Send me a message

Daniel Hulme
Daniel Hulme

What is: Artificial intelligence?

Posted on 3 June 2016

Artificial intelligence was once firmly in the realm of science fiction alongside UFOs and humans with super powers, but few will have failed to notice the recent increase in commentary about AI from almost every news source. We have all seen Hollywood’s interpretation of AI, but in the real world, what exactly is it and how does it work?

The basics

There are a myriad of definitions of AI out there, not least because in order to define artificial intelligence, you first need to define real intelligence. That’s not easy. One which I have come to like is that of computers performing adaptive goal-directed behaviour. Goal-directed in the way that the computer works toward something: solving a problem or performing some task, and adaptive in the way that the system can change its approach appropriately to better achieve that goal.

What does it look like?

Providing a concrete example of AI is difficult. It is a common response that once some milestone in AI development has been achieved, people no longer regard it as AI but simply a computer system. Looking back to the famous Deep Blue computer that beat chess grandmaster Garry Kasparov in the ‘90s. This computer changed its strategy throughout the game in order to beat the best human player. This is certainly adaptive goal-directed behaviour, but a chess computer game no longer really feels like AI.

“#AI: computers performing adaptive, goal-directed behaviour”

Nowadays, examples of AI are really everywhere but easily overlooked. Case in point: the personal assistant on your mobile phone which learns your routine and serves up the right apps or content for you. That technology was unthinkable a few years ago, but now the status quo. If you look closely enough, AI according to the definition above is everywhere and it has the potential to bring speed and performance improvements to many problems once only tackled by humans.

Cheat sheet

With the inextricable links between artificial intelligence and science fiction, unsurprisingly there are a lot of niche references in the related commentary. Here are a few to get you started.

Singularity A hypothetical point in time where AI performance surpasses human performance. Some people get very excited about this, so it’s good to know what they’re talking about.
Watson The AI system developed by IBM which famously beat the two human champions on the US TV show Jeopardy! Widely seen as a leader in AI technology.
HAL9000 Very common AI / Sci-Fi reference. HAL9000 is the AI computer that controls the spaceship in 2001: A Space Odyssey. As you might expect, HAL9000 turns evil and tries to kill the astronauts who want to disconnect it.
Turing test A test of a machine’s ability to emulate human behaviour. A machine passes the Turing test if an unknowing person can have a natural language conversation with it and believe that they are having a conversation with another person.

Something to add?

Let's take the discussion offline

Alex Lilburn

Alex Lilburn

Start a conversation

Send me a message

Alex Lilburn
Alex Lilburn

Message sent

Thank you for contacting us, we will come back to you as soon as possible