The new technology of Heuristic HervestingTM from
Mental Engines
shares the positive traits of both existing approaches to AI, namely
Classic AI
and Biological Simulation.
Classic
AI is based on logic. It uses rules to compute with certainty the
correct answer to a problem, Examples of successful "Classic AI"
include "Deep blue" (The world-champion chess program), Expert systems,
Google's search engine, and many more. In Classic AI, one can always
ask why the program produced a certain output, and the answer would be
meaningful - a list of observations and rules that pertain to the
problem. In the fields where these algorithms work, they are often
superior to Humans in that they never tire of considering all the
factors in great detail.
A
more recent approach in AI is Biological Simulation (including Neural
Networks). Examples of successful application are speech recognition
and image analysis, cryptographic analysis, and other problems
recognizing ill-defined patterns. In this type of program, there is no
sense in which you can ask "how does the computer know that?" - the
computer guesses using hundreds of factors that make no sense to us
Humans. In the fields where these algorithms work, they are usually
inferior to Humans, that can recognize word amongst noise (for example)
better.
Mental Engines'
technology modeled after the Human mind, in the sense that it
emulates basic (uneducated, innate) thought processes, implementing
learning mechanisms that are capable of picking out the appropriate
behavior in complex situations. The technology is, like a human,
adaptive to its environment, and modifies its behavior according to the
customs of its environment. Conceptually, it is in the middle between
the two existing approaches. The algorithms can provide an explanation
as to why an action was taken, but such an explanation is not
necessarily exhaustive.
Mental Engines' technology is particularly suited to replacing low-level uneducated labour in tasks such as manning gates, driving cars, delivering parcels, etc.