What is AI really?

Hello hub. I'm a person who often hears that AI will destroy the world or something like that. However, when I hear the word AI, I sometimes think about the intelligence of the Minecraft pig or the weather forecast, and not about neural networks. Why? Now you will find out.

Today I will teach you how to classify AI and make you more literate in this area.

There are many different types of artificial intelligence, but I will only talk about what currently exists in our world, and not about the hypothetical, not created strong AI, but about weak AI.

Here are the types of existing AI for working with data:

And types of AI based on the principle of operation:

• If-else programs (write a more precise name in the comments if you know)

• Neural networks

• Other methods

Yes, there are only 3 types of data processing and 2 basic principles of operation, but they themselves contain many different examples, from the mentioned AI pig from Minecraft to ChatGPT

I won’t go into whether the AI ​​uses neural networks, decision trees or if-else methods, since any example can use several options at once.

Data processing systems are programs that, based on the data received, produce a single result, which can be a specific answer or probability. The if – else method is often used. Here are examples of such programs:

1. Gaming AI

2. Expert systems

3. Chatbots

4. Computer vision systems

5. Many others

Game AI: Let's imagine a pig from Minecraft and a player. The goal of the pig is very simple – to follow the player if he has a carrot in his hands. Otherwise, don't go. In addition to the pig, you can also bring chess AI here.

Expert systems: Using an X-ray, the program automatically checked the integrity of the skeleton and highlighted the location of the fracture, indicating the type and approximate recovery time.

Chatbots: User Entered: Hello? Chatbot: If you entered hello, you need to answer “Hello, how are you”, otherwise answer “I didn’t understand the request” (Siri, Outdated version of Alice). Such programs are usually equipped with programs for determining meanings, since hippopotamus and hippopotamus are the same thing, but the words castle and castle are different words and you need to be able to distinguish between them all.

Computer vision systems: Describe the picture. This includes image generation.

This is only a small part of all types of information processing systems. And we are moving on to the next type of AI for working with data.

Data prediction systems are, as the name suggests, programs designed to predict data or forecast time series. Usually these are neural networks, but sometimes they are programs that use other approaches, like decision trees or adaptive systems. Here are examples of such programs:

1. Forecasting

2. NLP (Natural Language Processing)

3. Prediction of events

4. Many others

Forecasting: When predicting events, we usually talk about chaotic systems with many possible solutions. For example, based on data on cloudiness, time of year, humidity, temperature, give a weather forecast for tomorrow or any other day. Another example: predict the incidence of ARVI this year.

NLP: ChatGPT and SORA are great examples of all this.

Event Prediction: Prediction usually refers to a single event, such as the likelihood of this asteroid hitting Mars given its trajectory. Another case is the prediction of permanent events, such as the next passage of a comet near the earth.

All of these AIs and examples are fairly well-known examples, but there are many more in general. I also did not mention the mathematical models designed to work with the function, although they are always with us and form the basis of all neural networks, even ChatGPT.

I hope this was quite interesting or useful for you.