[agi] Chess Master Theory Of AGI.

[Mike Deering]

As far as chess goes, anyone who doesn't think I know what I am talking about is invited to come to my house and face me "over the board".   The chess environment is complex enough to tax the human level intelligence.  Very few people who play chess have any idea what they are doing.  Chess mastery is an example of goal directed intelligent problem solving with five hundred years of written history and a competitive environment that produces the optimum strategies derivable by human intelligence.  A chess master is one who knows and practices the optimum strategy.  This strategy is generalizable to any goal directed problem.    The general strategy is:   1.  Learn everything you can about the environment that is causally related to the goal including all causal relationships between the elements in the environment that may be causally related to the goal.  In chess this means analyzing the board situation.  When I use the term "goal" here it may be a multiple complex goal.  In life this means learning everything you can about anything that may be important.  That's a lot.   2.  Determine the plan that is appropriate to the situation.  This can't be accomplished until you have analyzed the situation in #1 above.  Choosing a plan that is not appropriate to the situation is a classic error of good (not master) chess players.   3.  Determine the most effective moves to move the plan forward and begin accomplishing those moves.    4.  Continually re-evaluate the situation and the effectiveness of the moves.  Modify the plan and the planned moves as necessary.  This re-evaluation doesn't require starting from scratch, just keeping abreast of any changes to the situation.      Of course, in these four steps a lot of detail has been left out involving analysis, problem solving, and planning but the human programmer has written a lot of code solving a lot of different problems and I think the algorithmic problem solving space has been pretty broadly mapped out.  The solution to the AGI problem will not require any algorithmic theoretic breakthroughs other than putting all the pieces that we already have together in the right structure and running it on sufficient hardware.   Their are enough different efforts moving computer intelligence forward as to constitute a statistical pool that will not be significantly affected by the actions of any one individual or small group.  Although, a Manhattan style project could have an effect.  Human level AGI is coming as inevitably as faster computers.  Most new software is written for the largest installed computer market which at the moment is the $1000 desktop.   The intelligence of computer software keeps constant with the capability of the $1000 desktop.   When the $1000 desktop reaches sufficiency to run human level AGI it will be available.  This is an economic certainty.    This will occur before the predictions of the experts in the field of Singularity prediction because their predictions are based on a constant Moore's Law and they over estimate the computational capacity required for human level AGI.  Their dates vary from 2016 to 2030 depending on whether they are using the 18 month figure or the 12 month figure.  Moore's Law is currently at 9 months and falling.  My calculations based on a falling Moore's Law put the Singularity on April 28th, 2005.    This human level AGI in a computer will be quite superior to a human because of several advantages that machines have over gray matter.  These advantages are: upgradability, self-improvement through redesign, self editability, reliability, functional parallelism, accuracy, and speed.  This superiority will be quantitative not qualitative.  It will be superior but completely comprehensible to us.  The belief in a radically different form of advanced thought incomprehensible to present humans is philosophical in nature, not based on evidence.   Mike Deering.