On Some Philosophical Defects of The Matrix Trilogy and AGI 1.2.0

Abstract Here we identify the central philosophical flaw of The Matrix trilogy, which is the central philosophical flaw of AGI, AI, and the popular notion that the mind is no more than the brain.

 

The Matrix – a 1999 film by the Wachowski and Warner Brothers- has sometimes been lauded for its philosophical depth (1). Less attention has been paid to it’s philosophical defects, which are significant.

The Matrix can be seen as a metaphor for Gnosticism, a Neo-Platonic form of Judeo-Christianity. Gnosis.org:

Human nature mirrors the duality found in the world: in part it was made by the false creator God and in part it consists of the light of the True God. Humankind contains a perishable physical and psychic component, as well as a spiritual component which is a fragment of the divine essence. This latter part is often symbolically referred to as the “divine spark”. The recognition of this dual nature of the world and of the human being has earned the Gnostic tradition the epithet of “dualist”.

Humans are generally ignorant of the divine spark resident within them. This ignorance is fostered in human nature by the influence of the false creator and his Archons, who together are intent upon keeping men and women ignorant of their true nature and destiny. Anything that causes us to remain attached to earthly things serves to keep us in enslavement to these lower cosmic rulers. Death releases the divine spark from its lowly prison, but if there has not been a substantial work of Gnosis undertaken by the soul prior to death, it becomes likely that the divine spark will be hurled back into, and then re-embodied within, the pangs and slavery of the physical world.

According to the above account of Gnosticism, human beings are trapped in the world as we know it by a process that involves their incarnation in a physical body, and will continue to exist in this state of entrapment for lifetime after lifetime unless they come to an understanding of this situation and consequently reject the reality of this world and its values. A central figure in Gnosticism is Jesus of Nazareth, who taught contempt for, and rejection of, the physical world, and was on a mission from outside this domain to free people from the physical world. The Matrix can readily be taken to represent the physical world that we must escape from, and the Neo-character to be representative of Jesus of Nazareth. The Wachowskis make have had a figure such the Buddah in mind, more so than Jesus, but Guatama lacked the power over the physical world that Jesus possessed, as did every religious leader in history other than Jesus. Jesus could turn water into wine, walk on water, raise the dead… and was able escape the strictures of death itself. Jesus on his death:

No one takes it (my life) from me, but I lay it down of my own accord. I have authority to lay it down and authority to take it up again.

In other words, he could do what he wanted in “The Matrix”, and was immortal. The film can also be seen as championing the idea of Artificial Intelligence, since the The Matrix system is the work of computers, and computers have taken over the world, imprisoning human beings in life-support pods, and reducing them to batteries. The film has many philosophical flaws (including the the unnecessary pre-establish harmony there is between the perceptions of those in The Matrix, the possibility of free-will in the confines of a deterministic computer program, the peculiarly American idea that violence is a means of enlightenment/gnosis) but chief amongst them, is the errors of AGI Artificial General Intelligence = the intelligence of a machine that could successfully perform any intellectual task that a human being can) and Strong AI = the thesis that machine that could successfully perform any intellectual tasks that a human being can coupled with the thesis that they are capable of consciousness.

So what are the prospects for “strong” AI, which I describe as machine intelligence with the full range of human intelligence? We can meet the hardware requirements. I figure we need about 10 quadrillion calculations a second to provide a functional equivalent to all the regions of the brain. IBM’s Blue Gene/L computer is already at 100 trillion. If we plug in the semiconductor industry’s projections, we can see that 10 quadrillion calculations a second will be available for $1,000 by around 2020. (2)

But there are in reality tasks that humans can perform that machines cannot possibly perform, and moreover, consciousness is dependent of the ability of humans to perform these tasks, from which it follows that there is not, nor can there be, any such thing as AGI, let alone strong AI. These machine-unfriendly tasks concern a special class of problems known as “N-P-Complete” problems. Problems that can be solved by computers in an efficient amount of time (where efficient means that as the problem-size grows arithmetically, the number of steps required ti solve the problem does not grow exponentially) we classify as “P” for polynomial time; problems whose solution can be verified in any efficient amount of time we classify as N-P for non deterministic polynomial time; all problems in P are also N-P, but not all problems in N-P are in P. N-P-Complete problems are both in N-P and not in P, and the classic example is The Travelling Salesman, which involves a salesman who must, after starting from a home-city, visit a number of cities exactly once before returning the home city for a certain cost. To see that there is no efficient way to solve this problem, let the home city = the halt state of a computer, let every other city = an atomic instruction of a computers program. If an instruction can be executed, assign a cost of 1, if an instruction cannot be executed, assign a cost of 2. If and only if the salesman can complete a circuit that visits every city exactly once for a cost of the number of cities, there is some computer that will halt when run with some input. If the number of cities is the same as the number of instructions in the program of the machine evaluating the TSP instance, the evaluation is a self-evaluation.

