Replies to This Discussion

Permalink Reply by Aaron Claassens on September 15, 2009 at 12:19pm

Noel Sharkey states that the mind could be a type of physical system that cannot be recreated by a computer. Does anyone smell a bit of Cartesian dualism? That the immaterial mind and the material body, while being ontologically distinct substances, causally interact. Mental events causing physical events, and vice-versa. How can an immaterial physical mind cause anything in a material body, and vice-versa? This has often been called the "problem of interactionism".

In my view the "problem of interactionism" is false. When we talk of the "computational theory of mind" we are not stating that the brain could be run on a dell desktop, but merely pointing out that "some sort" of computational processes within the brain gives rise to mental thoughts. A neural correlate of consciousness exists and is logical. It may be that are current theories of computation are not sophisticated enough at this current time to emulate the process, however minds must somehow be the product of brains.

I see evidence of this every time I’ve watched a neurosurgery procedure. Damage or remove a small complex physical brain portion and the mind can suffer. Yes, the mind could be a type of emergent physical system however it owes it’s substantiation to an object which exists in our 3 dimensional world, the brain. In my view the mind is just the world’s hardest software problem.

The brain is not of infinite complexity however it is extremely complex. A total understanding will not be achieved through one simple breakthrough. Our ability to scan the brain is growing rapidly and we are beginning to see the internal connection patterns. Our mathematical models of brain neurons are getting better and IBM have recently modelled a small slice of rat brain on a supercomputing substrate. This is the first time in history for AI projects that we have had access to tools of such sophistication and they are exactly the types of tools which are needed to reverse engineer the most complex object we know of in the universe.

Arguments suggesting that true AI is a pipe dream are like stones in a stream. The water runs around them. Are understanding of neuroscience is doubling yearly. To quote Professor Markram of the IBM Blue Brain project – “100 years of neuroscience discovery has led to millions of fragments of data and knowledge that have never been brought together and exploited fully. ‘Actually no- one even knows what we already understand about the brain'”. Our biggest hurdle in creating AI will be bringing together what we already know about the brain into one comprehensive system. We are making progress with supercomputing and I believe massive data mining projects will be needed as we move forward.

We know the brain processes information using electrochemical methods 10 million times slower then electronics. So how are brains superior? A point that is often overlooked is they make up for this by being three-dimensional. Brains are massively parallel. Every interneuron connection computes simultaneously, so you have 100 trillion things going on at the same time. This is the direction we will need to go in. Right now, chips, even though they’re very dense are flat. Once the current paradigm of Moore’s law runs out of steam we will move to the 3rd dimension. Chips that are massively parallel and based on biologically inspired models, which will be devised largely by our understanding of how the brain works.

The first AI will be a profound achievement for humanity, akin to landing on the moon or building the Panama Canal. After building the first mind we will build better ones. Mind Design will be an art form. Just like the myriads of species on earth, we will have myriads of different AI minds intended for myriads of purposes. Some will embed themselves into mobile substrates of various forms while others will exist at fixed locations, others inside us. Once these minds design there progeny will be the start of the Singularity. An intelligence explosion beyond are current comprehension.

Thank God Noel Sharkey was not born during the development of aviation or planes may never of taken off. If we can’t engineer it then we just don’t yet understand it.

• ▶ Reply

Permalink Reply by Aidan J on September 18, 2009 at 8:59pm

Whoa!

I offered the piece as a great discussion on a scientists POV of the AI argument and the ensuing discussion amongst us ordinary folk!

I'm not sure how many neurosurgeons you know or the many procedures you have watched but the only qualified person I know tells me he does not believe that brain damage equals mind damage, only its ability to interact with the world as it used to...

To those who keep quoting "Moores Law" - go and check it out, its not a law, just a prediction by Gordon Moore in 1965 which has actually been altered since by Intel empoyees. It means precisely zero as a scientific belief, it really belongs to the IT marketing sphere and now that we have made the move to multi core processing,it is pretty much nearing the point of 'defunct' as a concept.

Don't get me wrong I'm not arguing against the notion of AI but as you will likely know already there are many opinions on where/how far it will go. I too agree the AI breakthrough will be profound for mankind, as profound as the discovery of life somewhere other than on this planet, I hope I am around to see both.

