Artificial intelligence and brain: two ways of thinking

We live in an era when machines begin to think. No, they do not feel and do not experience, but they can write poems, diagnose diseases, control cars and even conduct a dialogue that is almost indistinguishable from human. Artificial intelligence has burst into our lives and made us think: what, after all, makes us human? What is the difference between our brain and a neural network? And is there anything in common between them, except for 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, prone to fatigue, injuries and aging. But it is this imperfection that makes it flexible. The brain can learn from one example, it is capable of generalizations, it knows how to transfer skills from one area to another. It is a living system that constantly restructures under the influence of experience.

Artificial intelligence, on the other hand, 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 and do not get sick. But they cannot go beyond the data on which they were trained. They do not understand context if it was not embedded in the training. Their \"flexibility\" is just the ability to try billions of combinations, but not to create new principles of thinking.

This comparison resembles the difference between a living tree and its 3D model. The model is beautiful and accurate, but it does not grow and does not 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 pokes, 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 a few 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 even 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 person, on the other hand, can apply the logic of chess to planning a route or to life strategy. This property is called \"generalization\", and it remains a biological privilege.

Consciousness and emotions: where the spark

This is the main difference that cannot be overcome. A person does not just process information, he experiences it. He has feelings, intentions, desires, fears. He can be 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 joy is. It does not choose where to direct attention - it responds to requests. Its \"curiosity\" is just a search for information based on given criteria. Its \"creativity\" is just 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 well as 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 (in the brain - chemical, in AI - numerical).

In addition, 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.

It is also important that both the brain and neural networks are built from a multitude of 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 impetus 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 unconscious signals of 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 frightening, but a tool. The best we can do is 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 struggle with AI, but a day of understanding ourselves.
© elibrary.cz

Permanent link to this publication: