SoftBank is betting billions on the coming of robot supremacy and artificial intelligence in the next 30 years
According to a paradox described by Austrian researcher Hans Moravec, what is complex for us is simple for machines, while what is easy for us is complicated for them. And so they beat our chess masters and our Go masters and prove themselves incapable of walking up stairs, or even getting up after a fall. But Moravec’s paradox is quickly growing out of date. Small, bipedal robots can now navigate up and down stairs, transport heavy loads, and avoid all sorts of obstacles. Taller humanoids, like Atlas, navigate as well on snow as on rubble. The quadruped robot Cheetah is so fast that it outran Usain Bolt’s speed record. As for the two-wheeled robot Handle, it can jump 4-foot-high obstacles.
Just a few years after their buyout by Google–becoming Alphabet in 2015–the companies that designed and built these startling machines, Boston Dynamics and Schaft, are passing from the hands of the American tech giant to those of the Japanese telecommunication giant SoftBank. The total sum of the deal, announced June 9, was not specified. “Robotics as a field has huge potential, and we’re happy to see Boston Dynamics and Schaft join the SoftBank team to continue contributing to the next generation of robots,” was all an Aphabet spokesman had to say. But why did the California company trade away such “huge potential?”
“The decision is a mystery to a lot of people,” says Nell Watson, professor of robotics and artificial intelligence (AI) at Singularity University.
In 2013, Android software inventor Andy Rubin created an internal division at Google dedicated to robots. He called it “Replicant,” after the humanoids from Blade Runner. Google then bought out eight startups, including Boston Dynamics. The company, which was founded at MIT in 1992, remained largely independent. Its headquarters stayed in Boston, more than 3,000 miles from Google’s Silicon Valley office, a mileage that certainly didn’t help Boston Dynamics find its place in its new home. Or to forget its lack of profitability. In 25 years, the company had barely sold any of its machines, getting by only on the backs of Army contracts.
And even these weren’t always fruitful. In December 2015, the Marines refused to work with dog-bots produced by Boston Dynamics, arguing that they were too loud, and their alternate versions weren’t loud enough. That same year, Alphabet buried “Replicant” in its mysterious innovation lab Google X. The arrival of Ruth Porat, the former financial director of Morgan Stanley, brought an end to its more uncertain and eccentric projects. Rumor got around that Boston Dynamics was up for sale. Alphabet was taking a hard look at its priorities, and it decided to focus on self-driving cars, which could produce actual vehicles in three years.
Even if that meant letting SoftBank, which had now taken over the French company that produced the robot Nao, become a titan in robotics.
Christened Aldebaran in 2005, in homage to a star in the Taurus constellation, the company that created Nao took on a somewhat less romantic—though perhaps more representative—name in May 2016: SoftBank Robotics. Nao, a two-foot-tall humanoid, found fame in August 2007 when it was chosen to succeed Sony’s quadruped as RoboCup’s official robot. RoboCup is a soccer tournament played by robots and their designers—students of the best robotics schools in the world. Its several interactive capabilities and appealing face are particularly entertaining for children. One might imagine it living a simple life as a company robot, but it’s primarily used as a teaching tool in elementary and high schools. Nao is worth thousands of dollars, after all.
In March 2012, Aldebaran presented a prototype of Romeo, a 4’7″ humanoid created to aid the elderly and infirm. Made with round, comforting shapes, it’s tall enough to open doors, take objects off tables, and even carry human beings. Romeo was created with limited power, though, with the intention of making it somewhat safer to use. It can carry on brief conversations thanks to information gathered from the Internet, and its social-interactive capacities are now being sharpened in various European labs.
In June 2014, it was Pepper’s turn. At 3’11”, it was as tall as a ten-year-old child. It could show empathy and be polite. It began conversations with a friendly, “Hello, how are you?” and prepared subsequent responses. For positive reactions, Pepper expressed happiness with gestures. For negative, it tried to comfort its conversant: “Don’t worry, it’ll be better tomorrow.” Pepper could even simulate basic emotional needs: “Hold me, please.” But like Romeo, it didn’t exactly look like a human being. Its voice had a metallic timbre, a straight face right out of science fiction. “We weren’t looking for Pepper to look like a human being,” explained Steve Carlin, head of the American division of SoftBank Robotics, to Ulyces last January. “I don’t want that, personally. We only want to make you forget it’s a robot long enough that it becomes an ideal companion.”
Once combined with the impressive motor skills of Boston Dynamics’ machines, the friendliness of robots like Pepper, Romeo, and Nao might make SoftBank the world leader of robotics. According to a study published by Business Consulting Group in June 2016, it should be worth $90 billion by 2025. “In the span of a single year, from 2014 to 2015, private investment in robotics has tripled,” its authors write. “This increased interest was aided by lower prices, by rapid progress in capacities, and by certain elements being applicable in a vast number of industries and services, as many observers have predicted.” These artificially intelligent machines will play an increasingly important role in our future society, says Nell Watson.
The CEO of SoftBank, Masayoshi Son, seems to share this opinion, having invested in more than just robotics.
