What is Artificial Intelligence? Intelligence, Evolution and Types of AI

Hello students, welcome to Kwickprep. Your phone finds the fastest road home and suggests songs you enjoy. Is the phone really thinking? Today we learn what intelligence means, how AI grew, its three types, and how AI, machine learning and deep learning fit.

Let us start with a simple word, intelligence. Intelligence is the ability to learn from experience, like a child learning that a hot tawa burns. It also means understanding a problem and finding a way to solve it. An intelligent person makes decisions, like choosing the shorter queue at the canteen. Finally, intelligence means adapting when the situation changes, like taking an umbrella when clouds appear.

Now compare a human with a machine. A human learns from experience, while a machine learns from data, which means facts and examples stored in a computer. We sense the world with eyes and ears, and a machine uses a camera and a microphone. We decide by thinking, and a machine decides using a model, which is a pattern it has learned. And a machine has no feelings, even when it talks politely.

So here is our definition. Artificial intelligence, or AI, means a machine doing tasks that normally need human intelligence. These tasks include seeing faces, understanding speech, predicting results and making decisions. Pause and think: is a simple calculator AI? No, because it only follows fixed steps and never learns anything new.

Now a short history of AI. In nineteen fifty, Alan Turing asked if a machine can think, and suggested a test to check. In nineteen fifty six, John McCarthy and others named the field artificial intelligence. In the nineteen seventies and eighties, expert systems followed rules typed in by human experts. In nineteen ninety seven, the computer Deep Blue beat world chess champion Garry Kasparov. Around twenty twelve, deep learning made computers very good at recognising pictures. In twenty twenty two, chatbots like ChatGPT reached millions of ordinary people.

The big change in this story is from rules to learning. In a rule based system, a human writes every rule, and in a learning system, the machine finds the logic itself. Rules must be exact, while learning needs many examples, like thousands of photos of cats. When a new case comes, fixed rules often fail, but a learning system can adapt.

Next, AI is grouped into three types by how much it can do. Narrow AI does one task very well, like face unlock or a chess program, and all AI today is narrow. General AI would learn and do any task at human level, and it does not exist yet. Super AI would be smarter than humans in every way, and today it is only an idea.

Pause and predict the type for each example. A chess app that cannot play Ludo is narrow AI, because it knows one task. A voice assistant that sets alarms is also narrow, even though it can do a few jobs. A robot that could cook, teach and drive as well as a human would be general AI, and no such robot exists today.

You already use AI every day. Map apps study live traffic from many phones and predict the fastest route and arrival time. Video, shopping and music apps recommend what to watch or buy next, based on what you liked before. Voice assistants like Google Assistant, Siri or Alexa turn your speech into text, understand it and reply. Face unlock checks your face against the face it learned.

Let us look closely at one example, a map app. It collects data, the speed of many phones moving on each road. From past data it learns patterns, like a main road being slow at six in the evening. Then it predicts, and tells you to take the other road to save ten minutes.

Finally, picture three circles, one inside another. The biggest outer circle is AI, every way of making machines act smart, even with fixed rules. Inside it sits machine learning, where the machine learns patterns from data, like an email spam filter. The smallest circle, inside machine learning, is deep learning. It uses networks with many layers, loosely inspired by the brain, for jobs like face unlock.

The circles give us three exam ready facts. All deep learning is machine learning, because the inner circle sits inside the middle one. All machine learning is AI, because the middle circle sits inside the outer one. But not all AI is machine learning, since an old rule based chess program is AI that never learns.

Let us recap. Intelligence means learning, solving problems, deciding and adapting. AI grew from fixed rules to machines that learn from data. Only narrow AI exists today, while general and super AI do not. Maps, recommendations and voice assistants all use AI. And deep learning sits inside machine learning, which sits inside AI.