If you’re interested in further discussion on neuro-integration from all aspects and angles, I’ve set up a discord server here. Come join the fun! WARNING: The following contains MANY spoilers for the show Pantheon (2022) from AMC. You probably haven’t seen it yet, but reading this article will make you want to see it, so I highly suggest you watch it first if you care about spoilers. If not, let us proceed. On the 30th of November 2022, the technological landscape tilted on its axis. OpenAI released ChatGPT, and suddenly Generative AI was the only thing anyone could talk about. We became obsessed with Large Language Models, transformers, and the eerie ability of a machine to predict the next token in a sentence. Just two months earlier, in September 2022, a show called Pantheon aired on AMC. While the rest of the world was about to lose its mind over a chat bot, Pantheon was quietly laying out a roadmap for something far more profound: Computer-Simulated Human Consciousness, or “Uploaded Intelligence” (UI). It contained some of the most realistic depictions of mind uploads, and the potential effects on society, that I have ever seen. It grapples with the hardest philosophical questions head-on, such as the true meaning of consciousness, personality cults, the consequences of immortality and digital death, self-copying, and the universe-as-simulation theory. I don’t think many people saw it when it came out. (I didn’t) Yet there are so many fantastic, mind-bending, life-changing things about AMC’s Pantheon not just in its philosophy, but also the way it depicts cyber-security, software engineering and big data concepts; not to mention the cut-throat world of big tech both within and beyond Silicon Valley. It’s not perfect, but it’s easily the most accurate I’ve ever seen from a show of this kind. Watching it today makes it feel somewhat prophetic, but likely was the result of extremely good footwork by the producers and their team, getting the most up-to-date picture of the inner world of big tech at the time, between 2021 and 2022. Facebook became Meta at the end of 2021 as a result of Zuckerberg’s Metaverse strategy, and you see a lot of those ideas were depicted in Pantheon. We also see clear depictions of things Generative AI are doing today: chat interfaces with digital intelligence, for instance. The digital intelligences in Pantheon weren’t artificial neural networks, but fully scanned and emulated human brains. They call these “Uploaded Intelligence”, or UI (which is confusing in tech, since UI stands for User Interface, but we’ll roll with it.) In the show, a UI is created by laser-scanning a biological brain, layer by layer, down to the stem. The brain is destroyed in the process - vaporised - but stored apparently in its entirety in digital form. This is a human mind, stripped of its biological substratum which is replaced by silicon. The entire connectome is then reconstructed digitally, presumably with simulated sensory receptors; the rest of the central nervous system is not included in the actual scan. Somewhere in between all that, there is some magic fairy dust that allows the full emulation to happen, but since that is still one of the great millennium-type problems of our time, I don’t expect them to have that bit of detail. At one point in the show, a UI produces an 80-page patent in seconds. Although GenAI today would likely screw that up with hallucinations, you can see the resemblance. It feels eerily prescient. However, we’re not here to talk about GenAI, and neither does the show: the real focus is on the Uploaded Intelligence and the ability to fully simulate a human being computationally. This requires a few major assumptions: * The brain - that is, the cortex and brain stem - constitutes the entirety of who we are as individuated conscious beings * An individual’s behaviours, emotions, and cognition, can be simulated in their entirety from a scan of the cortical network using classical and quantum computing platforms Before we can even begin to tackle these assumptions, we need to understand what it is to compute, to be intelligent, and to be conscious. I Am, Therefore I Compute When we perform a mental task, sometimes we are conscious of the effort expended to perform it. Other times, we are unconscious of that effort, and an input gets processed and turned into an output which is re-integrated with our conscious awareness sometime later. To us, these results arrive as flashes of inspiration or insight; that is the moment of reintegration. In truth, the brain was likely working on that problem for a period of time without you consciously being aware of it. These unconscious computations are presently done in our biological circuitry. Technically, the physical medium of computation doesn’t matter, and could be electronic. However, there is an inherent problem in viewing the brain as equivalent to electronic circuits. Classical computing