At the beginning of the 1980’s, SoftBank was, as its name still indicates, a mere software vendor. Then the company invested massively in media, the Internet, and mobile phones. These investments proved lucrative, especially one in a Chinese e-commerce site called Alibaba. And yet, Masayoshi Son did not hesitate to sell SoftBank’s shares in 2016, worth more than $8 billion, in order to acquire British microchip specialist ARM, worth $33 billion. With an ever-greater number of transistors, microchips will soon overtake the capacities of the human brain, Son says to justify the astronomical expense. “One of the chips in our shoe will be smarter than our brain!” he said in a speech to Mobile World Congress in Barcelona last February. “Today, 99% of chips in your smartphones were designed by us,” he added. “Tomorrow, it’ll be the same for the Internet of Things.”
To stay ahead of the competition, Masayoshi Son created an investment fund dedicated to technologies like quantum computing. Based on the principles of quantum mechanics, quantum computing vastly increases computing power. And here as with robotics, Google was angling to be a pioneer. In 2015, the company came out with a quantum computing prototype 100 million times faster than modern computers. But IBM and Microsoft weren’t far behind. Google would now have to be wary of SoftBank.
Masayoshi Son’s “Vision Fund,” created to bankroll his ambitious plans, will close in five months. It has already pulled in $98 billion in commissions from prestigious investors like the Taiwanese Foxconn, the Japanese Sharp, Qualcomm, and Apple, as well as the Saudi Crown Prince Mohammed ben Salmane, who alone is worth $46 billion. In short, there’s a good chance the SoftBank CEO’s grand bet will pay off. For an investment fund to rake in $100 billion is unheard of. That’s as much as every venture capital dollar spent on startups in the entire world in 2016.
That, along with other opportunities, allows Masayoshi Son to invest more than $520 million in Improbable Worlds, a British virtual reality startup. But it doesn’t change the fact that SoftBank is built upon massive lendings. “I’m the king of debt,” Masayoshi has claimed multiple times. Last year, his debt rose to $124 billion, ringing alarms at the financial analyst company Moody’s. Unlike Google, which finances its own technological risks with ad revenue, SoftBank relies entirely on other people’s money. Which means if the giant falls, it’s gonna fall hard.
So why is Masayoshi Son taking the risk? “Because I have a vision. And I believe in the singularity.”
“In mathematical terms, a singularity is the point where exponential growth becomes infinite,” Nell Watson explains. And in the tech world, the singularity is the point in which AI programs are no longer controlled by human beings but by AI itself, ushering in a tidal wave of progress. Called the “technological singularity,” it was first described in the 1950’s, but it was made famous by mathematician and science fiction writer Vernor Vinge in the 1980’s and 90’s. From a scientific point of view, it’s predictable by Moore’s law, which states that processing power follows an exponential increase over time.
Some see the singularity as an opportunity for humanity. The cofounder of Singularity University, Ray Kurzweil, declared that he “couldn’t wait for it.” Others, more numerous, don’t hide their fear. Physician Stephen Hawking thinks “the development of full artificial intelligence could spell the end of the human race.”
“I am in the camp that is concerned about super-intelligence,” said Bill Gates. For Elon Musk, CEO of Tesla and SpaceX, developing it now would be “summoning the demon.” He’s concerned enough about the technological singularity that his new company, Neuralink, is attempting to increase our cognitive ability enough to create a brain-computer interface capable of countering the rising power of artificial intelligence.
“We are about eight to 10 years away from this being usable by people with no disability,” assured Elon Musk. “It is important to note that this depends heavily on regulatory approval timing and how well our devices work on people with disabilities.”
The question of time carries substantial weight, given the singularity’s creeping approach. Ray Kurzweil, who claims an 86% success rate in his predictions, pegs it in 2047. Masayoshi Son is less specific, but sets it on more or less the same horizon of “the next 30 years.” Nell Watson won’t say herself: “It’s an open question.” The AI and robotics specialist isn’t even sure it’ll happen at all.
And she’s not alone. Many experts contesting the technological singularity theory criticize its advocates of failing to account for the limitations of progress, such as finite available energy resources.
For Jean-Gabriel Ganascia, computer science professor at University Pierre and Marie Curie in Paris, the technological singularity is a “myth.” For Margaret Ann Boden, cognitive science professor at the University of Sussex in England, a “science fiction nightmare.” “Given [the] trend, it is not surprising that some people foresee a point known as the ‘Singularity,’ when AI systems will exceed human intelligence, by intelligently improving themselves,” she writes. “At that point, whether it is in 2030 or at the end of this century, the robots will truly have taken over, and AI will consign war, poverty, disease, and even death to the past.
“To all of this, I say: Dream on. Artificial general intelligence (AGI) is still a pipe dream. It’s simply too difficult to master. And while it may be achieved one of these days, it is certainly not in our foreseeable future.”
Let’s hope for Masayoshi Son, his creditors, and his investors’ sake that she’s wrong. As for the rest of us, let’s hope she’s not.