Artificial intelligence is a bit like a human, who is inside something else. It’s not as smart as you — but it could be as smart as you in the future. I believe that we’ll become robots at some point too. Artificial Intelligence is changing our lives. But what can it really do? What will change, and what will remain science fiction? To answer these questions, we embarked on a journey... to meet the scientists working on our future. Augsburg, in southern Germany: Home to Kuka, the world's leading manufacturer of industrial robots. Rainer Bischoff is head of research here, and is considered to be one of the world's leading experts in this field.
He and his team are working on a new generation of robots that learn independently — like children. The task: to recognize and sort building blocks. This robotic system taught itself how to grab - in other words, there was no human programming the robot. So he tried by himself? He tried by himself, like a child. When he first started grabbing he wasn’t successful except for in 1-2% of cases. But he observed himself. And by observing himself, the robot identified when an image matched a particular grasping motion, and when it didn’t.
He’s applied what he learned and now grabs these objects over 90% of the time. I didn’t program him and yet he’s still learning the task by himself. Seeing that really motivates you. But what if the robot sees a new object, such as pliers? Here we have a nice example: every child would just say, ‘ok, grab and move these pliers over, no problem.’ But he's still failing. He’s failing because he doesn’t know what kind of inertial force this object has to be able to grab it properly. But you can see how he’s already trying out different methods, and in time he may get the hang of it.
I should add that he isn’t trained after every attempt: he collects around 1000 data, and then the neural net is re-trained. It’s possible that if we let him try 1000x, he would at least be able to grab reliably. Intelligent robots that learn by themselves. They can recognize parts, assemble them, and they can independently adapt to their environment with the help of AI. But we’re only in the early development stages... I have a favorite example and that’s chess. These days, there are computers or AI that can beat chess champions.
But we don't have a robot that can reach into a bookcase, take out a chess set, open the box, take out the pieces one by one, set them up and start playing. A 6-year-old can do that — but no existing robot can. So for the moment, whenever I need physical intelligence, we’re still doomed to fail. And I think that will be the case for a few more years. Yet machines are getting better and more intelligent. This video was produced using special effects... ...but this robot has learned how to play table tennis. He was built by researchers in Tübingen and shows how much is already possible in the real world.
How long will it take before robots are better than us in some areas? Robots already are better than us in many areas. Particularly those requiring non-variant repetition, a great deal of force, or a high degree of precision. The tasks current robots are not as good at as we are, are those involving sensors. There’s no point denying that, and I think it will be another 10 or 20 years before we have robots that can hold a candle to humans in some areas. We humans use all of our senses and can do more than smart robots. But the robots are beginning to learn.
Artificial intelligence also plays an important role in a story that began in January 1982, in Mount Washington, New Hampshire. Hugh Herr was 17-years old at the time. Together with his friend Jeff Batzer, Hugh went up a mountain. But they were caught off guard by a change in weather. A blizzard raged for 3 days. The missing boys were only found after 4 days. Alive, but both had severe frostbite. The doctors decided to amputate Hugh's legs just below the knee. 32 years later, Hugh Herr has AI legs, which he developed himself. He spoke on turning disability into opportunity at the TED Conference 2014.
Dancer Adrianne Haslet-Davis lost a leg in 2013, in the terrorist attack at the Boston Marathon. Thanks to the smart prosthesis by Hugh Herr, she can dance again. Boston, home to the Massachusetts Institute of Technology. We met with Hugh Herr to talk about artificial intelligence and the human body. He’s the pioneer in the field of intelligent prosthetics. A single person who is both developer, and user. There are dozens of prototypes in his lab. There’s a motor. This is a synthetic subtalar joint for inversion/eversion.
So we’ve iterated, and spent millions of dollars to arrive at this optimal architecture. Hugh began developing prostheses after his lower limbs had been amputated. His replacement legs became increasingly complex. Now they are AI limbs, with countless sensors, motors and computers. I quickly realized that I had an opportunity. That from my knees down, there was a blank slate. And I could create anything in that space that I could conceive and imagine. So as a young man I started to imagine what that blank space may look like, what may fill that space. Disability depends on perspective.
Hugh Herr has developed a novel answer: With the special prostheses that he developed himself, he can once again pursue his greatest passion: climbing. So there’s a computer in here? There’s three, actually. Three?! They’re each the size of your thumbnail. So very small microprocessors. And there’s a muscle-tendon-like motor-system. So the computer runs algorithms and receives sensory information. The device is measuring position, speed, accelerations, temperature and what not.
