Sunday, 17 November 2013

Top 10 Interesting facts about Artificial Intelligence | The Future trends in A.I - Artificial intelligence of Robots

       The Brain serving as a powerhouse of thoughts, innovation & creativity is believed to be ‘unreplicable’ .To put it in simple terms,, we cannot make a computer feel for an event emotionally or think like we do. The element of intuitional choice’ and absolute randomness is what that draws a fine line between us and the machines.  Even though a computer can calculate a million chess positions against its human counterpart in a Chess game, it ‘does’ fail in a match with a human and the winning part is not assured. 






                                   Even, Google admits that the human brain still beats Algorithms. Computers can never become intelligent until they find a way to impart intuition and the ability to combine logic with imagination in the tasks they do. Einstein & Newton didn’t become the masterminds just because they were applying the right logic &formulas every time. Einstein got his insights through pure imagination and inspiration. Newton was inspired by an Apple. Geniuses stun the world by combining art & science. Having pinned that, we should not underestimate the power of technology on the other hand. 

                   Recently, the CAPTCHA system was successfully cracked by an algorithm developed by an A.I firm, Vicarious. It recognized the text CAPTCHA with 91 % accuracy. (Human Accuracy stands at 70 %!)  IBM Deep Blue was the first computer to defeat a chess Grandmaster, Garry Kasparov (Hailed as ‘The Mozart of Chess’). When scientists at Google’s secret Lab built a brain simulation with 16000 computers with 1 billion connections, the computers particularly identified pictures of Cats, recognized humans, their body parts and objects too!
                   Let us see some of the emerging trends & interesting facts about Artificial Intelligence and how it will change the future.

1. Deep learning
                   The brain’s neural networks are a wonderful creation.“Deep Learning” Typically uses the simulation of Brain’s neural networks, replicating neurons of neocortex. The simulation mimics neurons in the brain. They associate content instead of just analyzing the huge pool of data & arrive at the final result with great success.
                  

Graph showing deep learning

                   Deep learning can be viewed as a method that makes a computer solve a problem independently by going through multiple levels of learning. Instead of writing a code that instructs how to solve a problem, Computers can solve the problem on their own by studying simple basic concepts. It is much like Human understanding wherein several hierarchical levels of cognitive abilities are needed to completely understand a particular object/thing & relate it to something else in a meaningful way.
                   Google uses Deep Learning to optimize their search results, improve Android Voice Recognition and for better ad targeting .It is seen that the computers started outsmarting Google Engineers in certain tasks.  The computers had a very high success rate in identifying objects that humans didn’t notice. Google admits that at a certain level they wouldn’t be able to tell how the ‘process’ takes place. In Deep Learning there are many layer s of processing. In the first layer, changes in pixels are noticed. The second level may try to identify edges and shapes and the third layer gets the information from the second and makes it much more informative and so on. Facebook is beginning with its 8 member Artificial Intelligence group that will use Deep Learning to improve facebook’s News Feed app and predict the future behavior of people. Well why would facebook want a Deep Learning program to do that? It is for Ad targeting basically.

2. Simulating feelings
                
Picture of Mark Sagar's Baby

     Remember when we said computers can never ‘understand’ emotions? Well, they can at least imitate now. But still not like we do. Meet Baby X, a virtual baby developed by Dr.Mark Sagar (the animation expert behind Avatar & King Kong movies). Baby X is a Computer Driven simulation that shows emotions in its digital face! The virtual face reacts exactly like a normal baby would do in real-time. It is a ground-breaking advancement in human-robot interaction. The neurochemical reactions are simulated in the baby’s simulated brain which in turn drives facial expressions. The baby gets upset & cries if we move out of reach and gets happy if we are around. Oxytocin, Cortisol & Dopamine levels are simulated and vary depending on its social connection.  Many such experiments have been going on that could make human-to-robot interaction look more lively and real.

3. Chewing Big Data
    Graph connecting users
    
Big data refers to the collection of ‘huge’ amount of data that cannot be processed with conventional Database systems. Usually it’s a lot of unstructured data. This might include data such as images, status updates, profiles etc that are spread out in the social networking sites and such data in large companies. Big Data & Deep Learning are like pairs. Google used its deep learning on its vast collection of video clips that made its computers identify cat pictures on their own. And now, Facebook is beginning to build a virtual brain with its user’s personal data! One thing that we are sure about is that they are not identifying pictures of cats & dogs with that. Facebook is trying something that no tech giant has done. It is trying to characterize the behavior of people with their own personal data. Although we are not sure what facebook is up to, they might hold the key to something interesting in A.I.
4. Blue Brain
       
          Blue brain aims in reverse Engineering a biological brain & recreates it using a computer simulation. The simulation will be accurate at cellular level! IBM’s Blue Gene Computer is used for the simulation process and hence the name ‘Blue Brain’. It is claimed that a full human brain simulation will be possible by 2023. The scope of Blue Brain Project includes improvement in fields of healthcare to advanced computing & better A.I. The radically new method used by them studies neurons through the biological data obtained which is simulated in a supercomputer resulting in a complete blueprint. Although this method sounds promising, the power consumption of Blue Brain is 8 MW, an equivalent of power consumption by 1600 homes!