And if we assign a cost of 2, rather than 1, then this machine is required to determine that an input can only be run by a more complex program than itself.

This proves that there are some problems that a computer, no matter how powerful cannot possibly solve in an efficient amount of time. It also proves that the problem of whether there are some problems that a computer can’t solve in an efficient amount of time (called “P versus NP”) cannot be solved in any amount of time by a computational intelligence, and yet you and I have just solved it. This proves that AGI is false because it proves that there are intellectual tasks that a human intelligence can perform that a classical computational intelligence cannot.

To see see why humans are conscious, and computers cannot possibly be conscious, consider that Factoring is not in P. Factoring = the problem of finding the prime factors of an integer. While it doesn’t take long to factor 15 or 21, imagine trying to factor the 200 digit integer

27997833911221327870829467638722601621070446786955428537560009929326128400107609345671052955360856061822351910951365788637105954482006576775098580557613579098734950144178863178946295187237869221823983

You can easily check that it divides evenly into the primes

3532461934402770121272604978198464368671197400197625023649303468776121253679423200058547956528088349

and

7925869954478333033347085841480059687737975857364219960734330341455767872818152135381409304740185467

But although it takes a pocket calculator a spit second to do the multiplication it would take a single 2.2 GHz computer roughly 75 years to do the division. And this is a theoretical problem, not a practical problem, for every Travelling Salesman problem can be transformed into a Factoring Problem by identifying TSP problems with integers based on the number of cities in the problem = the number instructions in a computers program. Factoring expressed in terms of whether a number has a prime factor not greater than n is equivalent to Travelling Salesman expressed in terms of whether there is cycle whose weight not greater than n, and a classical computer can’t factor a number n such that n is greater than the number of instructions in its own program… But Shor’s Algorithm for a quantum computer permits the factoring of integers in a efficient amount of time, from which it follows that the limit there is on classical computers in virtue of which they can’t solve certain problems in an efficient amount of time, nor solve certain problems in any amount of time, is inapplicable to quantum computers. It follows that the human mind is, not simply a classical computer, but has quantum computational elements.

We ought not to talk about quantum computers without also saying something also about quantum physics. The difference between classical and qauntum physics be most simply expressed by reference to the the 2-slit experiment, in which particles are shot through a pair of slits and registered on a screen. If detectors are not attached to the slits -if there is no “2-path information”- then the pattern on the screen shows that the particles have the form of a wave, rather than isolated entities, and so that the particles in a sense travel through both slits at once, interfering with each other:

This difference between the one form of the experiment and the other gives rise to the notion of a “wave-function”, a mathematical object that describes all of the possible locations of a particle prior to measurement by describing the wave. Detecting a particle -measuring the particle¬† en route- is said to “collapse” the wave function. Anytime we are conscious of something, and thus find it to be one way or another and not both, this is because of an underlying collapsed wave-function. But behind the content of consciousness, which is derived from collapsed wave-functions, and possesses the duality of the measured particles in the 2-slit experiment, is the container of consciousness. Sartre challenged us to try and detect the mind independently of some impression, some perception, independently of some content, and he observed that this is impossible (4.) In the same vein, Wittgenstein analogizes consciousness to the eye, pointing out (5) that the eye does not see itself because it is not a part of the visual field:

Where in the world is the metaphysical subject to be found?
You will say that it is exactly like the case of the eye and the visual field.
But really you do not see the eye and nothing in the visual field allows you to infer that it is seen by an eye.

The invisible container of consciousness -the mind, the subject- possesses the singularity of the unmeasured particles in the 2-slit experiment. It follows that the container of consciousness is not, as strong AI proponents like Ray Kurzweil maintain, something that emerges from complexity (plurality), and therefore might be a property of a sufficiently complex machine. Rather, it is something that emerges only from simplicity (singularity), and cannot be a property of a machine. There will never be a conscious machine, nor is there any possibility that computer-intelligence -which is the passive aspect of an intelligence that is both active and passive- will outstrip human intelligence (6). Even the most sophisticated computer is moronic compared to even the least sophisticated human mind. A setup like that in the The Matrix it follows is impossible, yet the theme that the world is dominated by machines, not in the sense that machines are like the task-masters of a Orwellian regime, but in the sense the worst aspects of ourselves -the passive, material, selfish aspects that seek only self-gratification and to establish physical monuments to themselves- predominate in modern over the active aspects is a very valid one.

Download pdf

References

(1) Irvin, W (2002), The Matrix and Philosophy: Welcome to the Desert of the Real (Popular Culture and Philosophy)

(2) Kurzweil, R (2005), On My Mind Long Live AI

(3) Shor, P (1994), Polynomial-Time Algorithms for Prime Factorization and Discrete Logarithms on a Quantum Computer

(4) Sartre, J (1943), Being and Nothingness: An Essay on Phenomenological Ontology

(5) Wittgenstein, L (1921), Tractatus-Logicus-Philosophicus

(6) Kurzweil, R (2005), The Singularity Is Near: When Humans Transcend Biology