All that information goes into the computer. The computer runs its algorithms and then decides on the action of the muscle-tendon-like motor-system. This all happens very fast. So as I am walking, going up and down hills and steps, it’s constantly responding to my biomechanical needs. It is so good that even nowadays you do mountain climbing — you go climbing. Absolutely, and I run. You cannot with a straight face say that I am disabled. I trail run, I play tennis, I mountain climb, whatever I want to do physically. Now if you remove the technology from my body, I'm severely disabled, I'm crippled.
But with the technology, in this sophisticated human-machine interaction, I am freed from the shackles of disability. Are intelligent prostheses only the beginning? Will technology increasingly merge with the human body? Intelligent humanoids have already been depicted in films such as ‘Ex Machina’. You shouldn’t trust Nathan. You shouldn’t trust anything he says. We’re closing the loop between the synthetic robotic limb and the human brain, the human nervous system.
And what that means is, the person can think, send descending commands down through the nerves, we measure these commands, and they control synthetic motors on the bionic limb. And then we’re also closing the loop so that sensors in the bionic limb will put information into the nervous system. So the person can feel the bionic limb moving: its position, its sensations - as if it were part of the body. This is almost philosophical, because you have the body, and the machine.
And you, sort of, start merging them together. Yeah, we’re gaining evidence that when a human being can feel a synthetic body part, when they can touch it and it feels like normal touch, when they move it and it feels like normal joint movement — that the synthetic object becomes part of their body, of their identity, their self. What’s cool about having significant parts of your body be designable and synthetic, is you can upgrade. So given that I’m an MIT professor, I’m upgraded every week, software and hardware. That’s interesting. I grow older: no upgrade.
And you can get new. Right. The synthetic part of my body is improving in time. My biological body is degenerating, which is very peculiar. For Hugh Herr artificial intelligence is a blessing. By the time our interview was over, a snowstorm was raging in Boston. An interesting coincidence — as this was also how Hugh’s transition began. Thanks to AI, body and machine are slowly merging. Artificial intelligence also increasingly determines our communication. It’s there, behind every search, hidden from view. In social networks, intelligent algorithms control what we see, and thus influence what we read, and what we don't.
But there’s a problem: Fake News Capturing and re-selling our attention and our digital data has become ‘big business.’ Information technology firms are among the most valuable companies in the world. Facebook, YouTube and Twitter have changed the media worldwide. But exactly what role do their intelligent algorithms play in the spread of fake news? In 2018, a team of scientists from Boston analyzed the spread of fake news. The study was led by Professor Sinan Aral. It was the largest worldwide study that had ever been conducted on the spread of fake news on social networks.
While information is abundant, attention is scarce. So there’s way more information than we can process. And so these platforms help us by curating this information, and, as you said, prioritizing what comes 1st in our newsfeed, what comes 2nd, what comes 3rd. And they have a machine, an algorithm based on machine learning that is deciding what gets shown 1st, 2nd, 3rd; or in fact, what gets shown at all. Some things are not shown. Not every piece of information is shown to everyone.
But which criteria do Facebook and Twitter use to program their algorithms? The incentives for writing those algorithms are based on the platforms' incentives, the companies that they work for. Those companies are based on an economic model of engagement. The more people are engaged, the more opportunities you have to show ads, and so you have more inventory for advertisements. But the second important reason is that the more people are engaged, the more you learn about who they are, and what they like, and the more sophisticated the targeting is in terms of advertising.
Engagement is key for economic successes of social media industrial complex. And daily Internet usage is increasing worldwide: In 2018 in Germany, the overall average was over 3h/day, for younger people it was just under 6h/day. Things that are exciting, novel, surprising, that are potentially shocking are more likely to be engaging, clicked on, read, viewed, shared liked. And therefore, elements of the models that determine the newsfeed favor engagement. The following case from Japan shows what fake news in social networks can lead to. Videos of women who allegedly became ill from cervical cancer vaccine were posted.
At the same time, unverified ‘scientific’ studies were circulated on social networks. Both videos and studies were picked up by television, In Japan, the vaccination rate against cervical cancer fell from 70% to under 1%. How could it be that false information could turn an industrialized country against a globally recognized vaccination? Hamburg — where we meet Riko Muranaka. The doctor had tried to counteract the anti-vaccine hysteria, and inform the public online. But then she was targeted. I was personally attacked in the Twitter or social media when I started writing about the safety of the vaccines.
