AI program that can play complex video games like Zelda and call of duty

Discussion in 'Intelligence & Machines' started by mackmack, May 10, 2016.

  1. mackmack Registered Senior Member

    Messages:
    123
    The most advance Artificial Intelligence program in 2016 is Google's Deep Q network, that can play simple atari games. However, based on its design, it can never play complex games like the call of duty or Zelda. Currently there are no AI programs in 2016 that can play the call of duty. In order for an AI program to play the call of duty it has to have knowledge at a 6th grade level. Skills will include generating common sense knowledge, decision making, predicting the future, doing recursive tasks, induction/deduction skills, and 6th grade knowledge.

    My AI program can play the call of duty. In fact, it can play really complex video games like Zelda or super metroid. Actually, this AI program can do any human task. Here is a video link to the thoughts of the robot while playing metroid:



    Inventor: Mitchell Kwok
    The secret is that I'm not using machine learning, specifically deep neural networks like the ones used in IBM Watson or googles AI.

    The problem these software companies are trying to solve in 2016 is "how does the AI program reference domains". For example, if the AI program learns how to identify numbers and count, how can the system take that knowledge to learn addition or multiplication? Next, how does the system take knowledge on addition and learn how to do calculus problems?

    counting->addition->polynomial->trigonometry->calculus problem

    So, my AI program learns information in terms of a bootstrapping process, whereby old information (domains) are used to learn new information (domains) to form complex intelligence. And their are other problems that arises, such as how do you modify information in a recursive manner within these domains? All those problems are solved in my AI program. watch the videos and they will show you the secrets to robots with human-like intelligence.

    In conclusion, there are no AI program in 2016 that can play the call of duty. My Artificial Intelligence can not only play the call of duty and Zelda, it can do very complex human tasks like writing software programs for Microsoft. Patents were filed on this invention starting from 2006. You can go to my website or view documents online at:
    http://www.humanlevelartificialintelligence.com
     
    ajanta likes this.
  2. Google AdSense Guest Advertisement



    to hide all adverts.
  3. Q-reeus Valued Senior Member

    Messages:
    2,533
    Such a spectacular advance in AI with strategic implications suggests either landing a multi-billion dollar deal with any number of players out there, or being gagged by the military owing to said strategic 'winning edge' defense/offense implications. I don't get the impression either has happened. Care to explain why not?
     
  4. Google AdSense Guest Advertisement



    to hide all adverts.
  5. mackmack Registered Senior Member

    Messages:
    123
    those AI videos (about 200 of them) was in response to the patent offices request for further clarity regarding my invention. based on your last comment i think you see the importance of the technology and what it can do.

    The concept of human like AI is pretty much unheard of back in 2006. It will take some time for scientist and the general public to accept this technology. and besides, the internet (facebook, google, twitter, etc) controls the stories that go viral and not individuals.
     
  6. Google AdSense Guest Advertisement



    to hide all adverts.
  7. Q-reeus Valued Senior Member

    Messages:
    2,533
    Hmm....Well it's your life but seems to me given as so such an evidently proven-via-video-games hot technology, the world should have been your oyster for some time now. You have approached the likes of Microsoft, Apple, Google, IBM, maybe DOD? Those folks are savvy and always scan the literature, so strange to me they weren't first in beating down your door back there in 2006.
     
  8. mackmack Registered Senior Member

    Messages:
    123
    actually a few knew about my work back in 2007 (i didn't tell anyone until a year has passed). They tried to take my ideas and incorporate them into their technology.

    i publish a book called Universal Artificial Intelligence: book2 in late 2006 and if you read the back it says: The UAI is an AI program that can play any video game for any game console without any predefined rules or goals. Next, in 2015 Deepmind technologies (bought by Google for 400 million dollars) unveils the Deep Q Network to play simple Atari games. And this is what they said in their press conference: Our AI program can play any Atari game without any predefined rules or goals -- it is a general purpose AI.

    Deepmind should rename their AI program to Universal Artificial Intelligence because that is the most obvious term to use. Google doesn't want to be like IBM, where they spend 5 years designing watson to play jeopardy, then spent 8 more years to design watson to diagnose patients, and 10 more years to make watson drive a car. Google wants one AI program that can do everything. Thus, using universal artificial intelligence is the correct term to use to represent this type of Artificial Intelligence.

    The flaw with their AI is the deep neural network (3 layer structure). i have explored this option back in 2001 and found many flaws that prevent the software from achieving human like intelligence. My guess is that IBM and Google will continue with there Deep neural net for 5-10 years more, find out that it really doesn't work, and take a 180 in my direction. By then it would already be too late, my design was sent to the patent office back in 2006.

    The video above is what they hope to build. and from experience, using a deep neural network to build something like that would be IMPOSSIBLE.
     

Share This Page