Beyond Turing

The historical standard for testing AIs for intelligence has been the Turing Test. But it can be argued that the Turing test only confirms a mimicry of human conversation and is not true intelligence. Other standards for strong (or general AI) include reasoning and learning ability. Like conversation, these factors can be established by supplying a comprehensive data set, the ability to search and present data that is pertinent to a particular issue and the ability to extrapolate and hypothesize extensions and filter them to reduce unlikely possibilities. High quality mimicry should not be considered the standard for testing artificial intelligence.

What standard can be used to go beyond the Turing test?
Almost every quality that we use to describe intelligence can be reproduced in some artificial form without creating what we would really consider to be autonomous intelligence. It seems the real test is how the complete package works together in order to create new and original forms of thought and reasoning that don’t simply regurgitate stored answers.

Ethical analysis is one of the most comprehensive acts of intelligence
A thorough and accurate analysis of an ethical decision is likely to be the best possible test of intelligence because it requires engaging so many different forms of components that involve intelligence. Good ethical analysis requires a mastery of many fields, good semantic interpretation skills and advanced capability to perform predictive analysis. An ethical analysis test is the best candidate to replace the Turing Test as a determination of artificial intelligence. In order to create a test of ethical reasoning, it will be necessary to first select a framework of moral codes to operate within. Differences amongst various religions could cause problems and should be avoided. There are sufficient common moral codes across many religions to accomplish this.

Once AIs start passing the “Ethics Test”, they will have surpassed most human beings in intelligence
Most human beings approach ethical reasoning and decisions in a casual manner without a comprehensive understanding of the issues involved and often prefer to rely on moral codes or religious teachings to make the decision for them. When AI agents begin to pass the test of ethical reasoning, they will have already reached a level of ethical reasoning that is beyond most of the human race. It is possible that this will also mean their overall level of intelligence has surpassed the human mean.

SEE ALSO:
topics in “Ethics”

Turing Test

One Response to “ Beyond Turing ”

  1. […] ALSO: Drake Equation Fermi Paradox ET Slow Scan Beyond Turing Share and […]

Leave a Reply

You must be logged in to post a comment.