Pages

Tuesday 11 August 2015

10 AMAZING (SOMEWHAT TERRIFYING) FACTS ABOUT ARTIFICIAL INTELLIGENCE


wps5F6F.tmp
10 Amazing (Somewhat Terrifying) Facts about AI
By Robert Grimminck,
Toptenz, 11 August 2015.

Artificial Intelligence (or AI) got its start in 1950, when computer pioneer Alan Turing introduced the Turing test in his paper “Computing Machinery and Intelligence.” The test involves a judge who must communicate with two participants in two different rooms. In one room is a machine, and in the other is a real person. The judge is then supposed to ask each participant questions and figure out which one is the machine. If the judge picks the person less than 50 percent of the time, the machine would be considered “intelligent.” Since Turing’s initial theory, there’s been a steady march to create thinking machines. In the past couple decades, there’s been enormous progress in this field, but we have to ask: is that necessarily a good thing? Here are some interesting AI facts - some of which are a little…troubling.

10. Most AIs are “Female”

wps1090.tmp

One thing that you may have noticed when it comes to AI that you possibly interact with, like Google Now, Siri, and Cortana, is that the default voice is female. For the record, this isn’t one of the terrifying facts; we just find it fascinating that AI tends to be female. Why is that, exactly?

Well, there’s no specific reason but a few factors play into it. For example, studies have shown that males and females both like the sound of female voices a bit better. Another reason, according to Karl Fredric MacDorman, a computer scientist and expert in human-computer interaction at Indiana University-Purdue University Indianapolis, is that mostly men work on AI so they probably find females attractive and want their AI to follow suit. Kathleen Richardson, a social anthropologist, said that female AI would be less threatening than male AI. For example, compare Samantha in Her to the Terminator machines. Well, except for Terminator 3, but then again, who the hell actually remembers that movie?

9. Artificial Intelligence Pets

wps8C68.tmp

Pets are great, but they have a number of downsides. You have to clean up after them, they can be destructive, they need to be fed, and of course, they die. Something that will take care of all those downsides is AI driven pets. University of Melbourne animal welfare researcher Dr. Jean-Loup Rault says that there are already a number of patents for robot pets and they could be widely available by 2025.

Rault says that in the next 10-15 years, developers will work on a number of aspects of AI and robotics so that manufacturers will be able to build pet-bots that people will be able to make an emotional connection with. Rault believes that robotic pets will be one of the only viable options for most people as the world gets more populated. He theorized that by 2050, only the incredibly wealthy will be able to afford real live pets.

8. Artificial Intelligence Can Repair Itself

wps5E6B.tmp

On the terrifying end of the spectrum, there was a paper published this year about a robot that could rebuild itself, even after losing two of its six legs. The robot doesn’t know what’s broken, but notices that its performance has dropped. Then, using an algorithm based on trial and error, the robot can figure out what’s wrong and how to repair itself. The researchers who developed the robot said that as it fixes itself, it updates its database with all the things that will not work in a phase called “simulated childhood.”

This phase lasts for a few minutes and during that time, the robot processes 13,000 possible movements. Those movements are pulled from 10^47 different behaviours, which is an unfathomable number. For a comparison, that’s how many atoms make up the Earth. The implications of this type of artificial intelligence are essentially limitless. Some of the more exciting prospects include search and rescue and deep sea and space exploration.

7. Artificial Intelligence Can Write

wps5A80.tmp

The first piece for a major news organization that was written by AI appeared on the Los Angeles Times website after a newspaper article reading:

“A shallow magnitude 4.7 earthquake was reported Monday morning five miles [8km] from Westwood, California, according to the US Geological Survey. The temblor occurred at 6.25am Pacific time at a depth of 5.0 miles. According to the USGS, the epicentre was six miles from Beverly Hills, California, seven miles from Universal City, California, seven miles from Santa Monica, California, and 348 miles from Sacramento, California. In the past 10 days, there have been no earthquakes magnitude 3.0 and greater centred nearby. This information comes from the USGS Earthquake Notification Service and this post was created by an algorithm written by the author.”

The computer was able to write the article based on data that was pulled from seismographs, which turned them into figures and then plugged those figures into a story. The technology was developed in part by Larry Birnbaum, a professor of journalism and the head of the Intelligent Information Laboratory at Northwestern University. He was one of the developers of the Quill system, which is an app for companies that do minimal writing. It takes statistics and graphs and compiles them into written reports. As for creative writing, like novels and screenplays, that is the next step for AI, but it’s obviously complicated. Perhaps AI will start off at a Nicholas Sparks’ level and then work itself up from there.

6. Artificial Intelligence can be a Fierce Poker Player

wps6CF.tmp

A big step in the evolution of AI was when IBM’s chess-playing Deep Blue computer beat reigning world champion Garry Kasparov in 1997. Another milestone was in 2011 when IBM’s Watson appeared on Jeopardy and completely destroyed some of the best Jeopardy contestants ever. AI took another major step in May 2015, when a supercomputer called Claudico from Carnegie Mellon University competed in a game of Heads-Up No-Limit Texas Hold ’em poker at the Rivers Casino in Pittsburgh. The tournament lasted two weeks and 80,000 hands were played. In the end, Claudico came in fourth, and the “loss” was close enough to be considered scientifically valid, meaning that statistically, Claudico tied his human competitors. While it didn’t win, it was a big step in AI computing. The big difference between a game like poker and playing chess or Jeopardy is that poker has a lot of missing information, and bluffing is a major strategy and not a logical way of thinking.

