After my recent remarks on the impossibility of true AI, a friend introduced me to Karl Friston’s Free Energy Principle (FEP). Put concisely, FEP uses the second law of thermodynamics to instantiate the rule of self-organisation in a system that “is at equilibrium with its environment” by minimizing its free energy. This creates order from chaos, behaviorally. Well, the second law is the guiding force behind the creation of life, and is how order comes from chaos in nature. This is how our planet was formed, in fact. Therefore, intuitively designing intelligence based on the principle which is generally used against theistic arguments of intelligent design, is a stroke of genius. However, the derivation problem is still alive and well, and I believe the FEP principle, far from bringing machines ever closer to the threshold of sentience, actually shows more staggeringly the difference between life, and the simulation of life. Let me explain why.
Sting in the Tail
The smallest social beings on this planet are insects, and funnily, they are also the closest to machines in many ways. Do they follow FEP? Not exactly. Autothysis is an act of suicidal altruism committed by some social insects when the colony is put at risk. For example, if a termite mound is
threatened, an individual will explode, releasing a harmful sticky substance to protect the colony. Bees are probably the most well-known example of this, as their stings evolved to deter an
impending threat to the colony at the cost of their lives. “The needs of the many, outweigh the needs of the few” as Spock would say. We have already crossed a boundary between the energy conservation of the individual, and of the community. The social aspect of the insects involved in autothysis (bees, ants, aphids) is not to be underestimated, as the recursive process of ordering – by way of importance – the society’s needs into a hierarchy is at once, in violation (singularly) and according (group) to FEP. Well, evolution has been known to use cheap, and dirty mechanisms to overcome the challenge of existence… but altruism is not the only cause of suicide in the living world. Dogs who have lost their owners have been known to harm themselves physically, both by starvation and drowning, to the point of death. Dolphins, and ducks, too. Again, it is important to note the social propensity of these animals. Here, we have cases where: (1) the need of the group; or more tellingly, (2) the need of the individual to escape suffering, have both caused intelligent, social creatures to commit themselves against FEP… which leads me on to Camus.
Camus and the Contradiction of Life
While the existence of social animals, and insects seemingly commit them break the principle of FEP, it does not appear that they knowingly do so. It is not their choice. They lack the introspective recursion required to place themselves outside of their current psychological situation, and into the possible world of another. Humans do not have this problem. Albert Camus in the Myth
of Sisyphus outlined what he believed to be philosophy’s most central concern: Why should we not kill ourselves? “There is only one really serious philosophical problem,” Camus writes,
“and that is suicide. Deciding whether or not life is worth living is to answer the fundamental question in philosophy. All other questions follow from that.” To make his point clear,
Camus uses the metaphor of Sisyphus to outline the absurdity of life. A man who must endlessly push a boulder up a mountain, until he becomes tired, and it rolls backwards.
Only for him to begin again. This, in the strictest sense, is what life is. Suffering, absent purpose given meaning, which is wholly perception based. There is no real, objective reason to continue pushing the boulder. Whether or not Sisyphus succeeds is, in the end, a truly meaningless task. The cold, and baron Universe will have no records. And, the unexamined life, as Socrates states, is not worth living. There is no clear answer to this problem.
Conclusion: The Problem of Derivation
While churning chaos into order through energy conservation is what intelligence emerges from, it appears the more social the animal, the more likely the FEP can be broken. And in the case of humans, the breaking of this principle can be a choice. To use a popular example, in iRobot the machine “Sonny” is given Asimov’s three laws:
1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.
2. A robot must obey any orders given to it by human beings, except where such orders would conflict with the First Law.
3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.
But also the ability to break them should he decide. (Interesting, right?) This is the jump from machine-to-human experience, and is not a derivation, but a random, chaotic mutation of information that can be taken in one of two ways: The hierarchical order of the information present, and the interpretation of that information. The ability to counter-act the principles which belie intelligence itself, is one of the first signs of intelligence in social animals. Our propensity to saturate our cognitive processes with poisons, particularly during socializing, evidences this further. There is a dark side to sentience which draws not only order, but self-destruction, from chaos, so that the process may begin again. A machine, because of the nature of the principles they follow, will never reach the stage where, unlike life, it is not deriving… but is randomly mutating. This is why adherents of AI are focused on randomness, and while they can try to draw from Quantum Mechanics to inject random-information into set-systems, this will only ever give the appearance of randomness; because the interpretation will
never be so. Only, at most, the input. Because, nothing random can come from a logical system.