Kurzweil Debates Gelernter at MIT

Posted 6 Dec 2006 at 05:33 UTC by steve Share This

Rodney Brooks moderated a debate between Ray Kurzweil and David Gelernter on the subject of whether or not a conscious machine will ever be built. Not suprisingly, Kurzweil's position was "yes it will" and Gelernter's was "no it won't". Gelernter's position seemed to be that non-meat machines can only simulate real events but can't do anything real themselves. Apparently machines have to be made of meat to "really" do things or be conscious. Further, Gelernter believes any machine which "executes software" is not capable of creating "new nodes of consciousness". Gelernter also suggested that any non-meat machine that claimed it could think or feel was "programmed to lie". Both agreed that without a definition of consiousness and an objective way of measuring it, the whole debate is unscientific. MIT offers a recorded webcast of the event (unfortunately, only available in a proprietary RealPlayer format).

Consciousness in machines, posted 7 Dec 2006 at 18:00 UTC by jdietsch » (Master)

Consciousness, as in self-awareness, can be artificially programmed into a machine, or it can, with more difficulty be learned, as in a human.

Consciousness begins with discrimination between "me" and "not me." It starts when a child sucks his thumbs and toes and compares the results with sucking on a toy or on Mama. In one case is tactile feedback from the object being sucked (me). In the other cases, there is not. With appropriate sensors and programming, a machine, particularly a robot, could learn to distinguish between itself and other objects.

Jeanne Dietsch

Re: Consciousness in machines, posted 7 Dec 2006 at 19:51 UTC by steve » (Master)

The hardest part seems to be getting any two people to agree on what the word consciousness means. Just in this debate there were several definitions used. Kurzweil offered two which he distinguished as "apparent consciousness" and "real consciousness". His explanation of "apparent consciousness" sounds like what we would normally describe as social interaction. Gelernter defined it as internal, private mental states that have no known behavioral purpose. But there are dozens of other definitions and explanations of consciousness out there. Some philosophers will argue that the word and concepts the layperson associates with consciousness don't represent anything that really exists.

Along the lines of what you're suggesting, Christophe Menant has written some interesting papers about the possible path by which consciousness could have evolved from "self-awareness" or self-repesentation.

mutual exclusion.., posted 7 Dec 2006 at 20:30 UTC by slap.fish » (Journeyer)

With the current state of technology, I don't believe that programming and true learning can co-exist together. A radical shift in both hardware, and how 'programming' is encapsulated within it, is required before the stage is set for anything approaching a true analog of how actual organisms learn and achieve this rather amorphous concept of consciousness.

Gelernter's argument regarding simulation is a strong and persuasive one - I would tend to agree that with any current approach, only simulation of the intended behaviour is achieved - and even if 100% accurate, remains so.

That said, I think perhaps that it is, to some extent at least - narrowminded to conclude that only living organisms will ever be capable of the feat we refer to as 'consciousness'.. Who is to say what other mechanisms may come to pass that could one day have this potential - is it not arrogant to assume that 'meat computers' will retain their monopoly on the scale of complexity, interconnectivity and processing of sensory input (or whatever other brief summary you choose to represent the vital requirements) apparently neccessary for such emergence ?

Real simulations or simulated reality?, posted 7 Dec 2006 at 21:14 UTC by steve » (Master)

I liked Gelernter's comment that when your computer simulates rain you don't get wet. But I don't buy his argument that everything computers and software do is simulation. Is a word processor only a simulation or is it a real word processor? Is the email I receive every day just "simulated email" or is it "real email". I think the key question is whether consciousness is informational in nature like creativity and ideas or physical in nature like a dirt and rain. As Kurzweil pointed out, if your computer is creative (even if it's "simulated" creativity, whatever that is), real ideas come out.

A matter of location., posted 7 Dec 2006 at 21:55 UTC by MDude » (Journeyer)

"You can simulate a rainstorm and nobody gets wet,"

Only if no one is in the simulation with the rainstorm. If you can presume that simulated people in a simulated rainstorm arn't real, then you could presume that Pensylvania cannot contain conciousness because no one in PA is real, and a rainstorm in PA will get no real people wet.

missed intent ?, posted 7 Dec 2006 at 22:59 UTC by slap.fish » (Journeyer)

I think the context of the simulation argument has not come across properely - I'd expect he'd clarify by pointing out that the comment referred to computer models of real-world phenomena - i.e the 'real events' mentioned in the item above.

in this sense, email is not a real event - I'd expect he would counter that by pointing out it's just a computer model of a 'real' letter, and cannot physically drop on your doormat.

it would be nice if the resulting shock and annoyance of an unexpected emailed invoice were only simulated though ;) - I suppose this highlights that the information within is real enough, lol.

Never ever, posted 19 Dec 2006 at 05:59 UTC by marcin » (Journeyer)

I haven't listed to the debate, and so highlighting my ignorance; but it amazes me that, with all the history of human endeavour and acheivement behind them, people are confident to say that something will never be done.


See more of the latest robot news!

Recent blogs

30 May 2017 shimniok (Journeyer)
29 May 2017 wedesoft (Master)
28 May 2017 mwaibel (Master)
25 May 2017 AI4U (Observer)
25 Feb 2017 steve (Master)
16 Aug 2016 Flanneltron (Journeyer)
9 Jul 2016 evilrobots (Observer)
27 Jun 2016 Petar.Kormushev (Master)
2 May 2016 motters (Master)
10 Sep 2015 svo (Master)
Share this page