When we are talking about artificial intelligence, we need to define, what intelligence is. We can define it as “the ability to take effective steps to achieve a goal”. Goal can be for example playing chess. You learn the rules that thanks to your intelligence can be used effectively to defeat your opponent. Artificial intelligence called “DEEP BLUE” programmed precisely for this purpose defeated master of chess Garri Kasparov already in 1997. DEEP BLUE was a complicated algorithm that used the “Brute force tactic” when in real time calculated every possible move and picked the best one.
Another example is artificial intelligence from Google called “ALPHA GO”. This artificial intelligence is programmed to play a game called “Go” popular in China, North Korea, and Japan. Lee Sedol (Master of the Go) was defeated by this artificial intelligence in 2015. Go has easier rules, but the number of all possible moves is bigger than the number of all atoms in our universe. This means, that counting all possible moves is a waste of the time. ALPHA GO therefore uses neural networks and learning. This very basic neural network was supplied by real games played by professional players of the game Go. ALPHA GO could learn from this game’s effective moves and tactics. No one expected the results, and the surprise is evident from interviews with Lee Sedol before and after his game against ALPHA GO. Before the match, he said: “From what I’ve seen, I think I’ll win narrowly.” after the match was his answer different: “I felt practically helpless.”
These are only two examples of artificial intelligence and why it’s better and stronger than people. We can find a lot more examples, but one thing is the same. How artificial intelligence works. The most dominant artificial intelligence is called “Machine learning”. Machine learning is used everywhere. When recommending videos or posts on social networks, in banks or other institutions for detecting frauds, for setting optimal prices, for speech recognition, and more. Very popular ChatGPT also works on the principle of machine learning. How does this artificial intelligence work?
Machine learning is inspired by how our brains work. We don’t know much about how our brains work, but we know something. Our brains are made of neurons (about 100 billion neurons). These neurons or brain cells are connected by “synapses”. Every neuron can receive and send signals. When a neuron receives some signals, it sends its signal. One neuron can relate to thousands of other neurons, and this creates a very complicated network. The strength of individual connections changes over time. The main point is that the stronger the connection is, the easier the signal passes through it. When we learn something, in the brain are made new connections and existing ones are strengthened.
Here is an example. Imagine that you see a picture of a dog on your screen. Light from your screen enters your eye and your eye activates some neurons, which corresponds to the physical properties of the received light. The signal from the eye traveled to the part of the brain responsible for processing visual sensations, where several groups of neurons analyzed different aspects of the dog, such as its color, face, number of limbs, snout, and more. The brain then combines all this information and evaluates that what you see is a dog. During this whole process, the connections between the neurons involved in this process were strengthened.
The point of this process is that the brain learns from experience, and it is very good at pattern recognition, this is the concept behind artificial intelligence, respectively machine learning. Teaching artificial intelligence is about creating a network of artificial neurons. This artificial neural network is much simpler than the one in the brain, but the basic principle is the same.