From: Graham Clark (ggc01r@ecs.soton.ac.uk)
Date: Thu Sep 19 2002 - 14:09:06 BST
Hi Srinandan,
in case Stevan doesn't answer first, I can at least jump in on this one:
> I think I now understand what T2 and T3 are supposed to mean. Is there a
> T1? T0? T4? I am sure you'll point me to a cogprints site! :-)
I don't think there's any T0.
T1 (or t1) are systems that implement "toy" tasks, e.g. chess-playing.
Things that are called "intelligent", but aren't really. Maybe if every
single toy task implementation was melded together there'd be something
interesting, but probably not.
T2 systems are those that can/are designed to (?) pass the standard
pen-pal style Turing Test, i.e. operating purely on syntax.
T3 systems can overcome the symbol grounding problem as they need some
kind of (non-computational) sensorimotor inputs/outputs, so as to ground
symbols in the real, physical world. They are behaviourally equivalent to
us - they can do & talk about the things we can, so there's no real reason
to say that they don't have a mind, apart from the normal other-minds
problem that applies between humans as well.
T4 and above just add to T3, i.e. the system can now look the same as a
human on the outside. T5 (I think) covers things like implementation as
well, e.g. human-like nervous systems, synapses, etc. T3 is the main one
to be concerened about. The rest can come later.
cheers,
Graham.
_ __ _ _ ___ ___
This archive was generated by hypermail 2.1.4 : Mon Nov 04 2002 - 22:08:39 GMT