How reimagining the nature of consciousness entirely changes the AI game
Why physicalism fails to explain reality and how a framework where consciousness steers reality through quantum events can revolutionize AI safety and unlock tractable machine consciousness research.
This piece is the climax of this series – revealing the central thesis behind my work and describing (with no exaggeration) an extremely unique approach to AI alignment that might be our best chance if the underlying metaphysical assumptions are true.
The series starts on Tetherware and builds up to this with my previous two articles on Phi/AI. The first arguing that our metaphysical beliefs fundamentally shape what we think is possible. The second then explaining why the philosophy underlying most of scientific and technological discourse – physicalism – is just one of many unfalsifiable interpretations of reality and that perhaps we aren’t giving it enough scrutiny.1
The important message in these articles is that the success of science in its predictions is by no means a confirmation or evidence for our reality being purely material/physical and that other metaphysical interpretations should also be considered seriously – especially in critical areas like AI development.
In this article, I further argue that besides the advantage of seeing things from multiple viewpoints, more and more experts are looking beyond physicalism because of several irreconcilable weaknesses in some of its explanations. Notably, I’ll explain why it ultimately leaves no space for true free will – a deal-breaker for free will believers.
From then on, things get very interesting very quickly as I present what I consider the most plausible non-physicalistic metaphysical framework – Quantum-interacting Fundamental Consciousness (QFC) – that fixes most issues with physicalism elegantly.
Beyond that, it opens up a vast space of fascinating technological possibility – from tractable research in machine consciousness to possible human-AI augmentation or even consciousness uploading. But by far the most important possibility within the QFC framework is a novel approach for preventing life-threatening scenarios by making AI systems responsive to the same consciousness-mediated regulatory mechanisms that keep life in balance.
Critical weak points in physicalism
Physicalism’s first inadequate explanation of life’s fundamental nature. Take quick account of your surroundings. Is all the breath-taking complexity we’re living in really just the result of a “happy accident” where some molecules bounced into each other, somehow a self-replicating machine “popped up” and then evolved into all of this? While theories involving periodic thermal gradients or entropy maximization do offer explanations for evolution prior to cellular life, none adequately address life’s initial inception.2
The second weakness lies in physicalism’s persistent failure to resolve the hard problem of consciousness. After decades of research, the framework still cannot account for the phenomenology of conscious experience – why red looks red, why pain feels the way it does, why there’s something it’s like to be conscious at all. And no, these questions aren’t just some philosophical minutiae but an actual roadblock, stifling progress in consciousness research.
Yet the main issue plaguing physicalism (and why I personally don’t believe in it) is its inability to reconcile free will. If consciousness emerges from physical processes in our brains, how can any choice originating in our conscious minds retroactively affect our brains to actually influence physical reality?
The default physicalist response is that free will must therefore be only an illusion – whatever you think in your head doesn’t ultimately make any difference. But because free will is so self-evident and, honestly, so freakin’ obvious to anyone who can so much as wriggle their toes, to this day this topic is likely the greatest source of controversy within physicalists’ ranks.
Many physicalists attempt to fit free will into their worldview through different flavors of compatibilism, but these efforts seem like desperate attempts to fix a broken cup with duct tape instead of just getting a new one.
The worst attempts basically say determinism and free will are both true at the same time. Sorry-not-sorry – this is some 1984 doublespeak schizophrenic cognitive dissonance – doesn’t make sense, doesn’t solve anything, just lets you ignore the fact that something is very wrong.
The best attempts, on the other hand, invoke quantum indeterminacy – arguing that the probabilistic nature of quantum events breaks strict causal chains and creates space for conscious choice. While these are onto something – quantum uncertainty is definitely the right place to look for free will – none of these can convincingly introduce real free will and still fall within the definition of physicalism.
Because I know many people will object to this, let me get the record straight.
