What is a neural network, in simple words
In short: a neural network is a program that learned to answer questions and write text by 'reading' a huge amount of text from the internet. It does not think like a human, but it is good at guessing which word should come next.
How it works, plainly
Imagine a person who read millions of books and remembers how phrases are usually built. Ask them anything and they assemble an answer from what they have seen. A neural network does roughly the same: it does not store 'correct answers' but builds the text piece by piece, predicting word after word. That is why it is flexible — and why it sometimes errs confidently.
What AI can do
Answer questions, explain the complex simply, write emails and text, translate, review documents, brainstorm ideas, help with code and studying — basically anything to do with text and language.
What AI CANNOT do
It does not know what is happening right now (unless search is connected), cannot reliably do complex math, and — most importantly — may invent a fact, date or quote while sounding completely sure. This is called a 'hallucination'. So important answers are worth double-checking.
AI, neural network, ChatGPT — the same thing?
Almost. 'Artificial intelligence' is the general term. 'Neural network' is how it works inside. ChatGPT, Claude, Gemini are specific AIs from different companies, each with its strengths — so it helps to ask several, not one.
FAQ
Is a neural network a robot?
Does AI tell the truth?
Where should a beginner start?
Try it yourself — it is free
Ask any question and get an answer from several AIs at once, marked where they agree.
Start for free →3 questions free, no card