Meh. The whole thing really just illustrates a fundamental problem with our current conception of AI -at least as it manifests in such 'tests'. It is perfectly clear that the Eliza-like program here just has some bunch of pre-prepared statements to regurgitate and the programmers have tried to wire these responses up to questions in such a way that they appear to be legitimate, spontaneous answers. But intelligence consists in the invention of those responses. This is always the problem with computer programs, at least as they exist today: they really just crystallize acts of human intelligence into strict, repeatable procedures. Even chess programs, which are arguably the closest thing we have to computer intelligence, depend on this crystallized intelligence, because the pruning rules and strategic heuristics they rely upon draw on deep human insights that the computer could never have arrived at itself. As humans we resemble computers to the extent that we have automated our behaviour - when we regurgitate a "good how are you?" in response to a social enquiry as to how we are we are fundamentally behaving like Eliza. But when we engage in real conversation or any other form of novel problem solving, we don't seem very computer-like at all, the point that Craig makes (ad nauseam).
On Friday, June 13, 2014 5:20:16 AM UTC+10, John Clark wrote: > > On Wed, Jun 11, 2014 at 4:22 PM, <ghi...@gmail.com <javascript:>> wrote: > > > If the TT has been watered down, then the first question for me would be >> "doesn't this logically pre-assume a set of explicit standards existed in >> the first place"? >> > > My answer is "no". So am I a human or a computer? > > > Has there ever been a robust set of standards? >> > > No, except that whatever procedure you use to judge the level of > intelligence of your fellow Human Beings it is only fair that you use the > same procedure when judging machines. I admit this is imperfect, humans can > turn out to be smarter or dumber than originally thought, but it's the only > tool we have for judging such things. If the judge is a idiot then the > Turing Test doesn't work very well, or if the subject is a genius but > pretending to be a idiot you well also probably end up making the wrong > judgement but such is life, you do the best you can with the tools at hand. > > By the way, for a long time machines have been able to beautifully emulate > the behavior of two particular types of humans, those in a coma and those > that are dead. > > John K Clark > > > > -- You received this message because you are subscribed to the Google Groups "Everything List" group. To unsubscribe from this group and stop receiving emails from it, send an email to everything-list+unsubscr...@googlegroups.com. To post to this group, send email to everything-list@googlegroups.com. Visit this group at http://groups.google.com/group/everything-list. For more options, visit https://groups.google.com/d/optout.
