Mind or AI — who will rule?

In recent years, a growing number of leading AI experts have drawn striking parallels between artificial neural networks and the human brain, suggesting that machine intelligence may not only equal but eventually exceed human cognitive abilities. Geoffrey Hinton, often called the Godfather of Deep Learning, remarked, "Our brains are just big neural networks. There's no reason a neural network can't do anything a brain can do."

In the same vein, Yann LeCun, Meta's Chief AI Scientist, stated, "Intelligence — whether biological or artificial — is the ability to predict, plan, reason, and learn. There's no fundamental law that says machines can't surpass us at all of them." These assertions reflect a widespread belief in the AI community, that by mimicking the structure and learning mechanisms of the human brain, machines can become not only equal to but ultimately more efficient, reliable and capable than humans in the matter. From this perspective, the rise of superintelligent AI is seen not as a question of 'if', but of 'when'.

Fact is that machines have always been wonderful aiders for reducing human work, for enhancing human productivity and making human life easier. But it needs not to be forgotten that machines are engineering feats of humans; the human mind comes up with things it wants to do, collects resources and makes machines that would aid him to do the burdensome task in a better, precise and easier way. In this progression, the more science has been done, the more machines have been made, and the more resources collected. This continual process keeps widening the range of possible new ideas, of new works and of new machines to do that work.

So, will there be an eventuality that the human mind, which so far seems to be the most thinking, imaginative, creative and agency-bearing creature in the universe, will one day create a machine that is more capable than itself in thinking, imagination, creation and agency? To answer this basic question, one needs to dive deep into the actual 'work' the AI does and compare that to the 'work' of the human brain.

The human brain is a massively parallel, self-organising biological system composed of approximately 86 billion neurons, each forming thousands of synaptic connections. Working in parallel means that each of the 86 billion neurons can be active at the same time communicating with thousands of others through synapses. Therefore, the human brain processes millions of tasks simultaneously.

For instance, when a human sees a cat, the brain is recognising the shape 3-dimentionally, gauging its volume and weight, processing the motion in 3D environment, associating the image with memories, activating related emotions, etc — all at the same time. This parallelism allows the brain to process millions of tasks simultaneously and to be extremely fast and efficient in a complex environment. At the same time when the brain is processing the presence of the cat, it may also be talking to someone, doing chores, and also receiving synapses from the all the organs and systems of the body, wherein it is organising functions at multiple levels.

In contrast, artificial neural networks are simplified mathematical models with rigid architectures, typically organised in layered sequences. They learn by optimising numerical weights through backpropagation and gradient descent — abstract statistical processes far removed from the rich, embodied learning of human beings. While AI systems require vast amounts of labeled data to learn specific tasks, human children can grasp abstract concepts, infer meanings and generalise from just a few examples, often guided by curiosity, intuition and social interaction. These differences are not merely technical; they point to a deeper divide between AI and organic intelligence shaped by millions of years of evolution and engineered systems bound by human-coded rules and objectives.

To understand how artificial neural networks operate, consider the common task of identifying the image of a cat. When an image is fed into an AI neural network, the first step involves the input layer, which receives the raw pixel data - typically tens or hundreds of thousands of numerical values representing color and intensity. Each pixel becomes a numerical input to the network. These inputs are then passed through a series of hidden layers, where the network performs mathematical operations using weights, which determine the importance of each input; and biases, which shift the activation threshold.

Initially, these weights and biases are randomly assigned, but during training or machine-learning, the network uses a method called backpropagation combined with gradient descent to iteratively adjust them. After comparing its prediction with the actual label, "cat", the network calculates an error, and propagates that error backward through the layers to finetune the weights and improve future predictions. This process is repeated thousands or millions of times across vast datasets, gradually allowing the network to "learn" which patterns of pixels correspond to cats versus other objects. However, this learning is entirely statistical and lacks any understanding or awareness — the AI recognises a cat only as a mathematical pattern, not as a living being with meaning or context.

When more neural network, meaning more input-output sequences are added to AI machine learning, and the layers are allowed to process in different ways, it becomes AI deep-learning.

One can understand that no part of the AI system 'understands', 'recognises' or 'feels'. Rather with increasingly improving nano-chips, of humongous and pricy data-centres and huge amount of electricity to run them, computers have been developed that can manage trillions of nodes and parameters. Just to 'recognise' a cat, AI neural networks runs through these parameter-systems, taking millions of hit-and-trials, adjusting and readjusting weights to eventually get the answer 'cat'.

Ask yourself, did your mind go through millions of tries, touching trillions of parameters to recognise a cat, or just your mama told you, "baby that's a cat" and you instantly started loving the cat and playing with it; and then you could recognise a thousand types of cats apart from a thousand types of dogs or other four-legged animals of approximately the same size.