Mark Zuckerberg is hoping to build something that only exists in science fiction and comic books.
The Facebook CEO revealed in early January that his personal challenge for 2016 was to build a simple artificial intelligence to help run his home and assist with his work.
On Wednesday, Zuckerberg took to his Facebook page to share an update of his progress and the limitations of AI, at least for now.
"Simply put, today's AI is good at recognizing patterns and bad at what we would call common sense," he wrote.
Zuckerberg, who plans to train his AI to recognize his voice and face, will implement a training method called supervised learning. By showing an AI system thousands of images it can begin to decipher patterns and use that information to diagnose diseases, drive a car or play games.
"But there are lots of limitations of this approach," Zuckerberg writes. "For one, to teach a person something new, you typically don't need to tell them about it thousands of times. So the state of the art in AI is still much slower than how we learn."
In order to teach an AI system to anticipate and react to new situations, the Facebook CEO suggests a method called unsupervised learning. The same way that animals and humans observe and test boundaries, this method could help AIs to learn how to hold conversations or perform complex actions.
But, artificial intelligence has a long way to go before that will happen, says Zuckerberg. Scientists do not yet know how general learning works and how best to train AI systems to function autonomously.
To those who see this new form of robotics as a concern — Stephen Hawking and Bill Gates — Zuckerberg says not to worry.
"We should not be afraid of AI," he said. "Instead, we should hope for the amazing amount of good it will do in the world. It will saves lives by diagnosing diseases and driving us around more safely. It will enable breakthroughs by helping us find new planets and understand Earth's climate. It will help in areas we haven't even thought of today."