Can a computer become a mind?

Since ChatGPT’s launch in November 2022, the public’s interest in artificial intelligence has skyrocketed. Among many of the philosophical, ethical, and morals questions is whether we will achieve AGI (artificial general intelligence). In other words, whether a computer can become a mind. Today, I have a concrete, impossible-to-argue, 100% correct answer to this question:

Yes.

(but not for the reason you may assume.)

Let’s start with a first-principles approach: to create a mind, we must know what a mind is. More specificially, if we want to simulate a brain, we must understand how it works.

Brains process and manipulate data through action potentials, or electrical signals that temporally fire in response to stimuli. Stimuli could be nerve input, internal signals, or simply regular intervals. As we learn and grow, the dendrites of neurons branch out to connect with each other. Compared to our state-of-the-art implementation of artificial neural networks, we are pretty far off. We don’t simulate temporal firing, and once a model is trained, it is static. It has no way to manipulate weights, or change architecture at runtime (although there is a research field of this called neural architecture search). A popular consequence of this is that large language models cannot transition information from short-term to long-term memory. The temporal nature of neuron spikes, or an equivalent behavior, is needed (funily enough, there is research on this, too, called spiking neural networks that can be trained without backpropagation). Worst of all, we are not even close to the number of parameters of a human brain. The best language models have about half a trillion parameters, and while we don’t know the exact number of neurons in the human brain, it is estimated to be between 10 and 1,000 trillion. Unless we can simulate more accurately, the level of intelligence we create will be limited.

However, we are getting closer with each year. On one hand, machine learning architecture is being optimized to better-reflect biological reality each passing year. On the flip side, our ability to manipulate biology is just now allowing us to embed real biological computers for general use. This makes a strong case for the possibility of creating a mind, but I am not interested in using any of this as supporting evidence. Instead, I have a simpler prospect: To answer the question with another question, consider the following: Does it even matter?

“Of course it does!”.

Why?

“Because if it’s true, it will change the world.”

Why?

Maybe this is a better question. Why will it change the world? Why do we want it to? Maybe, the hard part is not finding the answer, but finding the question. An answer of “yes” will not only change the world, but spawn many more questions. An answer of “no” means… everything goes back to normal? For this reason, the answer to “Can a computer become a mind?” is “yes”, because living in a world of this reality is fundamentally more interesting and fullfilling than the contrary. I have no other evidence to offer other than the interesting prospect, but neither does anyone else.

Based on what I’ve provided, it may seem like recreating minds is a matter of simulating reality, but our exact understanding of minds still wouldn’t be complete, because we must solve the idea of conciousness. What does that even mean? The answer is “we don’t know”. As far as we can tell, the brain is an arrangement of neurons with some areas of specialization, but there is no host of a soul, no central core of existence, and no light of conciousness. Until Neuralink, BlackRock Neurotech, or another lab studying the fundamentals of neuroscience discovers the answer to this question, we can’t know the case for machines. Moreover, we can’t know the case for each other. We can’t even know the case for ourselves.

Circumstances strongly support the idea that computers can become minds, as one day, someone will figure out how to simulate neuron activity one-to-one. If this proves to be impossible, it will prove conciousness exists outside the realm of what is simulatable, changing our understanding of quantum mechanics. Other than that, it presents a dull outlook for innovation and the sustainability of life.

Funding for AI companies are at a record high. Billions have been invested with billions- maybe even trillions more to come. Firms are not investing to answer whether AI is possible, they are doing it to answer what will happen if it is. As a civilization, we can either ponder the implications of this reality, whether it presents a threat or an opportunity, or we can put down the book and quil, get our hands dirty, and build the future.

Sure, there are useless AI products, slop, and money wasting. Yet the value and productivity gain from AI already is hard to ignore. From massive efforts of automating entire workflows, to obscure solutions like handwriting recognition for lost post office letters, the future of AI is already here.

For one minute, don’t argue, and don’t over-think this: I’m here to tell you that computers can become minds. Now, with that knowledge, what do you want to with it? What do you want to build? What should you test? What should you explore? These are the questions of machines and minds. Build the future you want, and build the questions that give you the answers.

"Only when you know the question will you know what the answer means."

- A Hitchhiker's Guide to the Galaxy

Sources

Comparing artificial and biological neural networks

• Section "Anatomy of a biological neural network"

AI startups grabbed a third of global VC dollars in 2024

• "Funding for AI and ML startups accounted for 35.7% of all VC global deal value last year"

Chalmers, J. (1996). The Conscious Mind: In Search of a Fundamental Theory. Oxford University Press.

• "Consciousness is a feature of the world over and above the physical features of the world. This is not to say it is a separate "sub- stance"; the issue of what it would take to constitute a dualism of substances seems quite unclear to me. All we know is that there are properties of in- dividuals in this world—the phenomenal properties—that are ontologically independent of physical properties."

• “It is logically coherent to suppose that there could be a molecule-for-molecule duplicate of a conscious being that lacks consciousness entirely.”