Tech Trend 247

Losing jobs to technology is nothing new. Since the industrial revolution, roles that were once exclusively performed by humans have been slowly but steadily replaced by some form of automated machinery. Even in cases where the human worker is not completely replaced by a machine, humans have learned to rely on a battery of machinery to be more efficient and accurate.

A report from the Oxford Martin School’s Programme on the Impacts of Future Technology said that 47 percent of all jobs in the U.S. are likely to be replaced by automated systems. Among the jobs soon to be replaced by machines are real estate brokers, animal breeders, tax advisers, data entry workers, receptionists and various personal assistants.

47% of all jobs in the U.S. are likely to be replaced by automated systems.

But you won’t need to pack up your desk and hand over to a computer just yet, and in fact jobs that require a certain level of social intelligence and creativity such as in education, healthcare, the arts and media are likely to remain in demand from humans, because such tasks remain difficult to be computerized.

Like it or not, we now live in an era dominated by artificial intelligence(A.I.). A.I. can be seen as a collection of technologies that can be used to imitate or even to outperform tasks performed by humans using machines.

We might not first see it but we cannot avoid running into one or more systems that use some form of an A.I. algorithm in our day-to-day activities — such as searching for some information using Google, purchasing a recommended product on Amazon, or recognizing faces in an image uploaded to Facebook.

Deep learning

Recent breakthroughs in A.I. are largely attributable to a technique called deep learning. Often known as machine learning or neural networking, deep learning involves “training” a computer model so it can recognize objects from images. The power of deep learning-based A.I. systems lies in their ability to automatically detect noticeable features and use them to solve hard recognition problems.

It is often difficult for a human to explain the exact procedure at a sufficiently detailed level.

Although humans could easily perform such recognition tasks almost unconsciously, it is often difficult for a human to explain the exact procedure at a sufficiently detailed level so that it could be programmed into a computer.

With deep learning all this has changed. Now, deep learning-based A.I. systems can figure out the important features for solving difficult problems that were once thought to be solvable exclusively by humans.

And as a result, humans will have to mentally prepare for the fact that some of our jobs will be lost to A.I. systems. We might even have to call A.I. systems our colleagues or bosses in the near future.

But despite the deeper level of knowledge that our computers will soon acquire, losing our jobs to machines doesn’t have to be a bad thing. Letting machines do the bulk of the work means that humans will be freed from routine tasks that computers are better at performing with higher accuracy rates, such as driving cars.

This should enable humans to think like humans instead of machines. It will also free up time and energy for humans to engage in more creative and intellectually stimulating activities, possibly assisted by A.I.

Emotional intelligence

A.I. systems have already become far too complicated for the average person to understand, let alone repair, so there will be new roles created which will require people who can act as intermediaries between computers and humans.

Similar to professions such as medicine or law, where professionals with specialized skills are required to interpret technical details for everyday folk, we will need professionals who speak the language of A.I.

Ethical issues arising from human and AI co-working environments is a real concern.

These professionals may vary in their skills and are likely to consist of software developers, computer scientists and data scientists.

But ethical issues arising from human and A.I. co-working environments is a real concern. It is one thing getting a face incorrectly recognized in an image uploaded to Facebook, but a totally different matter if cancer is misdiagnosed by an A.I., which could very easily happen. After all, computers make mistakes, just as people do

Although A.I.-based systems are becoming smarter than humans in many fields, these systems are far from perfect and are unlikely to ever be perfect considering the unpredictable learning mechanisms they use.

That said, it is likely to be the social and cultural changes that will be the real challenge, rather than the technical challenge of A.I. itself. So while robots taking over our jobs can be a good thing, only time will tell if we are ready to accept them as our co-workers.

Danushka Bollegala is a senior lecturer at the Department of Computer Science at University of Liverpool.

A drone, or quadcopter as it's technically called, typically has four propellers to generate enough thrust to let it hover in the air. The propellers are arranged in such a way to counterbalance the weight of the drone itself.

Researchers at the Institute for Dynamic Systems and Control at ETH Zurich have created a new kind of flying machine (technically, not a drone) that only needs one blade to fly.

The flying machine is called the Monospinner and is billed as the "simplest, controllable, flying machine in existence" since it lacks actuators and aerodynamic surfaces. It only has one moving part: the rotating propeller.

The three-person research team (Weixuan Zhang, Mark Mueller and Raffaello D'Andrea) said the Monospinner was born after they asked "what is the minimum number of moving parts necessary for controlled flight?"

The researchers have provided an in-depth look at the math and science that went into getting its flying machine into the air here. The gist: "An unconventional equilibrium is found by analyzing the vehicle’s dynamics. For a certain constant angular speed and propeller force, the Monospinner is able to remain substantially in one position. Feedback control keeps the vehicle near this equilibrium."

To fly it, just throw it into the air like a frisbee:

Needless to say, you probably want to stay away from one of these guys. Just look at how it's spinning at an angle. It could easily take your eye out.

Have something to add to this story? Share it in the comments.

This week on MashTalk, we sift through all the rumors to try to decide: Is the iPhone 7 going to have a damn headphone jack or not?

Plus, we discuss the exciting (and possibly unsettling) news that robots are becoming capable surgeons. Lastly, our longest Rapid Reactions round ever features a debate about sex in self-driving cars. Yep. 

Listen to the episode on iTunes or on Stitcher. You can also subscribe to MashTalk on iTunes by clicking the button below to get our latest episodes, instantly. If you like MashTalk, please fill out our audience survey.

Have something to add to this story? Share it in the comments.

Researchers at the University of Science and Technology of China (USTC) in Anhui, China, have just raised the bar for realistic-looking robots.