I do however believe that great thinkers and innovators like Sharkey have our best interests at heart with articles like the one I posted, he is certainly well qualified to comment on the sphere of robotics and AI, I dare say moreso than most of us! (http://en.wikipedia.org/wiki/Noel_Sharkey)

The argument over AI and the extent it can take us to will go on for the foreseeable but for us to believe right now that 'such and such' will definitely happen is misguided because we simply don't know yet. We might as well just pick any religion at random and follow its beliefs. Someone tells us to believe it and it sounds nice so we will...!

• ▶ Reply

Permalink Reply by Aaron Claassens on September 19, 2009 at 3:33pm

I agree with you that brain damage does not always result in a damaged mind, however this is due to the remarkable neuroplasticity the brain has in it's ability to repair itself after some forms of traumatic injury. My remark was merely stating that sometimes extreme traumatic injury to the brain can result in severe injury to the mind. Some cases even resulting in brain death.

If your brain is injured enough you will have no way to interact with the world. This alludes to the mind being a product of the wetware which is the physical brain. If Noel wants to think of the mind as a physical system then his best bet will be trying to find it within the brain. This is exactly what AI researchers are trying to achieve. Of course I may be wrong, some modern philosophers assume consciousness to be the foundation of reality which brings into being the physical universe through observation of quantum phenomena. Quantum mechanics seems to make materialistic thinkers the "flat earthers" of today. Maybe science, which was long believed as materialism's best friend, may actually be the best friend of idealism in its true colors. There are however many competing interpretations of quantum mechanics which negate this view.

Moore's law is not a physical law as such, you are correct. We can extend Moore's law to encompass the next paradigm of computing so as a "general concept" it still continues to hold strong. A better description is Kurzweil's Law of Accelerating Returns which applies Moore's to not only information technology but other spheres of accelerating change such as genetics, robotics and nanotechnology. Having been an employee of Intel myself, I know that it's roadmap for processing power should continue until 2022, at which time key features on chips will be about 4 nanometers in size. Processing power will still continue to grow through new paradigms of computing such as 3D molecular computing, quantum computation, spintronics and even other more exotic forms such as DNA computing.

One scientist to study the limits of Moore's law is MIT professor Seth Lloyd. He spent several years working out the ultimate limits to computing with a 1 kg mass (remember with today's physics - things may change if history is any guide to the future). Lloyd shows how the potential computing capacity of a kilogram of matter equals pi time energy divided by Plank's constant¹. This equation generates an extremely large number: about 10 x 10^50 operations per second. It is estimated that this "ultimate laptop" could perform the equivalent of all human thought over the last 10 000 years (that is, ten billion human brains operating for 10 000 years) in one then-thousandth of a nanosecond². How such a laptop is built is an engineering challenge for post humanity. The example illustrates that yes there are limits to Moore's but there not very limiting. And remember this calculation is ignoring even more speculative forms of computation such as femto and pico technology (computing with sub atomic particles).

In relation to the general arguments about true AI becoming functional, I agree we don't know yet. We are however making small breakthroughs all the time which leads me to strongly believe the whole enterprise will materialise.

Ref:
1. http://puhep1.princeton.edu/~mcdonald/examples/QM/lloyd_nature_406_...
2. The Singularity is Near by Ray Kurzweil pg.134


Aidan J said:

Whoa!
I offered the piece as a great discussion on a scientists POV of the AI argument and the ensuing discussion amongst us ordinary folk! I'm not sure how many neurosurgeons you know or the many procedures you have watched but the only qualified person I know tells me he does not believe that brain damage equals mind damage, only its ability to interact with the world as it used to...

To those who keep quoting "Moores Law" - go and check it out, its not a law, just a prediction by Gordon Moore in 1965 which has actually been altered since by Intel empoyees. It means precisely zero as a scientific belief, it really belongs to the IT marketing sphere and now that we have made the move to multi core processing,it is pretty much nearing the point of 'defunct' as a concept.

Don't get me wrong I'm not arguing against the notion of AI but as you will likely know already there are many opinions on where/how far it will go. I too agree the AI breakthrough will be profound for mankind, as profound as the discovery of life somewhere other than on this planet, I hope I am around to see both.

I do however believe that great thinkers and innovators like Sharkey have our best interests at heart with articles like the one I posted, he is certainly well qualified to comment on the sphere of robotics and AI, I dare say moreso than most of us! (http://en.wikipedia.org/wiki/Noel_Sharkey)

The argument over AI and the extent it can take us to will go on for the foreseeable but for us to believe right now that 'such and such' will definitely happen is misguided because we simply don't know yet. We might as well just pick any religion at random and follow its beliefs. Someone tells us to believe it and it sounds nice so we will...!

• ▶ Reply

▶ Reply to Discussion