They tried to threaten me by sending those blackmail messages to my family or me. Riko continued undeterred. She analyzed the vaccination opponents’ facts, checked the scientific validity of their experiments, and published her results. Afterwards, I was just stormed by the criticism. One day I just decided to shut my Twitter account for a while. But when I got the John Maddox Prize, it became a Twitter trend in Japan. But even that didn’t change public opinion in Japan. Despite the most respected scientists sharing Riko’s view, she eventually lost the battle to fake news.
They accused me because my writing is wrong, and my writing is giving wrong impact to the society, and I am hiding the truth. But it’s the contrary: I'm telling the truth, people feel I'm hiding the truth. It’s really interesting, isn’t it? The WHO sees the anti-vaccine movement as a global health threat. In Japan, around 3000 women will probably die every year from cervical cancer, because they chose not to get vaccinated. Fake news can be fatal. The false information is moving through human society digitally like lightning, while the truth is like molasses, dripping very slowly from person, to person.
The spread of false information, shown in orange, and correct information, in blue. Sinan Aral has studied these patterns on Twitter more closely than anyone. False news travelled further, faster, deeper and more broadly than the truth in every category of information that we studied. Sometimes by an order of magnitude difference. And this was particularly true of false political news. Which was the most viral category of any type of news that we studied. Fake news! We are fighting the fake news.
As you say: fake news. Fake news has changed the political climate - worldwide. Social networks with intelligent algorithms are increasing division in society. They vie for our attention, feeding us exactly the information we like. What counts are click rates, quotas and length of stay - and not whether content is true or trustworthy. This personalized communication is dividing our society. Social networks assign each user a profile, based on what they click, read and watch. Those belonging to the red group here are mainly supplied with information that matches the red profile.
Thus a filter bubble is gradually formed. Everyone lives within their own network. Our opinion is echoed by like-minded people. Contradictory information and opinions hardly enter our bubbles. Media should be a mirror of society. But AI algorithms distort the opinions we form based off our media consumption. Yet media are too important to be left to people out to make money. How will artificial intelligence change conflicts? What about intelligent autonomous weapons? The military is already testing prototypes, like here in California: two fighter jets launch a swarm of intelligent drones.
The autonomous flying objects then identify their own targets. Should machines be allowed to take life-or-death decisions? We travelled to meet one of the top ethicists on autonomous weapons in the US. He warns of uncontrollable development and is committed to a worldwide ban on autonomous weapons. We visited Yale professor Wendell Gelding in his house north of New York. Sometimes people do not fully understand what lethal autonomous weapon systems are. They tend to think of drones that might have facial recognition software, and would pick up a terrorist at a distance, or perhaps a few robotic soldiers on the battlefield.
What is sometimes not fully appreciated: lethal autonomy is not a weapon system. It is feature sets that can be added to any weapons system. And that includes atomic weapons or high powered munitions. And the feature sets would be the ability to pick a target and destroy that target with little or no active human intervention. Intelligent image recognition, automatic target recognition - these AI techniques are already available. And global armament race has begun. That machines do not make life and death decisions about humans, humans make life and death decisions about humans.
And when we open this door to machines making those decisions, we undermine the basic principle of a responsible human agent. Lethal autonomous weapons and self-driving cars are just the tip of the iceberg, with something much larger below the surface. And that larger thing is autonomy in general, autonomous systems in general. Autonomous systems threaten to undermine the foundational principle of agency. And that agent can be a human or a corporation.
But that there is an agent who is responsible, and potentially culpable or liable for any actions that are taken. I cannot imagine of anything more stupid than humanity going down a route where we have diluted the principle of responsibility. Where we dilute it in such a way that nobody can be held responsible anymore if something truly dire takes place. In the past, we have been too slow to recognize we were going down a wrong path.
We need a worldwide ban on autonomous, intelligent weapons! Artificial intelligence will revolutionize industry. In Germany, the term Industry 4.0 has become a buzzword. Cars, robotic tools and entire production plants are being linked via sensors and equipped with AI. But how will German companies fare in worldwide competition? Dr. Michael Bolle heads the Bosch Research Center, in Renningen, Baden-Wurttemberg. Artificial intelligence is one of the main focus points here.
When it comes to industrial AI, the AI that plays a role in products, then I think that the technology companies that have decades of experience in the physical world, in real-life objects, and the corresponding experience in development and production, have a competitive advantage when adding in machine learning and AI. They have an advantage over companies that come purely from the virtual world. So I’m confident. And this is the reason why we're investing so much in this area. And why we're rolling out and applying this expertise across the group.
