This school of thought represents the best possible way of thinking.
What is it based upon?
Simply produce greatest number of different variations, and pick the best variation.
In chess game, this would be very simple. Just after opening move, play many variations based on that opening move, and see which variation wins the game. When you try enough different versions of that particular opening move game, you eventually end up with one which wins or at least results in draw. This is because chess game only has two types of moves: moves which allow opponent to defeat you if he is smart and moves which dont allow that. So you produce enough versions of game until you get the moves which cannot be defeated. This of course depends on what opponent plays too, but his moves are not unlimited, and thus, there is a correct move for each of his moves, and that correct move is learned by producing different variations until correct move is found.
The thinking process is very simple: try one variation, if variation fails, try different variation. Repeat process and you will reach variation which doesnt fail.
This school of thought likewise works with theories and explanations. Producing greatest number of explanations for some event, and then seeing which explanations fail and which stand.
Likewise, this school of thought explains origin of life. Simply put, given infinite space with infinite number of logical variations, at least one variation will end up with life because life is logically possible and must result in at least one of infinite number of variations. In this sense, what is logically possible must exist somewhere, in case of infinite variations.
In AI technology, randomness of AI can be controlled by producing great quantity of content and then selecting the best.
In programming, you generally create different versions of program and select the best.
In debating, you create different versions of argument and select the best.
AI will become super intelligent because AI will apply this school of thought to its maximum.
While human mind is limited and can only think in terms of dozens to hundreds of variations of something per day, AI will be able to think of millions of variations per day, and select the best. So new AI will be best singer, best artist, best debater, best chess player... as long as its system for determining what is best is correct.
For this system, 3 things are needed:
1. Greatest number of variations
2. Desired goal
3. System for removing variations which dont suit desired goal
A simple system, but it gets much more complicated depending on task. For example, desired goal might be incorrect or not universally desired. But that too is solved by having multiple desired goals. For example, not all people like same music, but almost all people like music. In this sense, AI can see what different people like, and produce music for each taste.