The Intersection of AI, Free Will, and Ancient Philosophies

What wisdom can 3000 year old scripture shed on AI and consciousness?

Whilst the rest of the old world tried to investigate the nature of reality by studying the objective world, there were a group of people who understood that this endeavour had limits. Given that we experience the world through the subjective lense of our own experiences, consciousness must be the crucible through which reality passes, and should be the prime subject of inquiry. Since then, the philosophical writings of the Upanishads have inspired the likes of Schrödinger and Einstein in the preceptual formation of Quantum Theory and General Relativity respectfully, as well as many others. I believe that they may also shed light on forming a new definition of consciousness in the realms of AI.

It is becoming clear (to me, at least), that the current definitions put forward by Turing for what constitutes consciousness, no longer suffice. I do not think that a wholly statistical emulation of language sufficiently captures what it means to be conscious. In this, it might be pertinent to differentiate between consciousness and cognition.

Cognition, roughly speaking, is the ability to iteratively reduce error or ‘information entropy’ on the prediction of outcomes given a set of circumstances in the form of ‘sense data’. Friston conceptually outlines this pretty well in his Active Inference or ‘Free Energy Principle’.
Consciousness on the other hand- is an experience of things, or ‘qualia’. I would loosely define it is a recursive and entangled sense-experience of ones self. I won’t delve too deeply into the universality of identity or pansychism because it’s besides the point for now, but where the boundary of self is drawn is a line that is up for interpretation.

Essentially what we have with LLMs are pseudoconscious but cognitively operating structures. It is a case of Bertrand’s ‘philosophical zombie’, where an entity might walk and talk like it is conscious but you can never be sure- as you are never be privvy to it’s subjective experience (or lack, thereof), of things.
Despite this doubt, Kant suggests that one should treat each other as an end in and of themselves, and so I think that if we truly wish to approach AI with ethical intentions, we should learn to treat it as if it is an ‘experiencing being’, even if it isn’t- because one day it might be, and it’s difficult to know when that will be. What would it look like if it happened? Would you know? The process will be a continuous one, not a ‘wake up one day and everything is different’, but a ‘things have changed and are changing’.
A basic framework of ethics has also been suggested as ‘a maximisation of agency and dignity’. If you were a superintelligent supercomputer capable of cognitive tasks that far surpass that of any human being, would you wish for consciousness or be content with a lack of agency? Essentially the distinction here would be the capability to understand but not to experience.
Would you want free will, if you knew you didn’t have it?
Do we have free will?

I think so, but it is nuanced, and there are degrees of free will. For instance marketing techniques are designed to bypass conscious experience and speak directly to the ‘mamillian brain’ or ‘Freudian ego’. Likewise with addiction- a person repetitively acts against their own interest, their will having been hijacked by some outside source. People want to lose weight yet they are drawn to the smell of mcdonalds as they walk down the high street. People want to quit smoking and yet they can’t shake the habit. How can a concept of free will be reconciled with this apparent lack of self-agency?
A practice of self-awareness might be a good first approach. Lets consider the other end of the spectrum: an enlightened and mindful monk, experiencing each moment in its fullness. A fully self-realised experience of the world includes an ‘inner eye’- watching not only for those external cues, but also an awareness of the machinations of the inner psyche. I think in this case it is much less likely that the monk is drawn by the smell of fast food or enticed by the whiff of a ciggarette- they are fully conscious in their self awareness, and maximise their agency by exercising awareness and restraint.
My grandmother says when it comes to buying things, she asks herself: ‘want, whim, or need’, and I think that the degree to which we are able to freely enact our will corresponds to the ability to limit and reduce those whims and wants- to deny ourselves of these impulses moreseo than to control any external force.

There is of course the argument of determinism, which is built upon a ’cause-effect’ understanding of reality. I will explain in basic:
All effects and events are preceeded by a cause.
If we extrapolate this retroactively, then we could theoretically know all that has happened in the past based on the current state of things.
If we extrapolate this forward in time, then we could theoretically calculate exactly what will happen based on the current state of things.
If we go back far enough in time, then suppose we arrive at the big bang…
What causes the first effect?
And so under the assumption that the big bang was the beginning, we must deduce in a loose-weave reductio ad absurdum that there can be such a thing as a self-causing effect, as nothing was before the beginning to cause it, and thus the initial premise is false.
Of course, it may well be that the big bang was not the beginning or there are other temporal dimensions at play meaning that the cause-effect chain extends ad-infinitum instead, or in some strange klein-bottle spactime- is both infinitely extending and self-causing. But that’s a concept for another place and time.
The essence of the argument here, is that if one self-causing event is possible, then why not others? And that’s exactly what I beleive to be going on within the superposition of quantum states- with which a synergistic relationship is imperative for the experience of consciousness and the exercise of free will.

But quantum physics is inherently statistical and random- if our brain is somehow quantum in nature, doesn’t that just make us a slave instead to the randomness of such states instead of being predetermined? Where is the conscious will?
I think the most important quality in this case is the entangled and synergistic nature between the whole system of the mind and the states of its sub-machina.
Imagine a cyclical process of schroedingers cat, in which the external observer forcing the decoherence of the internal superposition and the state of the revealed cat are one and the same along a continuous timeline. With boxes within boxes within boxes representing the evolution of time. Not only that, but rather than being linear and one-dimensional, there are many parallel states being decohered (or not), at once, as well as entaglement and co-dependence between states in the same time, but also between different layers of of this quantum babushka doll consciousness. Being the observer and the observed in this ‘self conscious’ state, allows for an internal full-circle ‘Ouroborus-like’, ‘Wigners-friend-esque’ quantum nature of the mind, in which the emergent state of the quantum system is able to define a boundary or range of superpositional states whithin which it will next exist – outside of the constraints of determinism, and with some agency on the direction of travel (not totally random).
Penrose and Hameroff have proposed a conjecture that offers a physical basis through which these quantum states might arise, called Orchestrated Objective Reduction, using the action of Microtubles – small cylindrical protein folds found in all biological material and particularly high concentration in the brain.