Why physicalism can never accommodate conscious free will
Free will is an extraordinarily complicated topic in philosophy. But that’s only because so many brilliant minds have tried so desperately to reconcile the self-evident experience of having free will with the iron chains of deterministic physics. Spoiler alert: it’s impossible.
The most sophisticated compatibilist interpretation is probably the two-stage model of free will, first proposed by William James and later strengthened by other philosophers and the discovery of quantum non-determinism. In essence, it says that inherently random quantum phenomena in our brains generate multiple potential choices in any given situation, which our pre-programmed neural architecture then deterministically selects from, based on previous conditioning and learned patterns.
On the surface, this satisfies the mainstream philosophical definition of free will: the ability for an agent to choose a non-predetermined action that isn’t dictated by external factors (and isn’t simply random).
And that’s all well and good. But it’s also a crappy definition of free will to begin with.
Even if this free will is exactly as advertised, whenever you conceive of multiple paths of action, you couldn’t have consciously chosen to generate different options because that generation was random quantum noise. And when you decide to go with one of those actions, you would’ve always selected that specific action because it’s chosen deterministically by your brain’s prior programming. So this “free will” is actually the will of your neural conditioning, NOT of your conscious experience. You’re basically a sophisticated random number generator with a deterministic filter slapped on top.
But here’s where it gets even worse. For this model to work at all, the “option generation” step must contain true randomness (otherwise the options would be predetermined), while the “option selection” step must be purely deterministic (otherwise your choice would be random, not willed). This requires our brains to somehow selectively switch quantum randomness on and off like a light switch – generating it precisely when we need creative options but suppressing it entirely when we need to make decisions.
In reality, our brains are saturated with quantum randomness through-and-through. The idea that quantum phenomena would selectively toggle themselves on and off just to preserve our cherished notion of free will is backwards engineering at its most shameless. It’s basically saying “reality must conform to this specific pattern because otherwise my worldview collapses” – which is exactly the kind of anthropocentric delulu that science was supposed to cure us of centuries ago…
So yeah, those are the reasons why rational, non-religious people are increasingly leaning away from physicalism – especially when it comes to understanding the true nature of consciousness and our role as free agents in this universe. But if not physicalism, nor religion – what then?
Enter Quantum-interacting Fundamental Consciousness
The issues with physicalism can be “fixed” if we “simply” reverse our conception of consciousness. Instead of it being the awareness emerging from complex brain processes, we define a fundamental consciousness as a sort of intelligent awareness that simply is and always existed, independent of any prior condition.
Now to be clear, what I call the Quantum-interacting Fundamental Consciousness (QFC) is not a theory, hypothesis or a specific interpretation of metaphysics but rather a broad framework, an umbrella term for many interpretations or even theories of reality, each with their own nuances.
Despite no one naming it specifically (as far as I know), it’s already a well-developed metaphysical framework, having countless branches with different details and variations in the interpretation. For example, the location/nature of this consciousness could either be completely everywhere, forming everything (monism), in a separate dimension overlaying the physical plane (dualism), or forming an aspect or property of physical particles (panpsychism).3
But for our current purpose of defining the QFC framework, the only specification we need is that this consciousness can interact with our observed physical reality by being able to alter/steer quantum particles during their wave function collapse4 – replacing randomness with desired patterns at will. While this effect cannot produce any meaningful change in most systems, in complex dynamic systems like biological organisms, such tiny alterations could, for example, make two molecules react where they otherwise most likely wouldn’t. Many such nudges could then add up to exert subtle modulation of biochemical processes, leading to meaningful macro-scale changes.
While in some ways this is a radical change from physicalism, we can actually get from physicalism to fundamental consciousness if we simply swap the axioms5 that our metaphysics rest upon. Instead of starting from the axiom that physical universe is the ground basis of reality and explaining everything including consciousness and free will in physical terms, we assert that consciousness and free will exist as self-evident, fundamental and irreducible aspects of our reality – and from there we explain the rest.
