Sign up

What Is Artificial Intelligence?

Welcome to your first lesson! Here you'll find out what is really behind the term artificial intelligence (AI) and why it is less mysterious than you might think.

What does "artificial intelligence" mean?

Artificial intelligence describes computer programs that can carry out tasks that normally require human intelligence: understanding texts, recognizing images, making decisions, or translating language.

Good to know: AI is not a single program but an umbrella term for many different technologies. ChatGPT, Siri, the face recognition on your phone: these are all forms of AI.

Strong AI vs. weak AI

In movies, we often see AI systems that think and feel like humans. This is called strong AI (or AGI: Artificial General Intelligence). A complete AGI does not exist yet, but the boundaries are blurring: current models are becoming increasingly versatile and can handle more and more different tasks at a high level.

What we have today is weak AI (also called "narrow AI"): systems that solve the tasks they were built for very well, but without human understanding and without acquiring new abilities on their own. A chess program can beat world champions, but it can't do anything else. ChatGPT, which masters many tasks, is also considered weak AI: it does not understand what it is doing the way a human does.

Strong AI (not yet achieved)

Thinks, feels, and learns like a human. Can solve any task. Not yet achieved, but research is making rapid progress. The line between it and weak AI is becoming increasingly blurred.

Today's AI (more and more versatile)

Specialized, but increasingly versatile. Current models such as ChatGPT, Gemini, or Claude process text, images, and speech at the same time and can think through tasks in several steps. The boundaries between tasks are dissolving.

How does an AI "learn"?

Modern AI systems learn from data, similar to how you learn from experience. The process is called machine learning:

  1. Collecting data: The AI receives huge numbers of examples (texts, images, numbers)
  2. Recognizing patterns: It finds statistical patterns in this data
  3. Making predictions: Based on the patterns, it can solve new tasks

Language models like ChatGPT (also called LLMs, large language models) have learned in this way from huge amounts of text which word is most likely to come next in a sentence. Complete answers are built from this.

Easy to remember: An AI is not "programmed" to give a specific answer. It learns from millions of examples which answers are likely to make sense.
What best describes the current state of AI?
Correct! Today's AI systems are "weak AI": specialized, but very powerful in their field.
Not quite. Today's AI is "weak AI": it can solve certain tasks extremely well, but it cannot think like a human.
Key Takeaways:
  • AI is an umbrella term for programs that solve "intelligent" tasks
  • Today's AI is "weak AI": specialized, but powerful
  • AI learns from data, not from preprogrammed rules
  • "Strong AI" like in the movies does not exist yet

This lesson is part of a course with final exam and certificate. View the course "AI Starter Course"

Next →