and electronics are based on gates. Gates allow you to perform simple, but exact, operations on incoming electrical signals. For example, an OR gate takes 2 inputs, and so long as at least one of those inputs is receiving a signal, the OR gate outputs a signal. A simple way to model this would be like: 0 OR 0 = 0 1 OR 0 = 1 0 OR 1 = 1 1 OR 1 = 1 The number 1 denotes an electrical signal, while 0 is no signal. Meanwhile, an AND gate takes 2 inputs, and only outputs a signal if both inputs have a signal: 0 AND 0 = 0 1 AND 0 = 0 0 AND 1 = 0 1 AND 1 = 1 Gates like these feel intuitive to us. They follow a simple logic. They’re also exact and about as deterministic as it gets. They have no hidden influences outside the two expected inputs which can affect the result; so 0 AND 1 should never result in 1, just as 2 + 2 should never equal 5. Brains, and biology in general, are nothing like that. The thing about biological computation is that it is fuzzy. For most of us, doing novel arithmetic in our head is a combination of heuristics learned from repetition in similar tasks, which we use to get a sense of approximating a value; then more heuristics refine it down until we have a value in mind that we feel confident enough about. Let’s do a quick experiment. Solve the following 2 problems: * What is half of 10,000,000? * What is half of 8,626,400? Which one required more time/mental energy, the bigger number or the smaller one? If we followed a purely computational way of thinking about things, our expectation should be that the bigger number would be more computationally expensive to calculate than the smaller one. However for the human brain, it’s not dependent on the size of the numbers we’re working with, but on their composition. In the first problem, although the number was larger, its composition was vastly simpler, made almost entirely of zeroes. To solve it, we could reduce it by 6 decimal places, and then the problem becomes “What is half of 10?” The second smaller number had many more non-zero digits, meaning we could not reduce it in the same way. Instead, our natural inclination is to solve for each non-zero digit separately: “What is half of 8? What is half of 6? What is half of 2?” and so on. 1 problem instantly turns into 5. “Therefore, our brains are not computers, therefore, our brains cannot be simulated by computers.” Woah hold up there cowboy, not so fast. Plenty of things which are not computers are simulated on computers literally all the time in every field ever; we just haven’t simulated literally everything in the universe that exists. This argument, that the brain is not a computer and therefore could never be modelled by one, always drives me a little insane: just because it doesn’t follow gate-based logic does not mean it is not performing computation, and does not mean that it cannot be simulated computationally. It is, and it can. The problem here is one of dimensionality. To simulate the human brain, you need way more than merely its connectome. We know this from simulations of C. Elegans, the Nematode Worm whose species all have exactly the same number of neurons: 302, no more, no less. We have been working to simulate its entire set of known behaviours for decades computationally, and we have made progress; but consider how incredibly simple C. Elegans is, and yet we still haven’t figured it out? How complicated can a near-microscopic worm possibly be? Some have reached the conclusion that the complexity is not in simulating the worm, but rather, simulating chemistry itself. Chemistry is, in my opinion, the most sophisticated and most powerful phenomenon in the physical universe. Chemistry is like the operating system, the essential firmware, upon which the software of human minds can run. Even firmware requires something firmer: hardware. That’s quantum mechanics. That’s effectively the architecture upon which the Firmware must execute. Indeed, if the answer is that we need to simulate chemistry itself, then think of our predicament this way: Imagine you were trying to simulate a classical gate-based Turing machine on some highly exotic computer, and you only managed to implement AND and OR gates; then you proceed to try running DOOM on it (as is tradition.) You wouldn’t get very far, would you? Technically, we can build a complete computing machine using any combination of “NOT” operator with “AND” or “OR” gates, but we don’t even know the NOT operator exists, let alone what a universal Turing machine even is. That’s more or less where we are in the grand scheme of things when it comes to simulating all of chemistry. I Am, Therefore I Intellect Another very common assumption made by just about everyone is that the ability to think implies intelligence. It’s also extremely common to conflate intelligence with emotion, with motivation, with the survival instinct. This is part of the problem which has often taken debates on simulated