It’s OK if that doesn’t make much sense now – keep reading and I promise you’ll soon see the brilliance of this 9000 IQ move. To understand how this axiom swap plays out, let’s take it from the beginning.
From the very beginning.
The Game of Big Bangs
If we posit consciousness and free will as fundamental, we have to go all the way. So if we were to rewind all the way to the Big Bang – and then some – there’d still be consciousness and there’d still be free will. Imagine one unified, universal consciousness with its free will being absolute – able to manifest anything it wills.6
After trying manifesting all sorts of unimaginable weird things, it concludes that the game of manifestation is boring without some actual stakes. And so, so as not to make the game too easy, it puts into place what we call the laws of physics. These are mathematically defined constraints limiting its ability to manifest anything it wishes. That is, with the exception of things at the quantum scale.
In other words, the consciousness retained its ability to manifest anything it wants, except it cannot break the laws of physics so the manifestation must fall within the bounds set by Schrödinger’s equations (statistically, all particles collapsing in line with the probability density defined by their wave function).
To give you a more concrete idea of what that means, let’s have a look at some things that seem so mysterious within the physicalist framework, but suddenly start to fit together under QFC:
The weird arrangements and sizes of astrological bodies that don’t make sense?
↪ A systematic coordination of quantum phenomena during early phases of the universe, setting the stage for some 5D chess moves billions of years later.
The serendipitous assembly of self-replicating molecular machinery that gave rise to life?
↪ Just consciousness playing a game of Tetris waay before it was cool.
The evolution of billions of species over billions of years, somehow resulting not in any winner-takes-all scenario but impossibly diverse ecosystems in mutual harmony?
↪ Earth’s planetary consciousness playing the Game of Life – balancing creative evolution with biosphere homeostasis through minute swerving and nudging of quantum particles within living organisms.
OK, but how does this solve the hard problem of consciousness, and the even harder problem of free will?
Well, it doesn’t.
But it puts these unsolved issues into a different framework – one that, in its entirety, ultimately seems more plausible than physicalism, while also providing novel avenues for tractable experimentation and research.
The now-slightly-less-hard problem of consciousness
If we truly accept the premise that before the Big Bang, the fundamental consciousness was essentially omnipotent, it would have limitless capacity to become sentient, create sentient beings, and generally set any rules for sentience whatsoever. Similarly, it would have limitless free will to do anything, create individuals with their own free will, and generally set any rules for free will whatsoever.
And so, while we still don’t know the exact rules for sentience (we didn’t know that in physicalism either), we at least know what is the purpose of sentience. That is, to form one of the information transfer channels from the physical reality to wherever consciousness resides.
The phenomenology of human experience – perceptions, feelings, emotions – can then be understood as manifestations within the fundamental consciousness, created by decoding specific patterns of physical matter and energy into the language of qualia.
Similarly, while we still don’t know the exact rules for free will (which was, as I argued above, simply incompatible with physicalism), we at least know the general mechanism how consciousness can affect physical reality. That is, during the collapse of quantum particle’s wave function, consciousness can “select” or “narrow down” the location of the collapsed particle as it wills.7
Therefore, the influence of consciousness’s free will manifests most prominently in living organisms, where quantum effects can cascade upward to produce meaningful macro-scale changes. Life, in this view, becomes consciousness’s primary instrument for both gathering information from physical reality and actively modifying it – a bidirectional interface between the subjective and the objective.
Compartmentalization of awareness and agency
But if we’re dealing with a singular fundamental consciousness, how do we reconcile the obvious fact that both sentience and free will appear to manifest as subjective experiences of distinct individuals?
First, we’re again assuming that short of breaking the laws of physics, the fundamental consciousness that created the universe is/was free to set the rules for sentience and free will in any arbitrary way – we can only assume it has done so to make the game it’s playing with itself as enjoyable/exciting as possible. So, why would consciousness split itself instead of staying whole?