One of the teams has showed off a humanoid robot named Jiajia, whose features are modelled after those of five students from the university.

Jiajia's skin is made of a silicone material, and she's able to blink and nod in response to questions.

However, despite looking realistic while standing still, her gestures need work. From the demo video, it's clear that her facial control around blinking and speaking is far from convincing.

The team boasts that she's able to recognize voices belonging to different people, and has some capability of processing speech and responding intelligently — like Siri come to life.

Chinese netizens on Weibo are commenting how realistic she appears at a glance, but some pointed out that her stony facial expressions and mechanical hand movements make her seem pretty creepy.

A user, called "Thousand snow", said: "Too weird, I have goosebumps."

太诡异了，鸡皮疙瘩掉一地

Have something to add to this story? Share it in the comments.

Don't call it a drone. Though the Hover Camera has four propeller blades and a camera on its front, Meng Qiu "MQ" Wang, CEO of Zero Zero Robotics, prefers to call it a "flying camera."

Enclosed in a rigid, lightweight carbon fiber chassis, the Hover Camera hovers in the air, capturing your life. Whether that's walking down Fifth Avenue enjoying an ice cream cone on a sweltering hot day or mountain biking off some cliff, the Hover Camera wants to be right there next to you without you needing to fiddle with complicated remote controls to fly it.

I agree with MQ that drones are too complex for most people to fly without crashing them and endangering people around them. Drone makers like DJI are already making the user experience easier with things like tap-on-touchscreen-to-fly-to-location modes and auto-follow modes, but drones with their exposed blades are still somewhat dangerous if they malfunction by accident.

The Hover Camera — which people will inevitably call a drone, anyway — is relatively small. It only weighs half a pound and doesn't require an FAA drone registration to fly.

"We wanted to build products that are portable," MQ says. "It's very light. This is significantly smaller than any other drone devices out there on the market in this class. The main design consideration is portability, user friendliness and safety."

The device's carbon fiber enclosure protects its blades from giving anyone unexpected haircuts and the foldable design makes it easy to travel with. The Hover Camera, battery and charger fit neatly into a small protective case.

The Hover Camera is also unlike drones in that it truly hovers after powering it on. You simply let go of it and it'll hover in the air at the altitude which you let it go at.

"We wanted the user experience to be very natural — easy to use," MQ explained. "We wanted to make sure the learning curve for users is minimum. So there is no remote control or anything. There's no calibration process."

In auto-follow mode, the camera and software track people's faces, locks onto them, and then follows them.

In auto-follow mode, the camera and software track people's faces, locks onto them and then follows them. For the most part, the tracking worked in my demo with MQ, intelligently following me around as I moved around in a room. There were a few times the camera locked onto other people in the room and got confused, though. The Hover Camera maintained a safe distance of a couple of feet away, but I never worried about it slamming into me thanks to its enclosure. 

The Hover Camera's coolest feature is, no doubt, its ability to autonomously follow a person and record video of them, but you can control it manually with an app using gestures.

A two-finger swipe up and down on the screen controls the altitude. A one-finger swipe up and down moves the camera's one-directional gimbal up and down. Swiping left and right with one finger controls yaw and swiping left and right with two fingers shifts it left and right, respectively.

More fun: You can throw the drone into the air. MQ and I played a round of "hover frisbee" where we threw the drone back and forth. 

The Hover Camera's spinning propellers are exceptionally loud — you won't be sneaking around with this little guy following you — for such a little flying machine, but like everything else about it, it's still being worked on. 

On the front is a camera that can take 13-megapixel photos and record video in up to 4K resolution. By default the camera records in full HD video resolution. When I asked MQ if recording in 4K resolution would hurt battery life dramatically the way it does on smartphones, he was quick to say they've optimized the software and power consumption to ensure it wouldn't.

The camera's equipped with electronic image stabilization to smooth footage out. From a few sample clips I saw, the EIS works, but again, it's difficult to judge an unfinished product.

On the bottom of the Hover Camera is a 3-megapixel camera and a sonar system that tracks its position and objects around it so it doesn't crash. While the Hover Camera doesn't have advanced obstacle avoidance, its computer vision is intelligent enough to run into things.

"It has a Wi-Fi hotspot on the device," MQ explained. "If I have 4G network, I get to keep my 4G connectivity, too. So I can post photos and videos to social networks."

MQ also demoed the "360 Pano" mode, where the Hover Camera spins 360 degrees while hovering in place. Here's a look at one 360-degree panorama video it shot.

The Hover Camera has a battery that's easily swappable. MQ says each battery is good for about eight minutes of flight time. That's pretty short, but the Hover Camera will come with four batteries for up to 30 minutes.

Before the flying camera launches later this year, Zero Zero Robotics is inviting beta users to come up with creative ways to use test and use it.

"So we're working with a bunch of interesting activity groups; for example, some people in Beijing do parkour," MQ gushed. "So they're jumping from building to building and we're using the auto-follow mode to follow them to capture clips of them jumping from building to building. You can't have someone jumping with them. It's just impossible. And we're also working with a group of grassroots basketball players in China. And they can do crazy dunks. [The Hover Camera] can shoot them doing the dunk from meters away. Those shots weren't possible before. Bigger drones are going to be dangerous [to get that close and capture footage]."

The Hover Camera will launch later this year. MQ said the company's looking to sell it for less than $600, which would make it more affordable an attractive to consumers. Zero Zero Robotics has so far raised a total of $25 million dollars in investor funding to produce the Hover Camera. 

The technology is smart from what I saw. But ultimately, it'll be about how creators use it to create content and tell stories.

Have something to add to this story? Share it in the comments.