Turing Test

[Hans Blom, 960202]

(Bill Powers (960131.0800 MST))

An X behaves like a human if its visible outputs "looks like" what
would be generated by a human. So is X a human? This has always
been my problem with Asimov's Three Laws of Robotics, of which the
essential element is how a robot can recognize whether an X is a

Excellent point. The nature of the system producing the behavior
can't be determined just by observing the output of the system.

The remainder of your discussion proceeds to discuss a test of how to
establish whether an X is a human. But you forget one essential
point: such a test will often have to be made in a split second. Is
that something that just started to drop off the building a human or
a well-dressed straw man? Will the robot make a run for it, assuming
that it could make it and save the X? If you were a robot, how would
you decide? And why? Would probabilities enter your decision?

Why this point? Well, many of our decisions have to be made based on
insufficient information, often due to a lack of time (cognitive
overload) or an unwillingness to investigate further because other
things are deemed more important. The Test procedure that you propose
might just take too long to be much use in daily life. It just takes
too darned long! Although that is fine in science, where we have some
century-old conjectures that are still being studied. But are we
trying to model science or daily life?



[From Shannon Williams (951226)]

Mary Powers 951224--

language is far too complicated a phenomenon to get to in one jump.

I jumped in language because I am interested in modelling learning. If
the jump is even close to doable, it will be a much easier jump than to
jump to 'lower' levels. Give me a sketch of how we would program learning
in a 'lower' level and we can compare sketches.

why on earth are people chasing after
a verbal test before they know anything at all about modelling
the living systems that support the ability to be verbal?

Bill insists that all behavior is the result of control for a perception.
If this is true, then verbal behavior is no different than any other
learned behavior.