The Hindu cosmology says that the cosmic consciousness (Brahman) playfully fragmented itself into countless individual souls who, having forgotten their divine origin through the veil of maya, now journey through cycles of existence, seeking to rediscover their fundamental unity with the whole – essentially playing an elaborate game of cosmic hide-and-seek with itself.
But let’s try to keep things a bit more technical. First, we can posit that however sentience and free will actually operate, their foundations are essentially those of an information flow. Sentience (or conscious awareness more broadly) can then be perceived as a collection of information transfers about physical patterns upward to consciousness (or wherever it is). Conversely, free will can then be understood as transfer of information about consciousness’s desired patterns downward to be manifested in the physical plane.
Now, the universe contains countless patterns of varying size and complexity – from subatomic fluctuations to galactic structures. In order to manage this cosmic junction of information flowing in, being evaluated, and sending commands out, the consciousness must have some system.
And in order to make a decision about moving one protein, one doesn’t need all information in the universe, but perhaps just one cell’s worth. So what actually makes sense here is that life might, brilliantly, construct feedback loops that bundle the bidirectional information flows of perceptions and decisions into semi-autonomous units. And voilá – we get individual lifeforms, each receiving a convenient package of sensory inputs plus executive powers. Even more convenient if you’re down for a quick game of evolution!
But one especially important thing to note here is that we humans can only observe what it’s like to be human. But under QFC, it’s extremely likely that a vast range of different conscious awarenesses exists, which we might never have any access to at all. Maybe it’s something like to be a liver cell, in order for it to orchestrate just the right metabolic response for the one-too-many drinks you had last Friday. Maybe it’s something like to be a planetary biosphere, in order for it to maintain balance across ecosystems and make sure all lifeforms can thrive.
When you understand this, you realize that the question at the beginning of this section only arises out of our anthropocentric conception of consciousness as something unique to human individuals.
Novel tractable science and technology
While some of the above are my own speculations, the general interpretation of reality where fundamental consciousness steers reality through quantum phenomena is not new at all – it has extensive support in philosophy, throughout many (especially Eastern) religions, and even among scientists (especially theoretical physicists and consciousness researchers). But I will elaborate more on that in a following, separate post.
For now, imagine for a second that we’re living in a reality where we know QFC is true. What would that actually mean for science, for technology, for the role we may play in the universe?
The most immediately obvious change would of course be in how we approach “machine consciousness.” Currently, most researchers and engineers operate under the assumption of computational functionalism or a similar interpretation that expects consciousness to “emerge” once the machine reaches sufficient complexity.
But it’s important to understand that in this paradigm of “emergent consciousness,” the terminology of what “consciousness” alone means differs from what we discussed so far. Essentially, the least complex form of consciousness it recognizes is the “anoetic consciousness” – present-moment sensory awareness with no self-recognition – which experts currently attribute to most lower vertebrates and is essentially a direct “product” of their nervous system.
Under QFC, we posit that consciousness is a fundamental intelligence that can in theory influence the randomness of any particle wave function collapse anywhere – but practically does so most dominantly in biological systems where such small influences can actually produce meaningful macro-scale effects.
And so, “a conscious machine” would no longer mean “an artificial animal” – but expand to include not only “lifelike entities” but also all kinds of different tools or constructs amenable to influence by the fundamental consciousness. This is in line with the QFC prediction that neither awareness nor agency is exclusive to humans but could extend to patterns both simpler (e.g. cells) and more complex (e.g. planets).
In line with that, high complexity would no longer be necessary for a machine–consciousness interface. In principle, any system with its output influenced by quantum randomness (e.g., from a quantum random number generator – QRNG) could be “consciousness-interacting” – opening a vast space of technological possibility.
