Vicarious, the mysterious company that's been funded by Tesla and SpaceX CEO Elon Musk, Facebook's Mark Zuckerberg, and actor Ashton Kutcher, wants to do something completely radical — build the world's first human-level artificial intelligence (AI).
"Vicarious is building a single, unified system that will eventually be generally intelligent like a human," co-founder and computer scientist Scott Phoenix wrote in a World Economic Forum Q and A.
Right now, we have AI that's very good at narrow tasks, like playing chess or Jeopardy. But computer scientists still have yet to produce a computer that can do as many things as a human can, as well as a human can do them.
A computer that has human-level intelligence wouldn't just be a breakthrough in AI, it would answer one of the most fundamental questions in science — how do you build an intelligent machine? It would also completely change society — imagine being surrounded by artificial beings as smart as you at work, at home, and at the grocery store.
In 2014, Musk invested in Vicarious, which was founded in 2010 by Phoenix and neuroscientist Dileep George, during a $40 million funding round. Phoenix told Bloomberg that Vicarious and their funders are in it for the long-haul — their general purpose AI probably won't be ready for several years, and they don't expect to make much money on top of the $70 million they've raised until then.
"We're fortunate to have the freedom to take a 10-plus-year time horizon," Phoenix told Bloomberg.
Most researchers think building a human-level AI will likely take longer than one decade. Philosopher Nick Bostrom surveyed 550 AI researchers to gauge when they think human-level AI would be possible. The researchers responded that there is a 50% chance that it will be possible between 2040 and 2050, and a 90% chance that it will be built by 2075.
While Vicarious isn't forthcoming about their timeline, they want to build human-level AI as soon as possible. They're doing this by building an AI that emulates how the brain works — specifically the neocortex, the area of that brain that's responsible for perception and information processing, and they're making incremental progress.
Related story
In November 2013, Vicarious said that it had built algorithms that could see and solve CAPTCHAs, the codes used by secure websites to filter out robots from genuine human users.
Vicarious - Turing Test 1: Captcha from Vicarious Inc on Vimeo.
But they still have a long way to go before they can build a computer that can do more than one very specific task and also learns to improve on its own. Phoenix wrote they first have to figure out how to build a computer that matches the immense processing power the human brain uses everyday.
They also have to figure out how the human brain creates intelligence, which largely remains a mystery to neuroscientists and computer scientists alike. So Vicarious is trying to build a "mathematical model of the human brain that enables our systems to learn how to solve problems the way a person would."
They're starting with vision. Phoenix wrote in the World Economic Forum post that vision and perception are "the gateway[s] to higher reasoning."
"What seems like abstract thought is often stimulated by perceptual ability," he wrote in the World Economic Forum. "Suppose I tell you 'John has hammered a nail into the wall,' then ask you 'is the nail horizontal or vertical?' It might seem at first glance like a logic problem, but actually your ability to answer it comes from the mental image you've imagined of John hammering in the nail."
Humans would naturally imagine the nail being hammered in horizontally, say, to hang a picture. But a computer wouldn't logically think of why John hammered the nail into the wall, and could picture him hammering the nail into the top of the wall.
Now if Vicarious could build a system that could understand the scenarios and questions like that, they might well be on their way.
