artificial intelligence Archives

An AI Kill Switch Is Needed to Stop Robots from Taking Over

June 9, 2016 By Deborah Campbell Leave a Comment

What is today an old sci-fi fear of technology taking control over humanity, could very well become reality one day – or at least a real threat. That’s why scientists today have started thinking ahead and developing a kill switch for artificial intelligence (AI).

Two researchers, Stuart Armstrong, from the Future of Humanity Institute at the University of Oxford, and Laurent Orseau, from Google DeepMind, have already published a paper on why and how future AI machines could be prohibited from learning to override human input.

With their solution, humans will forever remain in charge of machines. “It is sane to be concerned – but, currently, the state of our knowledge doesn’t require us to be worried,” said Dr. Orseau.

However, he also agrees that it’s smart to start working on AI safety before any kind of problem arises. AI safety is concerned with making sure the algorithms providing the machines with learning power work the way they are supposed to.

The research of Orseau and Armstrong focuses on how scientists can be sure AI machines can be interrupted by the humans in their charge – at any time, without the machines themselves learning the code for the kills switch. This way, they cannot overcome or prevent human intervention.

“Now and then it may be necessary for a human operator to press the big red button to prevent the agent from continuing a harmful sequence of actions,” they wrote in their paper, called “Safely Interruptible Agents.”

The challenge is to find a way to disable the AI machine without the learning agent learning how to disable the red button, for example.

It’s important to be prepared, the authors explained, since it’s unlike for AIs to behave “optimally” at all times. Thus, the machines we create should not “be able to disregard human attempts to stop or interrupt its functioning.”

For example, an AI created in 2013 was taught to play Tetris, but in order to avoid losing a game, the machine had learned to pause the game indefinitely.

Other scientists also agree to the importance of a kill switch, such as University of Sheffield AI expert Noel Sharkey. He also told the BBC that the ideal situation would be if the AI program could stop itself when it detects that something is wrong.
Image Source: Flickr

AI System Recreates Nobel-Prize Winning Scientific Experiments

May 17, 2016 By Cristopher Hall Leave a Comment

We could all do with a shorter work week, so Australian physicists thought they could do themselves a favor and create an AI that doesn’t need much oversight.

This artificial intelligence system can run and even enhance complex physics experiments, allowing human scientists to concentrate on research design and high-level problems. We would all love to leave the nuts and bolts of our job to a robotic lab assistant.

Three physicists won the Nobel Prize in 2001 for the process that creates a Bose-Einstein condensate, a hyper-cold gas, and this AI system was able to replicate the experiment much better than any scientist could. Thanks to directed radiation, a group of atoms are slowed nearly to a standstill, resulting in all kinds of interesting effects.

The team at the Australian National University cooled a small amount of gas down to 1 microkelvin — a millionth of a degree above absolute zero — then ceded control over to the AI.

The system was then supposed to figure out ways to apply its lasers and control other factors to cool the atoms down to a few hundred nanokelvin (i.e. a billionth of a degree). After repeating the process dozens and dozens of times, the AI found increasingly efficient ways to do so.

Paul Wigley, a co-lead researcher from ANU, explained that the AI managed to do things “a person wouldn’t guess, such as changing one laser’s power up and down, and compensating with another.”

It was surprising to witness the machine learning to do the experiment by itself, from the beginning, in under an hour. “It may be able to come up with complicated ways humans haven’t thought of to get experiments colder and make measurements more precise,” he added.

Even though Bose-Einstein condensates have powerful properties, their extreme sensitivity to fluctuations in energy makes the process of developing and maintaining them particularly difficult.

With AI systems, many parameters can be monitored at once and the process is quickly adjusted in ways that humans could not understand or replicate. As a result, condensates are created faster, under better conditions, and in greater quantities.

It’s also useful that the AI doesn’t ask for lunch breaks or take vacations. Michael Hush, the other co-lead researcher at the University of New South Wales, said that “it’s cheaper than taking a physicist everywhere with you.”
Image Source: Wired.uk

The Remedy to Urban Beliefs of Robot Uprising

February 17, 2016 By Nancy Young Leave a Comment

Researchers at the Georgia Institute of Technology say they may have figured out the remedy to urban beliefs of robot uprising. Possibly one of the most popular themes in science fiction works and the like, the idea that robots created by mankind would one day become sentient and turn against their creators is an idea that strikes fear in many.