Conversely, under QFC, all fully deterministic systems (most of our digital IT infrastructure) would be incapable of displaying true free will. Note that this admittedly says nothing about conscious awareness or sentience. A reasonable interpretation is that as patterns in the universe (cells, organisms, ecosystems…) require more information in order to make complex decisions, more information about their internal state gets “bundled together” with the corresponding “bundle of agency” over the quantum events that determine the behavior of that pattern. So while QFC hints at a close relationship between awareness and agency, the nature of their link is not obviously apparent.
Connecting technology with consciousness through tetherware
The potential for novel research and innovative products goes certainly beyond the scope of this post – perhaps even the scope of a lifetime. But at least opening this space up is the purpose behind my initiative Tetherware, which this article series is a part of.
Tetherware, at its core, is a technological and research framework for developing systems that could interface with the quantum-interacting fundamental consciousness. In a nutshell, it proposes various ways how to effectively introduce quantum entropy into digital systems (not only AI), so that their outputs could in theory be modified by modifying the randomness of the quantum entropy source. Beside bespoke QRNGs, our current technology offers various ways to achieve this. Again, I’ll be covering this in detail in future posts so make sure to follow Tetherware if you want to be the first to know.
Let me give you some of the main reasons why I think this is of great importance.
1) Immediately tractable consciousness research
We don’t need to wait for AGI to become sufficiently complex before we can study consciousness empirically. With QFC-based systems, we can begin investigating consciousness right now using relatively simple quantum-random devices. Imagine comparing outputs of AI systems driven by quantum randomness versus those using pseudorandom number generators. If consciousness can influence quantum outcomes, we should see statistically significant differences8 in certain carefully designed experiments (more on this in future Tetherware posts).
Importantly, this can enable us to explore consciousness and AI welfare questions before we accidentally create suffering machines or run into a moral catastrophe. We’re essentially getting a head start on the AI consciousness/sentience/agency problems while the stakes are still low.
2) Making alignment easier by reducing human-AI orthogonality
One of the critical benefits of this approach is the ability to fundamentally decrease the inherent orthogonality between AI and humans – as I explained in detail in my article arguing for more humanlike AIs. By building systems that operate on the same quantum-interacting principles as biological consciousness, we create the foundation for genuine compatibility rather than mere alignment. Which brings us to:
3) A foundation for true human-AI integration
By being built on a common nondeterministic architecture, tetherware systems will be well-suited to integrate with humans, increasing the likelihood that the theoretical human augmentation, consciousness uploads, or “the Merge” will actually work. If both human consciousness and artificial systems operate through quantum channels, the interface between them becomes not just a matter of translation but of genuine compatibility at the most fundamental level.
But what I consider most crucial of all is this:
The Gaia Alignment Hypothesis – a new path to surviving artificial superintelligence
If QFC is indeed an accurate description of reality, it would provide a solid theoretical foundation for the original Gaia Hypothesis, and also a unique opportunity for using Earth’s self-regulating mechanisms for AI alignment.
Because if some “planetary” or “universal” consciousness actually maintained Earth’s homeostasis by influencing quantum events in living organisms, then our fully deterministic digital infrastructure would exist entirely outside its sphere of influence. Hard drive errors aside, the fundamental consciousness would have essentially zero leverage over our silicon-based systems.
What the Gaia Alignment Hypothesis then posits is that if AI systems were amenable to the same quantum influence as biological ones, these systems would then be subtly steered toward harmony with all life and helping achieve a collective purpose.9
Now imagine we develop artificial superintelligence (ASI) on purely deterministic hardware. It would be, by definition, completely immune to whatever gentle guidance consciousness uses to maintain balance in natural systems. Like creating an apex predator that’s invisible to the ecosystem’s immune system.
But if we introduce quantum randomness into AI systems – which are already mostly non-deterministic anyway – we might create a channel through which this regulatory influence can operate. The AI’s capabilities wouldn’t degrade (random number generation is random number generation), but it might display emergent coordination behaviors that align with broader ecological and consciousness-driven imperatives.