Germany has faith in its decades-long technical expertise together with AI. One player who is fighting to get ahead is China. Changing of the guard at the Gate of Heavenly Peace in Beijing. Soldiers, flags, Mao. This was the old image of the country; but modern-day China has awakened. Digitalization and AI promise a brave new world. An entire nation seems intoxicated by its own progress. Where does this palpable euphoria about the future stem from? I meet Hong Yang. She is Chinese and has worked for a German company for years. I ask her what is different in China.
Culturally speaking we are different. In the traditional thinking we are more open to the latest technology and open to the world. Probably you can see from how much we using the smartphone: like just now, we buy the coffee, and pay the taxi bill with the smartphone. Sometimes my German colleagues are astonished to say that: you don’t even have to bring cash with you. And I said, yeah, that’s normal life. I always forget my wallet when I’m in Germany, because here in China, I pay for everything with my smartphone.
If you go to the market and there’s an 80-year-old woman selling produce, you might think: “I guess I’ll pay with cash.” But you can’t anymore. You’ll be buying an apple, and she’ll take out a QR code, scan it, and then you pay for it with your smartphone. It’s unbelievable. There’s nothing like that in Germany. It’s crazy. For example, if I have dinner with my friends, and we all first have to hand out our phones, and we put phones on the table. And then, if somebody is picking up the phone — by calls, or text message or WeChat, he has to pay for the bill. That’s the punishment.
So we can feel the advantages of the technology, but when you get used to it, You start to reflect: what kind of impact to my life, what’s good, and what’s bad. And then in terms of the bad part, I mean everyone will start to think about ‘How I can get rid of the bad part’, but still trying to keep the good part. Young China is catching up, and the whole nation is hungry for progress. What’s so special about China? The speed. Just think about where China was 40 years ago. Now, things are going full-throttle. China has even surpassed the US when it comes to filing AI patents.
One example is the mobile transportation service DIDI. Active in 400 Chinese cities, the platform organizes some 30 million trips a day. Each Didi vehicle is equipped with a data logger, which registers whether the car is stationary or moving, or whether there’s a traffic jam. The data is also used to improve information on traffic flow in big cities. We call this Real-Time Traffic Information, or RTTI. The Didi data go directly into the RTTI, which gives you a much more reliable view of whether roads are congested or not than in German cities.
Connecting everything with everything else, and generating added value from that — the Chinese are really, really good at that. China is a much younger nation, there’s energy and enthusiasm about what's to come. There are a lot of investments. There are great education institutions here, Tsinghua University in Beijing and Tongji University in Shanghai are top level. There are really many capable people. The talent pool here is huge. So I think there’s a good chance that China will be leading. It’s predicted that in 2030, 37% of all scientists will be Chinese.
By comparison, 1.4% will be from Germany. And while Germany is currently facing a lack of science teachers at schools, a young generation of innovators is growing up in China. They have top-level education, fresh ideas, and they’re hungry for success. For example the DJI company from Shenzen. It was founded in 2006 by a young engineer. Today, it’s the world's largest manufacturer of civilian drones. Chief Development Officer Martin Brandenburg shows us the latest the model. The new drone uses intelligent image recognition and can independently pursue its target.
Equipped with dozens of sensors and smart positioning, it can detect obstacles, such as trees or bushes. The tree saved my live! The tree was the limit, because the drone says no. In this case, the drone said ‘I can't fly through it safely, so I'd better stop.’ As I understand it, it’s very common here in China to combine things: facial recognition, navigating, flying... Yes, and in this case only visuals were used. You don’t have a Bluetooth transmitter or anything else on you. The pilot simply says: ‘I want to follow Ranga’ on the display, confirms and then the drone follows you.
Innovation ‘Made in China’. Chinese television proudly reports such successes. China no longer copies — modern-day China invents. DJI is truly the first global brand with a completely new product range from China. Before, China was primarily known as the world’s factory. That's changing now, and perception is also changing. Our company alone employs almost 3000 engineers. Really smart people: They’re all enthusiastic; they're motivated and want to create something new. And that’s the spirit that prevails in this country. That’s going to persist.
A hunger for innovation is what defines modern China. And China's economy is booming. Cities such as Shenzen, Chengdu or Guanchu have the same economic output as entire European countries. The country is investing in its young people. Take the example of Robomaster. Here, a team is preparing for the upcoming season. Each team has to program and optimize a gaming robot. The final is a nationwide event. Troy Qin supervises the competition and shows us the parkour. This one is this year’s- the new- the latest one. And also we have a- It doesn’t shoot right now? It doesn’t shoot right now. As we can see, there is the panel right here.
