Artificial intelligence and the brain: two ways of thinking

We live in an era when machines are beginning to think. No, they do not feel or experience, but they can write poems, diagnose diseases, drive cars, and even conduct a dialogue that is almost indistinguishable from human. Artificial intelligence has burst into our lives and made us ponder: what, after all, makes us human? What is the difference between our brain and a neural network? And is there anything common between them, besides the word ‘neuro’? World Brain Day is the perfect occasion to delve into this depth and try to understand where biology ends and code begins.

Architecture: biological chaos versus mathematical order

The first and main difference is how both ‘processors’ are structured. The human brain is the result of millions of years of evolution. It is not designed, but grows like a living organism. Its neural networks are not perfect: they are noisy, slow, subject to fatigue, injury, and aging. But it is this imperfection that makes it flexible. The brain can learn from one example, it is capable of generalizations, it can transfer skills from one area to another. It is a living system that constantly reconfigures under the influence of experience.

On the other hand, artificial intelligence is created by engineers. Its neural networks are mathematical models operating on digital carriers. They are accurate, fast, and predictable. They do not get tired or sick. But they cannot go beyond the data on which they have been trained. They do not understand context unless it has been encoded in the training. Their ‘flexibility’ is just the ability to try billions of combinations, but not to create new principles of thinking.

The comparison here is reminiscent of the difference between a living tree and its 3D model. The model is beautiful and accurate, but it does not grow and bear fruit. The tree is chaotic, unpredictable, but it is alive.

Learning: experience versus data

A human learns through interaction with the world. A baby does not receive labeled data — he stumbles, tries, falls, cries, and based on this chaos, builds models of the world. His learning is continuous, without a teacher, in conditions of uncertainty. The brain learns all his life, and every new experience changes its structure. It does not require billions of examples to recognize a cat — it is enough to see it several times in different angles.

Artificial intelligence learns on huge amounts of data. To make a neural network learn to distinguish a cat from a dog, it needs thousands, sometimes millions, of labeled images. It does not ‘understand’ what a cat is, it simply finds statistical regularities in pixels. Its learning is the optimization of the error function, not the formation of an internal model of the world. It does not know that a cat meows and catches mice — it knows only that there is a certain correlation between the shape of the ears and the label ‘cat’.

In addition, AI does not transfer knowledge from one area to another as naturally as a human. A neural network trained to play chess cannot play Go without retraining. A human, however, can apply the logic of chess to route planning or life strategy. This property is called ‘generalization’, and it remains a biological privilege for the time being.

Consciousness and emotions: where the spark is

Here is the main difference that cannot be overcome. A human does not just process information, he experiences it. He has feelings, intentions, desires, fears. He can feel bored, happy, sad. He is able to be aware of himself, to ask questions about the meaning of life, to worry about the future. This is called phenomenal consciousness, or qualia. We do not know how it arises from neural activity, but we know that AI does not have it.

Artificial intelligence is an algorithm. It can imitate emotions, respond in a rhetoric that seems empathetic, but inside it there are no experiences or subjective experiences. It does not know what pain, sorrow, or ecstasy is. It does not choose where to direct its attention — it responds to a request. Its ‘curiosity’ is just the search for information according to given criteria. Its ‘creativity’ is the combinatorics of known elements.

Consciousness makes us vulnerable, but it also makes us human. It is precisely it that allows us to love, doubt, dream. And as long as we do not know how to recreate it in silicon, we remain the only creatures capable of asking questions about the meaning of our existence.

Similarities: common principles of operation

Despite all the differences, the brain and AI have important similarities. Both are information processing systems. Both use parallel data processing: neurons in the brain work simultaneously, as do layers of neural networks. Both learn through reinforcement and error correction. The principle of backpropagation of error in AI was inspired by ideas about how the brain regulates its connections. And in both cases, information is transmitted through excitation and inhibition (chemical in the brain, numerical in AI).

Moreover, both the brain and neural networks are efficient in image recognition. They can find patterns in noise, classify objects, predict sequences. Both systems can ‘remember’ information, although the mechanisms of memory are fundamentally different (synaptic plasticity versus weight coefficients). Both systems can make mistakes and both need ‘rest’ — the brain in sleep, AI in breaks for retraining.

Also, it is important that both the brain and neural networks are built from many simple elements working together. In this sense, they are examples of ‘emergent’ intelligence, where complex behavior arises from the interaction of simple parts. This similarity has given a boost to the development of the entire neuroscience, because AI has become not only a tool but also a model for understanding the brain.

Boundary: what will remain for humans

Today, AI surpasses us in solving narrow tasks: it calculates faster, plays chess better, translates texts more accurately. But it cannot make decisions in conditions of uncertainty without data. It cannot adapt to a completely new situation without retraining. It does not have intuition, which is based on many years of experience and subconscious signals from the body.

The boundary between man and machine does not lie in the level of intelligence, but in the way of being. We live, we suffer, we create meanings. Artificial intelligence is a tool. Powerful, useful, sometimes terrifying, but a tool. The best we can do is to use it to expand our capabilities, but not forget that true wisdom, creativity, and freedom remain with us. World Brain Day is not a day of fighting against AI, but a day of understanding ourselves.
© library.ee

Permanent link to this publication: