Tuesday, June 10, 2014

Masayoshi Son's other reason for making emotional robots

Last week, SoftBank unveiled the Pepper emotional robot to the world, to much surprise. But did you know that SoftBank CEO Masayoshi Son already hinted to his dream to build robots with artificial intelligence and emotions, way back in 2010?

From SoftBank's "Next 30-Year Vision" 2010
In his 30-year vision, Son described his view of the future, and justified the need for artificial emotions. He imagined that the future would bring brain-like computers that would learn by themselves. Instead of being pre-programmed, these computers would have that special capacity to learn constantly, an ability that even the lowliest animals have, from bumblebees to chimpanzees.

He warned, though, that artificial intelligence purely driven by goals could have its consequences.

He gave an example of a robot-driving-AI developed at Honda Research. The robot car was designed to learn how to run the circuit in the most efficient, fastest way. Then, by accident, it ran into one of the barriers and smashed through the track walls...

"The car made a mistake once and bumped into a wall. Then, it broke through the wall, made a turn, and ended up finishing the course quicker. In later runs, that artificially intelligent car started hitting the walls deliberately."
He concluded,

"When we follow our instincts to simply satisfy our desire, that's when it becomes dangerous. That's where savagery comes from."
What happens when computational knowledge surpasses that of humans? he asked.

 "If we give only computers intelligence greater than that of a human, and give them a goal that they can only use their intelligence to achieve, then we'll see them doing the same thing that the car earlier did: breaking free to reach their goal in the shortest distance possible. So, to stop them from breaking free recklessly, I believe it's necessary for computers to have the highest of emotions. Namely, the richness, the kindness, of love for people. This will help computer brains have control."
So, while Son announced last week his passion for kind, gentle robots, he also envisions that emotions can be used to control AI.

Watch the full presentation here

Other memorable quotes:
@1:08:00 Brain-like computer chips can be used with moving motors. When we combine one of these chips with motors in that way, we get robots. Those artificially intelligent robots could be sent to dangerous places like disaster sites after earthquakes or fires. Snake-like robots could be used to slide into tiny spaces to call out "Hey, anyone alive in here?" Then they can communicate with the person in need until help comes." Or maybe we could have robots with the strength of an excavator, but not just robots that dig, but talk.

@1:09:26 In housework and medicine too we will see robots that are both smart and gentle. We don't want doctors to perform surgery on us just to mechanically say “Hang in there.” when we yell “This hurts!” We want artificially intelligent doctors that are kind. In that way, we could have robots that are gentle towards people but also have incredible intelligence and strength. Robots of various sizes and shapes will appear. 

@1:10:25 The best robot companies won’t necessarily be the ones that make the best robot “muscles” in the manner of a car manufacturer. Components such as motors will be easily made in a factory. The most difficult part will be the part that tells the muscles what to do. The part with intelligence. The part that does the thinking and feeling. SoftBank will continue to follow this idea with the information revolution.

@1:12:30 Intelligent, gentle robots would not be harmful, they’d probably be more gentle than humans. With them we can probably create an even better society together. We at SoftBank want to make that happen. We are not science fiction makers. We do not write novels. We want to make it a reality. We want them to be gentle, feeling, and through the information revolution, we want to help everyone. We want to spread the use of brain computers to make people happy.

@1:13:20 In 300 years there won’t just be one robot or appliance with a brain computer in each house. There will be 10 or 100. Maybe we will work together with these robots to achieve things that humanity has never experienced. To solve problems that have been beyond our abilities. Maybe together we will be able to cure unknown viruses, combat terrorism, protect people against meteorites and great earthquakes and towering tsunami. Together maybe we can solve these problems that humanity has not yet succeeded at doing alone.

No comments: