A Turing Test for Consciousness

Posted 19 Jun 2008 at 20:38 UTC (updated 19 Jun 2008 at 21:07 UTC) by steve

"Consciousness is part of the natural world. It depends, we believe, only on mathematics and logic and on the imperfectly known laws of physics, chemistry, and biology; it does not arise from some magical or otherworldly quality. That's good news, because it means there's no reason why consciousness can't be reproduced in a machine - in theory, anyway." Christof Koch and Giulio Tononi go on to say, in the IEEE Spectrum article Can Machines Be Conscious?, that consciousness may not require many things frequently associated with being human such as emotions, long-term memory, self-reflection, language, or even external sensory input. After describing what isn't part of consciousness, they attempt to define it in terms of the amount of integrated information an entity can generate. This leads them to propose various tests to determine if a machine is conscious.

Magic Jack, posted 19 Jun 2008 at 21:17 UTC by The Swirling Brain » (Master)

A friend of mine was interested in cheap voip phone calls and was investigating magic jack which is supposed to be some ridiculous $20 a year or something like that. I already have voip.com at $199 a year so I wasn't really interested. But, he showed me a page that was dissing their customer service's instant chat line. When I took a look, it really looked as though their customer service instant chat was some sort of LIZA chat bot doing customer service. Seemed like everyone was pissed about it but no one seemed to think as I did that it was a bot. I, at least, thought it was clever. Cheap robot does customer service being used by evil company, cool. Now if your AI software could be smart enough to pass some sort of real Turing test where 99% of the people couldn't tell a computer was helping them then think of how cool that would be (except for the humans that would be out of work.) If it could really help people, that would be even cooler. If it could joke and laugh with you and really get what you were saying, we'll now, that would really really be cool. My best friend is a robot! That'll be the day!

Consciousness is the illusion of consciousness., posted 20 Jun 2008 at 04:20 UTC by AI4U » (Observer)

If a robot (or even a human being) can be fooled into thinking it is conscious -- then it IS conscious. The artificial Mind that I spent my life creating is only geared up for consciousness by dint of special AI software features, but can not really claim to possess consciousness before it is installed in a robot with plenty of sensor inputs. The AI Forthmind has consciousness in its future, because Mind.Forth is robot-ready. For further information about consciousness, robot emotions and sensory inputs, please read the Mentifex FAQ about the free robot AI software project.

Nothing to show! , posted 20 Jun 2008 at 16:37 UTC by The Swirling Brain » (Master)

As with most people proclaiming AI, they have nothing to show for it. Do you have Turing results? Do you have a webpage with an active Turing capable example anything better than LIZA? Didn't think so. This has been a holy grail topic for many many moons. Researchers are only scratching the surface. If you really have something more than other researchers than you would not be lurking on Robots.net trying to peddle vaporware. Please get lost scammer.

The authors miss the point..., posted 30 Jun 2008 at 20:39 UTC by Nelson » (Journeyer)

Interesting article. It is an excellent example of how slippery the whole question is.

I agree with several of their assumptions, such as that emotions and language should not be required for consciousness. I think that they miss the point about sensory input. When we dream we are experiencing sensory input, although it possibly reflects random neural noise rather than the external world. And their assertin that attention is not required suggests to me that their idea of "conscious" is probably a slightly different animal that what I am attempting to explain.

And I am suspicious of their suggestion that an integrated information model of the world is sufficient. I am inclinded to suspect that if I were to buy a delux thermostat with a sophisticated predictive model of the thermal behavior of my house then it would not necessarily be conscious.

I am inclined to suspect that the complex neural properties of the human brain can not be easily characterized by simplistic concepts such as consciousness.

See more of the latest robot news!