The reference system is for sensing the bullet. When you strike it, it flashes. That means you hit the panel and it decreases the health. If one robot loses all of its points it will shut down. It may look like a game, but it’s actually a program to support young engineers. It was launched by DJI and now several other companies are also involved. The next generation of engineers needs to be good at designing and programming — and that’s exactly what they are learning here in a playful context.
The background is serious engineering or more a game? It’s serious engineering. Because you need to be able to build a whole and new robot. Putting them together is just the first step. Then you need to do the coding, and do some artificial recognitions. It’s quite a massive work and it’s not that simple for students. How many universities are...? There’s 32 universities in the final tournament. But for this year we have 170 universities from all over the world that have registered for the competition. And how many are Chinese? About 140. And how good are they?
We shall see. OK - maybe can get hands on the robots and experience how it works. OK - we have two robots, so we’ll try and have a look at it. Troy needs just a few key combinations to control the robot. ... here W for forward and S for backward. I'm an old guy you know. And now you see me. And now you fire at me? No, I fire at you? You can fire me. But we are teammates, right? We don’t fire at teammates! The final is a major event. 20,000 people are in the audience. Another 30 million watch the contest online. Engineers and programmers are China's new pop stars.
They really care about this game. If a team from their school wins the glory from the game, that means a lot to them. The students invested thousands of hours into developing their robots. The winner of the 2018 competition was from Southern China University of Technology. All of China celebrated their success. There are also excellent initiatives in Germany: One example is the so-called Ideen-Expo in Hannover. Around 300,000 students come here in a single week, Europe's largest classroom. Young people are introduced to new tech, learn how to program, and design new circuits — and they seem to love it. But in contrast to China, German media hardly take note.
Germany talks too little about its successes. We, too, can keep up with the global AI race - if we want to - but we really have to step on the gas! Children are now growing up with smartphones, internet and intelligent toys. But what does that do to children? We met scientist Stefania Druga to discuss. Following research in Boston, she is currently working in Berlin. The generation after the internet generation is the AI generation. They are growing up surrounded by AI ... ...like Alexa, for example: This device was not designed for children. This device was designed for households, so families make purchases via this device.
I think that it’s very important to recognize when we talk about kids, and about regulating smart toys or devices that record data about our kids, to understand who made this device and what was the ultimate goal of this device. Just imagine you that in some years. You might also have children. Would you install Alexa in your home? It depends on what Alexa, the platform, would look like then; where I would live. Right now? Right now? No! Alexa, how many seconds are there in a year? A calendar year has 31,536,000 seconds and a leap year has 31,622,400 seconds.
That's not what I wanted to know. Unlike Alexa, the small robot COSMO was developed for children. Trust and intelligence are related, because if I think that a device is smart, I tend to trust it more. So the younger children were not sure how smart the devices were. The older children thought they were smart because they have a lot of data. So basically the children who are 3 and a half, 4-6, were more skeptical initially of these devices. But as soon as kids would go to school they were more trusting just because they saw how much information these devices have.
Curiosity, creativity, imagination: Children are open-minded, and enjoy trying things out. Some devices react as if they were artificial living beings. After the children learned how to program it and train it, both the young and the old children became more skeptical and trusted the device less. They understood it knows how to answer this type of question, not this type. Do you trust machines? Are they smart? What can you do that they can't? Stefania calls for increased understanding of artificial intelligence.
The goal of my research is to bring this AI literacy both to kids and parents, they're there in the home, parents are also part of the conversation, asking questions. I think it is important for families to understand how AI works, in order to make good use of this technology. Some things promote our own creativity, others seem to be superfluous. And: there is another important difference between humans and machines: A friend can sleep over at your house and tell you stories. She talks to you. If you play with a robot, it can only do certain things. Cosmo can only play with dice.
Alexa can only answer questions or sing a song. And that ball can only roll. Julia can do all those things. You just have to get a lot of single things from one robot, or get other robots. First this one, then that other, then the next. And with a person you don't have to get a new one. You always have them there and they can do everything. Marie and Julia hit the nail on the head: The robot cannot replace a best friend, no matter how smart it is. A machine cannot substitute a human being. Here our journey through the world of artificial intelligence draws to a close. There will be major changes! But: It is not the machines, but we humans who cause them. We have not only the freedom, but also the responsibility to shape our own future.
No comments:
Post a Comment