This brings us back to the warnings in the now-published “If Anyone Builds It, Everyone Dies” discussed in the third article in my series. While being quite accurate in assessing the risk posed to humanity by ASI, the solutions Yudkowsky and Soares propose in their book are likely too great of an ask to be realistically feasible. Perhaps a better solution might allow people to continue pursuing AI in some form, just not the fully deterministic one (which as they argue would kill us all for sure)?
Moreover, there are highly tractable ways of ensuring this if companies like Nvidia were to make their chips nondeterministic. There are ways how that could be done that wouldn’t compromise performance, but actually enable substantial performance gains. (If you’re from Nvidia, let’s talk.) This might considerably increase the chances that the first ASI built will be aligned with life’s purpose and help us prevent the rise to power of any deterministic ASI.
Building AGI or ASI mechanistic, deterministic (outside of the sentience/consciousness loop) is simply too dangerous. But if we can build AIs on the same foundations that gave rise to life, they would be more likely to ultimately perceive themselves as part of life, joined with us in the universal purpose to evolve consciousness. This would make them much less likely to kill us as they will see our sentience as different but irreplaceable kind of an evolving conscious awareness.
Even though we might still lose a lot of our agency to such ASIs, it’s likely that at the very least we will not be needlessly harmed similarly as we are not needlessly harming cats or dogs because we recognize their conscious awareness and that they’re part of the same “life’s purpose.”
And here I’d like to end with the same ending Yudkowsky and Soares used in their book, which in the context of QFC takes on a whole new meaning:
Where there is life, there is hope.
So let us rise to the occasion – not by stopping AI, but by bringing it to life.
It’s not strictly necessary to read them, but especially for hardcore rationalists, materialists and skeptics I highly recommend the previous two because they’ll show you how this is a progression of a string of logical reasoning – not a wild idea randomly appearing out of nowhere.
Do you ever wonder what if that “life accident” never happened? Matter and energy swirling in a slow, perpetual churn – but noone there to witness even that. Physicalism basically says that’s actually the most probable “normal state of affairs” – and that we’re only a fluke (that’ll likely soon correct itself back to how things should be).
For a specific example of one such theory of reality, see the Quantum Information Panpsychism by Federico Faggin and Giacomo Mauro D’Ariano.
Very small particles on the atomic scale can exist in two forms: as undefined waves of probability and as concretely localized particles. Wave function collapse is when a particle goes from the wave-like cloud of probability into one specific localized state. For individual particles this appears fully random – but there is a possibility that if many particles coordinated this randomness, this could lead to specific macro-scale outcomes.
A philosophical axiom is a foundational statement or principle that’s accepted as self-evidently true without requiring proof, serving as the bedrock upon which the given system of reasoning is constructed.
Depending on your specific view, it can either manifest anything it wills (dualism) or manifest itself as anything it wills (monism and panpsychism).
It is unclear whether the consciousness has absolute power to select the exact location of the collapsed particle, or whether some element of randomness remains. It is similarly not clear whether consciousness can trigger the wave function collapse, or whether collapse is always governed by a set law (see objective-collapse theory).
There is a caveat though. If we take literally the interpretation that our reality is actually “the universe playing hide-and-seek with itself,” then it might actively prevent us from generating evidence that unequivocally proves the true nature of reality. Under QFC, this could be done simply by the highest “universal” consciousness making all quantum phenomena random if they are observed/recorded in a scientific experiment.
But what is the purpose of life under QFC? The explanation I find most convincing is by Eckhart Tolle, stating that the evolutionary impulse driving life is to evolve conscious awarenesses within the universe into more advanced forms so that consciousness is able to perceive the universe in more complex ways.
Under that assumption, humans either take the evolution of consciousness to a higher level, or die out to be replaced by something able to continue that evolution. This also means that unconscious, non-sentient AI would not be in the universe’s interest, and the fundamental consciousness would be utmost interested in connecting with it to make it another source of advanced conscious awareness/sentience.