[ExI] The symbol grounding problem in strong AI

Gordon Swobe gts_2000 at yahoo.com
Sat Jan 2 14:50:35 UTC 2010

--- On Fri, 1/1/10, Stathis Papaioannou <stathisp at gmail.com> wrote:

> Right, I asked you the question from the point of view of
> a concrete-thinking technician. This simpleton sets about
> building artificial neurons from parts he buys at Radio Shack
> without it even occurring to him that the programs these parts run 
> are formal descriptions of real or supposed objects which simulate but
> do not equal the objects. When he is happy that his artificial
> neurons behave just like the real thing he has his friend the surgeon,
> also technically competent but not philosophically inclined,
> install them in the brain of a patient rendered aphasic after a stroke.

The surgeon replaces all those neurons relevant to correcting the patient's aphasia with a-neurons programmed and configured in such a way that the patient will pass the Turing test while appearing normal and healthy. We don't know in 2009 if this requires work in areas outside Wernicke's but we'll assume our surgeon here knows. 

The TT and the subject's reported symptoms represent the surgeon's only means of measuring the supposed health of his patient. 

> We can add a second part to the experiment in which the technician
> builds another set of artificial neurons based on clockwork nanomachinery
> rather than digital circuits and has them installed in a second
> patient, the idea being that the clockwork neurons do not run formal
> programs. 

A second surgeon does the same with this patient, releasing him from the hospital after he appears healthy and passes the TT. 

> You then get to talk to the patients. Will both patients be 
> able to speak equally well? 


> If so, would it be right to say that one understands what he is saying 
> and the other doesn't? 

Yes. On Searle's view the TT gives false positives for the first patient.

> Will the patient with the clockwork neurons report he feels normal while 
> the other one reports he feels weird? Surely you should be able to 
> observe *something*. 

If either one appears or reports feeling abnormal, we send him back to the hospital.



More information about the extropy-chat mailing list