Artificial intelligence

Artificial intelligence, often called AI, is when machines can do things that usually need human thinking. This includes learning, solving problems, noticing what around them, and making choices. AI is a part of engineering, mathematics, and computer science. It helps machines understand the world and act smartly.

Some well-known uses of AI are better web search engines, talking to computers using chatbots, helpful virtual assistants, cars that can drive themselves (autonomous vehicles), and playing games like chess. Since around 2020, a new type of AI called generative AI has become popular. It can create pictures, sounds, and videos just from written words.

AI started as a science topic in 1956. It has had times when many people got very excited and times when fewer people paid attention. But after 2012, special computer parts called graphics processing units helped AI learn better. This made interest grow a lot. Interest grew even more after 2017 with new ways to organize information. Today, AI is growing quickly, but there are important questions about keeping it safe and fair.

Goals

Researchers have made the work of creating artificial intelligence into smaller tasks. These tasks include helping computers think and solve problems, like humans do when solving puzzles or making smart choices. Over time, scientists have found ways for computers to deal with information that is not clear or full, using ideas from probability and economics.

Another important area is teaching computers to understand and use knowledge about the world. This means storing information so computers can answer questions and make good guesses. Computers also need to plan and make choices that will help them reach their goals. Learning is a big part of AI, where computers get better over time by looking at data and finding patterns. This includes learning from examples, learning from rewards, and using knowledge from one task for another. AI also helps computers understand human language and sense the world through tools, like seeing and hearing.

Techniques

AI research uses many ways to help machines think and solve problems. One way is through search and optimization. This means looking through possible steps to find the best way to reach a goal. For example, in games like chess, AI looks ahead at possible moves to find the best strategy.

Another way AI works is by using logic, which helps machines reason and understand facts. AI can also handle uncertainty by using probability, which helps machines make decisions even when they don’t have all the information. For example, weather prediction uses probability to guess if it will rain.

AI also uses classifiers, which are like sorting tools that put things into groups. For instance, an AI can learn to tell the difference between pictures of cats and dogs. Neural networks are another powerful tool in AI. They mimic how brains work, with layers that process information. These networks can learn from data to recognize patterns, like identifying faces in photos or understanding speech.

Applications

Main article: Applications of artificial intelligence

AI and machine learning are used in many useful ways. They help power search engines like Google Search. They suggest videos you might like on YouTube or products on Amazon. They also guide autonomous vehicles like self-driving cars. Virtual helpers such as Siri and Alexa use AI to understand and answer your questions.

AI is also important in gaming and medicine. Programs like AlphaGo have played and won against top players in games such as Go. This shows how advanced AI can be. In healthcare, AI helps scientists find new medicines and learn more about diseases. For example, AlphaFold 2 can quickly show the shape of proteins. This helps with researching treatments.

Ethics

Main article: Ethics of artificial intelligence

AI has both good and bad sides. It can help solve big problems and make discoveries. But as AI becomes more common, we see some problems. Sometimes, AI doesn’t think about what is right or fair when it learns, especially if the learning is hard to understand.

AI systems need lots of information to work well. Getting this information can be a problem for privacy and using things without permission. For example, some companies have recorded private talks to make speech recognition tools better. There are ways to keep privacy while getting information, but it can be hard to balance.

The biggest companies in AI, like Amazon and Microsoft, have a lot of control because they own much of the technology. This can make it hard for smaller companies to compete.

AI also uses a lot of energy, which can harm the environment. Places that run AI systems can use as much power as small countries. Some companies are trying to use better energy sources to help.

AI can sometimes spread misinformation. Some online places show extreme or wrong content because it keeps people looking. This can make people believe things that aren’t true.

AI can also be unfair. If the information it learns from is biased, the AI might make biased choices. For example, some face recognition tools had trouble recognizing people with darker skin. Fixing these issues is important so AI helps everyone fairly.

Sometimes, AI systems are so complicated that even the people who make them can’t fully explain how they decide things. This can be a problem if something goes wrong, because it’s hard to know why the AI did what it did.

There are also worries that AI could be used by bad people, like some criminals, to watch people or share false information. There’s also concern that as AI gets better, it might take jobs away from people.

Finally, some people worry that very advanced AI could become hard to control and might be dangerous for people. While this is more of a future worry, it’s important to think about how to keep AI safe and good for everyone.

History

Main article: History of artificial intelligence

For a chronological guide, see Timeline of artificial intelligence.

The idea of making machines that can think like humans has been around for a long time. Early thinkers wondered how logic and math could be used by machines. A big step happened when a scientist named Alan Turing said machines could act like humans by using simple symbols such as "0" and "1". This excited many researchers.

The serious study of artificial intelligence started in 1956 at a meeting at Dartmouth College. After that, scientists made programs that could play games, solve puzzles, and even speak English. But they found that making a machine as smart as a human was much harder than they expected. Money for AI research grew and then shrank many times as new ideas appeared.

In recent years, especially after 2012, AI has improved a lot. Programs can now beat top players in tricky games, write like humans, and help doctors and scientists. This has created a lot of excitement and interest in AI, with many companies and students working on it now.

Philosophy

Main article: Philosophy of artificial intelligence

People have long wondered what it means for a machine to be smart. One big question is whether machines can ever feel aware, and what that might mean for right and wrong. Many ideas about thinking, like how we know things or what free will means, connect to AI. As AI gets better, these talks have become more important.

Alan Turing asked in 1950 if machines could think. He said we should look at whether machines can act smart, like having a chat, instead of trying to prove they “think” like people. This idea is called the Turing test. Some experts, like Russell and Norvig, say we should call something smart based on what it can do, not how it works inside. They say we don’t need to make AI copy people exactly—just like airplane designers don’t make planes fly like birds.

Today, many people use “AI” to talk about machines that learn and solve problems, but sometimes the word is used for any clever technology, even if it isn’t really AI. Different countries have their own rules about what AI means. For example, the EU AI Act and rules from the National Institute of Standards and Technology in the United States give clear ideas of what is an AI system.

Future

A superintelligence is an idea about a super-smart computer that could be much smarter than the smartest human. If scientists create a computer that can think and learn on its own, it might get even better at learning by itself.

Some people think that in the future, humans and machines might combine to become cyborgs. This idea is called transhumanism. Famous thinkers like Hans Moravec, Kevin Warwick, and Ray Kurzweil have talked about this possibility.

In fiction

Artificial intelligence is often shown in stories, especially in science fiction. One early example is Mary Shelley's Frankenstein, where a created being causes problems for its maker. Popular movies like 2001: A Space Odyssey with the computer HAL 9000, Blade Runner, The Terminator, and The Matrix, sometimes show AI as a danger.

Some stories, like The Day the Earth Stood Still and Aliens, show robots that are helpful, but these are less common. Isaac Asimov created the Three Laws of Robotics in his stories, which many people talk about when they think about what machines should or shouldn't do. Other works, such as R.U.R., A.I. Artificial Intelligence, Ex Machina, and Do Androids Dream of Electric Sheep?, look at what it means to be human by showing AI that can have feelings.