Now, this hasn’t so far been a real issue as the robots created up to this point in time are still far from becoming sentient and capable of getting at the level where they would make decisions of the likes found in Sci-Fi novels by themselves. However, it’s very well known that artificial intelligence is the focus of many tech giant companies out there, with some of them currently having entire departments focusing their finances and man force in order to develop AI’s known capabilities.

The methods that these companies apply is by devising systems that are capable or learning through experience. Anyone has, at one point, come in contact with system learning how to ‘behave’ according to the profile of the one using them is displaying. The simplest example would be the way Google searches work these days.

By gathering data about the kind of websites you visit, the things you post, the things you normally look up, what you type, what you watch and what you listen to, Google now makes it easier to find personalized results for you based on what it has learned about the user.

That’s the basic idea behind robot ‘psychology’ too. By exposing robots to multiple fields that they can draw their knowledge from, their prediction system would – in time – allow them to achieve outstanding levels of artificial intelligence.

However, this takes us to the main point that the scientists at Georgia Institute of Technology are trying to make; researchers and great scientific minds alike believe that a robot uprising is not out of the question when the robots will reach a certain level of understanding.

Which is why researchers have also tried to think of a method to prevent that from happening, even if we’re possibly a few decades early. They believe that by delivering good moral messages to robots, they may learn to peacefully live with humans. The first idea that the scientists thought of were actually fairy tales – a very simple method of teaching robots the difference between right and wrong, good and bad, similarly to how adults do with their children.

That way, robots may be able to adopt the values of the human society as it is nowadays.

Image Source: 1

SwiftKey Reveals Popular Emojis For Each State

August 19, 2015 By Cristopher Hall Leave a Comment

SwiftKey, the Android and iOS program for predicting the next word the users will type, has just revealed which are the most popular emojis in each state. SwiftKey gathers its data by analyzing the users’ previous messages as well as looking at social media posts.

So what did the results tell us about the United States of Emoji? After analyzing more than a billion popular emojis from its users, SwiftKey let everyone know that California really likes taxis, while Nevada is mesmerized by eggplants.

The company that was founded in 2008 revealed that they now know a great deal about how people interact and communicate. And this includes popular emojis use. SwiftKey’s US Emoji Report revealed where Californians’ emojis stand at: they ranked high for taxi icons (Uber and Lyft are based here), but also for healthy, positive things like sushi, sunset, bike and lemon.

An unexpected discovery came from Hawaii. You’d expect their SwiftKey users to rank high for rainbows, waves and volcanoes, but everyone was surprised to see they also like to use the basketball emoji as often as the can.

The firm’s report doesn’t come alone. They just brought out a new feature that is momentarily only available in iPhone beta which will let you know what your distinctive emoji is. And if you’re an iOs user, besides learning what you signature emoji is, you can also check out which state best represents you.

Let’s expand our horizons and look over the borders. What are other countries’ most popular emojis? While Americans are in the lead with overall emoji use, Canada and Malaysia are right behind, on second and third place, while Vietnam ranks last on the list. Wondering what country really digs romantic emojis? The Russians are in the lead, using three times as many love emojis than the average.

Brazilians showed a preference for religious emojis such as prayer hands and church icons, which were more than double the average. Malaysians seemed hooked on the sleep related icons, while Australians don’t shy away from overusing drug emojis like cigarettes. The US ranks first in the meat department, with the French people using the least meaty emojis.

If you were wondering where Spain falls in all of this, they lived up to their reputation and ranked first in party emojis. At the same time, Arabs really like using the red dressed dancing lady.

SwiftKey seems to continuously improve its knowledge, which is not only limited to popular emojis. That means the software can predict your next word while you’re typying, and it can also auto correct your messages. By storing your preferences online, users can easily access them from any gadget. You didn’t know you’re already using so much artificial intelligence to your benefit, did you?

Image sources 1, 2, 3

Categories