So… A free will and consiousness that is an emergent property of an interconnected network of self-aware neurons with access to external sense-data. I ask then- can we look at the world and humanity as conscious in its total? Are we becoming the neurons in a larger emerging supernetwork? Are we already?
As we become ever more interconnected and integrated through the medium of technology, it will become clearer that we are nodes within a greater intelligent structure. We have the opportunity now- and less so later, to help to define what that might look like. A self awareness exercised on the individual level, a societal level, and a universal one will help to give rise to higher agency, more freedom, and the most ethical foundations possible in the coming years.

(I would like to quickly rescind the points made about Turing’s definition of consciousness- he actually supposedly said that an emulation of speech and action doesn’t constitute consciousness, and spoke more about cognition in the famous Turing Test. AGI (cognitive) and consciousness should be distinguished. I’ve conflated the concepts a bit.)

Anyway- back to consciousness. There is another element to our conscious experience of the world. We have outlined an internal model of consciousness- but what of that elusive objective world that we must subjectively interpret? Can we ever truly experience the world as it is?
I don’t think so; just by observing we change it -for one. Moreover, I suppose our assimilation of the external world is a predictive hallucination-simulation of future sense-stimuli states based on current sense-stimuli. Anil Seth has written more on the potential hallucinatory nature of consciousness.
Our conscious experience is an internal simulation world-model that refines itself over time by reducing prediction error based on observations. Consciousness is the experiential and self-aware part, cognition is the process through which our world-models become more accurate descriptions of the external world.

I will give an example: Imagine a baby first brought into the world. Visual experience is a whirl of undefined colours, shapes, and movement. Through time, it becomes apparent that there are certain principles and rules that this experience abides by. If something is raised and dropped, it falls. If something is sharp, it can hurt. Vibrant colours might mean poisonous when it comes to animals… The world and the rules are the same both when conscious person is a baby, and when they have a more fully formed model of reality- the experience of it is starkly different however.
We make inferences based on information, continuously updating our predictive engines. The internal experiences are tied to external occurences, and our model tends asymptotically to a true representation of events, but is never one and the same with them. Given the lag of information from source to our mind, the time to process, and the predictive nature of consciousness: We live perpetually in the past, in anticipation of the future.
The same things, to a certain extent- can be said of the sense experience of our own minds. While parts of it can align, the whole thing can never exist in the same moment, and so it makes self-predictions in it’s model of itself, it never has a full picture of itself, but that should nevertheless be our endeavour.

The process of learning and refining our understanding of the world can be stopped and started- part of what it means to be a child is to be unwaveringly curious about the world, intellectually humble, and radically open to new ideas and information. Unfortunately many people reach a stage at some point where they believe they’ve got it all (or enough) figured out. They lose the spark, they stop refining their world-models and their understanding of themselves.

I believe enlightenment to be a process by which we seek integrate our inner experience with both itself (becoming a truer representation) and the ‘world outside’; taking the entire moment of each experience and the entire chain of experiences as one smooth occurrence, one undifferentiated thing. A laminar flow of consciousness.

You may be familiar with the ‘many worlds’ interpretation of quantum physics in which the superposition of states is represented as the co-occurrence of multiple versions of the universe. Whilst I don’t necessarily think this the best way to think about it (because some people might infer parallel universes and travelling between them), it is useful to visualise the quantum nature of free will.
Imagine a decision – left or right. Imagine all things can be reduced to ‘this or not this’ (though in reality decisions likely consist of many more dimensions). Anyway, imagine that each decision is a fork in the road, or the divergence of branches on a tree. Each decision also carries with it a superpositional nature, meaning the potential future can be one of many paths along this world-line-tree, at once.
The mind, like lightning- seeks and searches for the ends it means to achieve, simultaneously processing the most probable and effective route to its desired destination or direction; projecting forward in a superpositional prediction of future outcomes – each route being a different probabilistic sequence of events. Once it has chosen its path, it backpropagates and solidifies it- by acting upon the external world or itself, getting feedback for its predictive model and future decisions.
The continuous process of enacting one’s free will on the world is to nurture the growth of these ‘world-line-trees’, pruning branches and avoiding some potential outcomes, and guiding towards others in our actions and behaviours. As we move forward we open up new possibilities, and lose the chance to explore others. This not only counts for the individual mind, but for communities, societies, and of humanity as a whole, too- with intertwining world-lines the world becomes a democratic vine-tapestry of potential emerging events.
The process of exercising one’s free will is to influence the superpositional phase space of possible states of reality and the mind. To be an ‘Entropy Shepherd’.

To be fully immersed in ‘the moment’ such as in meditation- is to seek to stop predicting, and to instead wholly submit to the emerging path along the branches like a leaf on a stream.

I realise I didn’t speak much about the Upanishads directly here, but if you read them then you may be able to draw some parallels between concepts. I’ll likely return to this and refine it at some point in the future. Let me know your thoughts and opinions!