A Turing performant type

The Turing Test is a bit like a driving test, it represents a demonstration of performance capacity of a particular set of behaviours under a certain set of circumstances.

As pointed out by Searle et al, passing the Turing test is a demonstration of behaviour but it doesn’t elucidate any of the underlying mechanisms of the non-human agent.

The agent could be a simulation, replication or emulation. It could be silicon or organic. It is not impossible that 2 humans would play and the judge consistently identifies one more often than the other, the implications of which are interesting, however Turing explicitly excludes such a test in the original conception.

Turing also limits the game to digital computers, and while that technology is still the most favoured, it would seem artificial to restrict say Living Computers from having a pop at the game in the modern version.

Clearly its worth identifying the plausible range of properties that such an agent which passes the Turing Test would posses, and consider the implications of those properties.

It seems to me to be reasonable to describe such a classification as a Turing performant type.


performant is not a word.

Yes I know. Lots of people know. It's funny to annoy people who don't like the word. I want my own word, I want ai researchers and students to put "Turing performant Type (Hodder 2011)" in their dissertations.  Yes I am an ego-maniac.

No comments:

Post a Comment

Don't be nasty. Being rude is fine.