Re: T-scale

From: HARNAD, Stevan (
Date: Fri Jun 02 2000 - 12:46:03 BST

On Thu, 1 Jun 2000, Shaw, Leo wrote:

> Stevan, I realise this is a bit late in the day, but could you remind me
> (us) of precisely how the T-scale is defined: am I right in thinking (1)
> is toy, (2) is implementation independant, (3) has integrated
> sensorimotor parts, (4) is internally indistinguishable from a human,
> and (5) is some kind of biological replica?

Hi Leo,

You are right in all the essentials here, but kid-sib would never be
able to figure out what you meant. So, on the exam, remember you're
explaining this to kid-sib, not to me. (For me, you have confirmed that
you know the levels of the T-hierarchy.)

> Also, how are the tests defined - you mention in the Skywriting that T2
> is offline, but T3 necessitates interaction, what other requirements are
> there?

t1 is just an arbitrary, SUBTOTAL fragment of our Total capability:
chess-playing, walking, recognizing faces.

T2 is TOTAL, Turing-indistinguishable pen-pal capability (EVERYTHING and
ANYTHING a pen-pal could do via email, symbols, and symbol-manipulation,
including: playing or learning chess, WRITING about walking, and
writing about faces).

T3 is TOTAL Turing-indistinguishable sensorimotor (robotic) capability -- that
means everything that you and I can do, in the real world of objects and
people, INCLUDING t1, T2, and everything else we can do.

T4 is T3, plus Turing-indistinguishable in inside structures and
functions, not just outside ones (doesn't just look, walk, quack like a
duck from the outside, but also if you open it up and look at what
happens on the inside when you peek/poke/simulate/ablate/record its
brain activity). But T4 can be all synthetic function.

T5 is us, and clones of us, made out of the real stuff.


This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:36:29 GMT