What is artificial intelligence and machine learning?
Artificial intelligence (AI for short) is the term given to machines that mimic human intelligence. Machines with AI are capable of making decisions and can change their behaviour and learn. Computers, robots and some software can be trained to solve problems and complete tasks, learning from their mistakes to improve their performance over time.
Human intelligence and how we learn
Most people think of intelligence as our human ability to learn, make sense of things, solve problems and use language to communicate. Learning is a big part of what makes us intelligent.
We are learning when we acquire new skills and knowledge. Humans learn in a number of ways but the most common way we learn is by repeating actions. For example when learning to play a musical instrument, a person will start with simple notes and songs. Then they keep practicing and repeating until they’ve mastered them, before moving onto something harder. It’s the same with learning languages, sports, driving a car and anything new. We repeat until we have learned how to do something.
How do machines learn?
Machine learning works in a similar way to human learning. To train a machine, we have to give it a set of data. With the data as an example, the machine can use algorithms to predict how best to behave. If the machine makes the correct prediction, it learns to repeat it. And if it makes a mistake, it tries not to do it again. In this way, as long as the machine has been given the correct way to measure success, it is learning better behaviours.
AI in everyday life
Phones, computer games and even hospital equipment use AI to improve how they work. Services like Netflix, Spotify, YouTube and Amazon use data-driven AI to learn what people like and make recommendations. Computer games can see how a player behaves and adjust its behaviour to make the game harder. A lot of research is going into using AI in healthcare so that doctors can diagnose health conditions better and predict illnesses before they have even happened.
Google translate uses AI to make more accurate translations. And smart assistants on our phones like Siri, Alexa and Google Assistant all make use of AI to give us better information.
The dangers of AI
Scientists like Sir Stephen Hawking have expressed concerns about machines becoming too smart. Machines that are smarter than humans could cause a lot of problems.
Another problem is the risk of bias in the way the AI is programmed. To teach AI, we use data and experience to tell the computer what a good outcome is. This is not unlike how parents teach their children how to behave. Machines, like children, can only learn from the information they are given. If they are given data that is bad or biased, the algorithm the machine uses will make that worse. So people working in AI have the challenge of creating AI that is helpful, not harmful.