Blue gene super computer

5. Neuromorphic Approach
Neuromorphic computer chip
       Neuromorphic engineering aims in replicating the ‘brain’ but there is a slight difference between this approach and the ones that use muscular super computers. Neuromorphic computers are sophisticated computer systems similar in size, speed & energy consumption to that of a brain! Human Brain consumes only 20 W of energy whereas the supercomputers of today are consuming power in Megawatts. Still the intellectual crispness of human brain is not seen in Supercomputers and stays unmatched. Neuromorphic engineering aims in creating electronic circuits that fire neurons like human brain does and the dimensions of these circuits remain very small. Consuming just 5 Megawatts of power, they can mimic 1 million neurons in a NeuroGrid! If this hardware is combined with the ability to Deep Learn, Terminator’s SkyNet will become a real possibility.
6. Mind Uploading
      
Brain computer interface
   Mind Uploading involves transferring our brain to a digital storage medium and moving conscious memory from one to another.  Although it sounds too fictional, Google futurist claims that humans will be able to upload their entire brain to computers and become digitally immortal! Mind uploading and emulation could make us super intelligent, super fast and immortal in real world. The Future will be like ‘The Matrix’ movie where everyone will be plugged in to computers with Brain Computer Interface BCIs. Assuming that A.I would not have us as slaves or do anything evil, mind uploading sounds really cool. Although some argue over the impracticality of this approach, mind uploading remains the ultimate embodiment of machine and mind.

7. Transportation
       
Highway in chennai
         One thing that strikes our mind when we see ‘transportation’ is the term ‘traffic’. Traffic is everywhere, even on the internet. To control traffic we used the traffic signals. Without traffic signals we all would end up in the hospital or in the rear window of another car. Studies reveal that a traffic signal system controlled by human brain is very efficient than the conventional system. Since the human brain is good at traffic stuff, researchers are building ‘machine learning’ Artificial Intelligent traffic control systems that could control the traffic efficiently. They also look forward to the future where vehicles will be equipped with GPS and Wi-Fi to make traffic control a simple task.
8. The Singularity
       

Megatron
            Here is where Artificial intelligence gets evil. Technological Singularity or simply ‘Singularity’ means ‘The End’ to the humans. Artificial Intelligence will one day (for sure) overtake us. Singularity is the theoretical moment in time where the machines will be smarter than us. According to Irving John Good, the last invention of the human race will be the ultra intelligent machines. Since an ultra intelligent machine would have surpassed all human capabilities, it will build machines that are far beyond the human intellect.  That definitely sounds spooky, isn’t it? Unfriendly A.I is much more likely to happen than friendly A.I. But there is a slim chance that we beat them if they turn against us. Either we become the machines that we fight / get some help from friendly A.Is.
9. Self Improving A.I
     
HAL 9000
   Since the basic nature of Intelligence allows further creation & invention, a super intelligent system would further evolve through recursive self improvement. It’s a symbol of strong artificial intelligence. Imagine how would it be if a computer software codes itself? This results in adaptation to the situation. And anything that adapts to its situation survives it. The redesigning of cognitive algorithms and updation to a better self could turn it into something that it was not programmed initially. It becomes ‘self aware’.  Self Improving A.I could start Intelligence Explosion. After that, nothing is guaranteed for sure and cannot be predicted.

10. You’re fired!
Japan's worker robot
        Have you ever imagined that you might lose your job because a robot is qualified to do your task better than you? Well, that might become a reality in the near future. Artificial Intelligence has created better robots and it continues to do so. Japanese firm Kawada has created Robots that could work alongside us creating “Automated Labor”. There are automatic self driving cars & cleaner robots too. Even as Google’s computers become smarter each day, they result in the reduction in number of employees. To any investor this might seem a fair advantage, reducing the number of heads while increasing performance with computers that you don’t have to pay salary for. So watch out! A.I might steal your jobs!

Share if you found this useful & don't forget to like us on facebook.


Comment on this post for suggestions/improvements!

3 comments:

  1. very nice information about Artificial Intelligence, you can also improve your article by simply checking my articles on my blog www.techora.net , hope you'll like it.

    ReplyDelete
  2. The video games millions of people enjoy on a daily basis employ this type of Artificial intelligence. Very interesting information about AI.

    ReplyDelete
  3. Your blog is very informative and gracefully. Your guideline is very good. Thank you
    Aido | Aido Robot

    ReplyDelete