Saturday, 7 April 2018

Computationalism: the Threat to Buddhism in China (and everywhere else)

In a recent comment on the article on intentionality,  a Buddhist from China states that 'Currently, Chinese Buddhism, traditionally a bastion of Mahayana, is booming but also undergoing tremendous intellectual challenge from computationalist philosophies (similar to the issues you addressed in some of your other posts)..'

So we Buddhists need to disprove computationalism,  (aka the 'Computational Theory of Mind), which 'holds that the mind is a computation that arises from the brain acting as a computing machine...  So the computational theory of mind is the claim that the mind is a computation of a machine (the brain) that derives output representations of the world from input representations and internal memory in a way that is consistent with the theory of computation'.

Computationalism is a form of materialism (in fact is the logical foundation of all forms of materialism, but more of that later). It denies any spiritual or transcendental dimension to human life. Everything is mechanistic.

The popularity of the Computer Theory of mind seems to be linked to the intensity of hype surrounding Artificial Intelligence.  We are currently going through the peak of another Artificial Intelligence (AI) hype cycle (an 'AI summer').   AI-hype assures us that computers are now so powerful that machine simulation of the human mind is imminent (as it has been for the past 60 years).  

Hype and reality

These AI-hype bubbles seem to occur every decade. In the 1950's the hype was about 'electronic brains' which could do dozens of calculations per second and would soon surpass humans in all mental functions. Then in the sixties, computers were on the verge of being able to understand human language, enabling true nuanced translation between languages, rather than clunky word-substitution. 

In the seventies, the microprocessor revolution was predicted to enable  massively parallel brain-emulators to be built, bringing about the long-promised AI revolution.  The eighties were the decade of expert systems, which would replace human technical and medical professionals by omniscient computers. 

In the nineties, we were assured that neural nets would at last provide the solutions to the failures of previous hype cycles.  In the noughties, the interconnection of millions of computers via the internet would finally give enough power to produce a global brain to surpass the human mind, and solve the problems that neural nets couldn't.  The current decade's hype cycle concerns robotics, which promises us that connecting computers to sensors and actuators will at last solve the Hard Problem  and lead to true artificial sentience, within a year or two at the very most - and this time it's really going to happen!

Although 'AI-hype' and 'computationalism' aren't totally synonymous, there is a close relationship (known as 'strong AI').  To see why AI always fails to deliver, and computationalism is a non-starter, and always has been, and always will be, we need to consider the following factors:

Computationalism implies that all mental processes can be modeled by suitable combinations of the the members of the instruction set of a general purpose computer, the instruction set being functionally identical to a Turing Machine (TM).     

So, if we can demonstrate any functions and activities of the mind which are beyond the capabilities of a Turing Machine (and hence all its derived instruction sets), then we have demolished the foundations of all forms of computationalism*.

When we examine the components of the TM in detail, we find that none of them...

(i) are capable of holding meaning. They do not possess any semantic capabilities, or 'intentionality' to use the technical term.

(ii) are capable of registering qualitative states ('qualia') such as sensations of pleasure, pain etc.

And when we check the basic repertoire of operations that the TM can carry out (and the derived and hence equivalent instruction-sets of general purpose computers) we fail to find any combination of these operations that can operate on the components to produce semantic or qualitative phenomena.  Thus two major functions of the sentient human mind are beyond the capabilities of machines.  These arguments are explained in more detail here, here, here and here.

This is why the Mother of all Algorithms is itself NOT an algorithm, and never could be programmed.

Tail wagging the dog?
The Turing Machine was, and is, an object of consciousness, a mathematical thought-experiment  first imagined by Alan Turing. For many years it remained just that - a mathematical structure with no physical instantiation, existing only in the minds of theoreticians.

Computationalism claims ontological primacy for this disembodied object of the mind, making it more basic than everything else, including the inventiveness that first created it.  The computationalists claim (explicitly or implicitly) that the Turing Machine is the foundation of consciousness, yet the TM is itself a product of consciousness. 

Buddhists would believe that there is a phenomenon even more fundamental than the TM, since the TM is an object arising out of thought.  That more fundamental phenomenon is the experience of creating, imagining and thinking about the TM in the first place.    

As the Kadampas put it: 

'..If we check, we can see that we cannot in fact separate out the objects of our thoughts from the thoughts or awarenesses holding them, any more than we can separate out a wave from an ocean or a reflection in a mirror from the mirror itself. There is no such thing as an object not known by mind, which is the definition of object, “known by mind”.

Can you even think of an object that is not known by mind? There is no world outside of our experience of the world. What is going on for you right now, for example, is your experience of what is going on – if you go looking, you cannot find anything going on out there. Your whole world cannot be separated out from your experience of the world – you cannot point to any world outside of your experience of it. As soon as you do, you’re experiencing it.

Waves are the nature of the ocean, not outside the ocean. Appearances are the nature of the mind, not outside the mind...'

Demolishing all forms of materialism
The Church-Turing-Deutsch Principle (CTDP) states that a universal computing device can simulate every physical process.

If the CTDP is true (and it's never been disproved) and computationalism is false (which seems to be the case) then all other physicalist and materialist theories of the mind are consequently false.  It follows that the non-computationalist mind exists outside the scope of the CTD, and hence it is outside the scope of all physical, mechanistic and material theories, no matter how they may be expressed.  By defeating computationalism we have removed the foundation of  all other physical/materialist explanations of the mind.  

Our next (and most important) task is to understand the true nature of this non-physical mind.

* Turing machines and instruction sets
Turing completeness is the ability of a system of instructions to simulate a Turing machine. A programming language that is Turing complete is theoretically capable of expressing all tasks accomplishable by computers  

In computability theory, a system of data-manipulation rules (such as a computer's instruction set, a programming language, or a cellular automaton) is said to be Turing complete or computationally universal if it can be used to simulate any Turing machine.

Although physical TMs can be built, they are generally demonstrated to students in the form of emulations using standard programing languages.     Hence, it works both ways,
TMs can emulate computer instruction sets, and computer instruction sets can emulate TMs.  

It's quite remarkable just how few instructions are required to provide full and complete computing capabilities - fewer than twenty:  SET, MOVE, READ, WRITE, ADD, SUBTRACT, MULTIPLY, DIVIDE, AND, OR, XOR, NOT,  SHIFT, ROTATE, COMPARE, JUMP, JUMP-CONDITIONALLY, RETURN

Increasing the power of a computer cannot expand its functionality beyond this instruction set, since any general purpose computer can simulate any other computer, albeit slowly.  Hence if machine sentience ('strong AI') were indeed possible, it could have been demonstrated in the 1950's!  Throwing newer, faster more powerful hardware at the problem won't produce any fundamental breakthrough in functionality, just as increasing the working temperature, spin-speed and capacity of your washing machine won't suddenly make it capable of mowing the lawn.

See also Buddhist Philosophy,  Why Materialism is Crap 

SWOT Analysis - Strengths. Weaknesses, Opportunities and Threats to Buddhism

No comments: