Artificial Soul

On November 17, 2010, in Blog, irregular, Philosophy, Technology, by Amiya

If you can recreate each and every neuron and its interconnections, you can (re)create an artificial brain which is exactly like ours - conscious, thinking, self evaluating, intelligent and capable of dreaming and feeling; put appropriately - an  artificial life with an artificial soul!

Some may deem the statement as outrageous and  absurd but as I see it, this a perfectly logical and sound statement.  It is as clear as any mathematical axiom.  Whether or not this can be done is an entirely different question and if it is ever attempted in this form, it sure will be the worst way of achieving our goals in artificial intelligence (AI). Whether or not you accept this statement boils down to which view you hold towards everything - Physicalist or Dualist. In dualism it is claimed that mental phenomena are, in some respects, non-physical. Putting it bluntly all that the dualist are saying  is that there exists things in nature that can never be understood and explained (hence the term non physical). This to me is an outrageous statement completely absurd.   All I can say to the people holding these views is - wait and watch, the future is going to rock your world!!! Contrary to these beliefs, the other set of people accept the philosophies of physicalism. According to physicalism, the language of physics is the universal language of science and, consequently, any knowledge can be brought back to the statements on the physical objects[1]. (As a dualist would) Put bluntly it says that everything is material. Now do not take the high school text book meaning of material.  Material here is anything whose sum of mass and energy is greater than zero - so light is material as it has finite energy (but no mass). Other way of saying this is that in physicalism everything can be explained at least in theory.  This is not only a belief, it is the single most strongest source of motivation for a researcher and a scientist.

Mr. Robert lost the Turing test

The progress in science and technology over the last few centuries is momentous yet when it comes to AI the smartest machine we have has the intelligence of a retarded cockroach[2].  The field of AI is comparatively very new and research in the field was sparked by Alan Turing in a seminal 1950 paper - Computing Machinery and Intelligence which is considered to be the backbone of philosophy of AI. In the paper he proposed a test to measure the intelligence of a machine. The Turing test  explained in simple words is a test where the task of the machine is to imitate human. So if you are chatting to a bot (you will get numerous ones on Skype) and cannot figure out if its a human or a machine then the bot is said to have qualified the Turing test.  There are many variants of the test and its interpretations but all of them operate on the same basic principle. Sadly not even a single machine has qualified the Turing test.  Although providing a solid foundation for defining intelligence the Turing test does not say anything about consciousness and self realization. What is it that demarcates consciousness from intelligence? And when a machine running on solid mathematical foundations suddenly venture into the world of high level thinking is still an open question.  The branch of AI itself is broken into hundreds of subfields like machine learning, neural networks, fuzzy logic, Bayesian framework, search and optimization, symbolic logic so on and so forth... Each subfield is a mammoth with years of high end research put into them. It is not so surprising that there is not much co-ordination between these sub fields which could be stopping us from gather a larger view of intelligence and conscious machines. Also it might have happened that our approaches to this are wrong till now - may be we need to take a radically different approach. There are many brave people who are trying out the radical like Kwabena Boahen who is trying to reverse engineer the brain at the neuro biological level (he took the first statement quite seriously!),  Jeff Hawkins who is trying out a new framework (actually the only framework) on which to build intelligent machines, Henry Markram who works on the Blue brain project where he simulates parts of brain on a supercomputer. I am quite skeptical about their approaches.

As of now when brain is compared with computers, it turns out that the brain is actually quite slow in terms of how fast information flows in it, slow by many orders compared to a computer. But brains are massively parallel and vastly distributed which makes them unimaginably efficient, robust and fast in terms of processing data. It is hard to imagine what will happen if a brain can be designed on silicon which will work thousand times faster than our brain and have the same intelligence! Which brings along all the moral issues. If computers have life what rights will they have? Would not they outrun us once they are more intelligent than us? What is life, pain, happiness and love - everything surely is just bunch of electo-chemical reactions, but then where to draw the line? May be one day a computer will tell me the answers :0 I am waiting for that day.

PS: I myself am a fan of Bayesian statistics and Bayesian machine learning approaches.  I use these tools of AI to study the brain using functional MRI techniques.

[1] Otto Neurath [2] Michio Kaku

Related Posts Plugin for WordPress, Blogger... Print This Post Print This Post

Looking for something?

Use the form below to search the site:

Still not finding what you're looking for? Drop a comment on a post or contact us so we can take care of it!

Visit our friends!

A few highly recommended friends...