic how the human brain works,” Peter Norvig, the director of research at Google, told Tech Insider in an email. “If the task is to multiply two ten-digit numbers, then using a human brain alone would be an error-prone mistake — it would be more intelligent to either reason like a computer to start with, or to use a tool, such as a calculator or computer.”
Shimon Whiteson, a computer scientist at University of Amsterdam, told Tech Insider there’s another, more practical reason to build AI that functions differently than the brain.
“There’s actually not much point in just replicating human intelligence,” Whiteson says. “We want intelligence that has different capabilities than humans so the two can work together in a complementary way.”
Business Insider Emails & Alerts
Site highlights each day to your inbox.