Saturday, June 30, 2012

AI, Business and you

AI is used to increase a companies bottom line first.

Progressive insurance has a GPS monitor for your car with an accelerometer.
This lets them have a policy agent in your car at all times.
The driving and accident data is matched to set rates and increase profits.

Grok is used by credit card companies to predict a better payback from those who qualify but will tend to make late payments.

It is as important for a business to have more rights than an individual person as it is for an organ to have more rights than an individual cell.


Beginning, Middle, End

How many ways can you divide up a chain?
What is the middle of 1,2,3?
Where is almost to the end of 1,2,3,4?
The "beginning, middle, end Markov algorithm" works on short sequences.
What letter is at 25% of the alphabet?
A little harder?
The invariant representation for alphabet is built from sub invariant representations for the alphabets sub-sequences. You are only able to compare lengths of certain invariant representations chains at face value. Changing focus, putting on a short term memory stack and moving lower or higher in the representation hierarchy facilitates chain and element comparisons.

Friday, June 29, 2012

EXTRA! EXTRA!
READ ALL ABOUT IT!
Public sector hindered in their development of AI!

Each person has his own set of knowledge and has to exit a habitual Markov chain pattern in order to work with someone elses idea. He is not able to do this easily due to only putting an effort into understanding anothers claims if he holds an admirable status.

In a business such as Numenta.com each employee has a shared set of knowledge. When a possibility is presented to the boss he asks his top employee if he is able to make it work. An employees Markov chain is already based on receiving work from the boss and his reward system is based on his ability to impress so he will work hard to make it work.

The Numenta forum was made to let the public sector work with a technology and they published an algorithm to innovate with for free.  Nothing came of the forum and everybody who posted detailed thoughts was left mainly hanging. After some time Numenta decided to make software and proved that a group working together with a common background under an authority has a far greater ability to produce.

When one understands a proposal based on its merits they need not judge the messenger nor deny the proposal based on appeal to an authorities limitations. The latter two behaviors exist as they are part of the filtering system for new memes.

When we wait for an experts approval we are using one of natures hacks to avoid having to find information with our own energy. Todays topics are far beyond believing the hunter vs. the cook about the location of an animal. Getting over this and letting knowledge and logic do the decisions about what is possible is being more of a civilized man and scientist than someone who is not. If we do not do things based on what Wolfram, Kurzweil or someone else says is difficult for computers to do then we will have some major problems producing new things or modeling efficient algorithms which exist in our minds tissues. 

The presentation link below of HTM for games was met with typical negative robotic responses. With no hard wired bias responses to the posts would may be "That is interesting sounding and a bit hard. You say x can do y, how does a,b,c do that- we will help refine a diagram if you post it!"
http://www.gamedev.net/topic/618710-neocortex-based-algorithms/

Numentas technology is treasure chest of gold under a bridge. Would you believe a bum who came up and told you there was gold under the bridge? Of course not. When some unknown guy online tells you those places have the building blocks of AI you feel the need to deny from emotional tribal programming. It is there to save time and protect us. We want to avoid crossing out of the way fields until someone more credible already has and approves.
"Human beings are Markov Machines" - Teddybot



The technology to accomplish the above exists and is spread around online.
This blog is an effort to cluster these technologies to help programmers get over the many fallacies and hardwired negative behaviors which hinder their AI software development.