Researchers said that the tournament was a great start and they believe that by 2020, AI will be able to beat the best poker players in the world. This type of AI also has a number of other applications instead of just trying to clean out casinos. The algorithms used in Claudico will be applicable anywhere there is incomplete information, including cyber security, medicine and negotiations.

5. Romantic Relationships with Artificial Intelligence

wps79ED.tmp

A question that’s bound to arise is that if AI is almost or completely indistinguishable from humans, will humans be able to have physical and emotional relationships with AI entities? David Levy at the University of Maastricht in the Netherlands had an interesting, but plausible, scenario where sex with a robot could become more mainstream. He said that at first it will come across as geeky, but then a story will appear in a publication like Cosmo in which someone will rave about how great sex with a robot is, and the mainstream attention could alter the prevailing attitudes.

But how close are we to something like that? Well, both Levy and Henrik Christensen, founder of the European Robotics Research Network, thought that by 2012, humans would be having sex with robots. We’re actually closer than you think, as there are currently a number of sex toys that use robotics and promise pleasure like you’ve never experienced before. As for serious relationships, like marriage, AI and Androids still need quite a few years of advancement. With that being said, Levy believes that by 2050, human and robot marriages will be made legal. Of course it could just be that Levy is way too big a fan of the movie Her.

4. Artificial Intelligence Can Learn

wpsC078.tmp

There’s a saying about computers that they’re only as smart as the person using them. However, with advancements in AI, computers are starting to learn by themselves. For example, Google developed an AI system that taught itself to play Atari 2600 games. After doing so, it beat some of the world’s best players.

Another learning AI system is an android developed by the United States Army that learned how to cook from watching YouTube videos. The robot was able to learn the skills from visual recognition and trial and error. While it’s unlikely that the military will use robot chefs, cooking uses a wide arrange of skills so it is an excellent demonstration of what the robot is capable of.

3. AI Will Become Smarter Than Humans

wps3319.tmp

With AI having the ability to learn, computers are getting to be pretty smart. As of 2013, AI was about at the same intelligence level as a four-year-old and there have been lots of advances since then. For example, in 2014 a supercomputer cracked a complicated math problem called Erdos discrepancy problem, which was published in 1930. The amazing thing is that humans can’t even double check the solution because the equation is too long. The file is 13-gigabytes, and just for comparison, all of Wikipedia is about 10-gigbytes.

According to renowned futurist Ray Kurzweil, by 2029 AI will be at about the level of intelligent adult humans. Beyond that, anything is pretty much possible, especially if AI can get exponentially smarter. For example, Kurzweil believes it could lead to something called singularity, which is where humans and machines will meld into one entity.

2. Nautilus

wps8764.tmp

An interesting development in artificial intelligence is an SGI Altix supercomputer called Nautilus. It appears that, to a certain degree, Nautlius can predict the future. For example, it was able to predict where Osama bin Laden was hiding within 125 miles, and was also able to predict the Arab Spring that started in December of 2010.

Nautilus gathered this information from over 100 million news articles from all over the world dating back to 1945. The articles would be analyzed for two different criteria: the mood of the article, and the location of the story. This information led to a web of 100 trillion relationships and the data was fed into Nautilus. From that information, the computer was able to piece the information together and create graphs that charted mood. For example, with the Arab Spring, sentiment in the area was low before the protests.

The author of the findings, Kalev Leetaru from the University of Illinois’ Institute for Computing in the Humanities, Arts and Social Science, is looking at making the information work in real time. While Nautilus won’t exactly predict the future, it may give mood forecasts similar to economic forecasts or weather reports.

1. The AI Apocalypse

wps87C.tmp

There’s little doubt that AI has the potential to greatly improve our lives. AI will make the roads safer, help in medicine, aid the disabled and the elderly, perform customer service and a number of countless other jobs. However, AI also poses an incredible threat, and this isn’t the stuff of science fiction, either. Top scientists and technologists like Stephen Hawking, Bill Gates and Elon Musk believe that AI is a very real, and dangerous threat to humankind.

It’s so dangerous that Max Tegmark, a physicist at MIT, compared it to the development of nuclear weapons and says we may only be able to do it right the first time. In fact, there has been a push to actually slow down advancement on AI and focus more on containment. Containment is important because if we were to ever lose control of AI, we may never get it back. Then it’s just a matter of time, since the AI could wipe out humanity because it could calculate that humans are a virus-like being, or it could kill humans as a way of self-preservation. Essentially, AI will either fix all of our problems or destroy us all. In other words, The Terminator and The Matrix aren’t nearly as farfetched as we previously believed.

Top image credit: GLAS-8/Flickr.

[Source: Toptenz. Edited. Top image added.]

No comments:

Post a Comment

Please adhere to proper blog etiquette when posting your comments. This blog owner will exercise his absolution discretion in allowing or rejecting any comments that are deemed seditious, defamatory, libelous, racist, vulgar, insulting, and other remarks that exhibit similar characteristics. If you insist on using anonymous comments, please write your name or other IDs at the end of your message.