What IS Artificial Intelligence?

Search for a definition of ‘artificial intelligence’ in Google or Bing and you’ll see definitions that all coalesce around the idea of machines doing the work of humans, of being ‘as a human’.

You’ll see references to robots that eventually turn on humans and become the last thing humans ever develop.

But all of those doom, gloom and ‘human-like’ cross-references are not helpful if you want to understand what artificial intelligence is today, right here in the here-and-now.

So we particularly like Malcolm Frank, Paul Roehrig and Ben Pring’s useful definition, from their book, ‘What to do when machines do everything’:

“AI is an area of computer science that focuses on machines that learn.”

It does away with all of the ‘machines replacing humans’ and ‘machines behaving like humans’ rhetoric, and strips the working definition back to its essential parts.

The authors slice through their definition even further, dividing AI into three subsets:

1. Narrow AI;
2. General AI; and
3. Super AI.

Narrow AI is the artificial intelligence we are seeing at the moment; it focuses its attention on a narrow band of activities and interests. Think self-driving cars, or ordering an Uber, or asking Siri to read you your emails. The system might be very good at handling those tasks, but in reality, it is doing one or two things very well, and it’s focused on meeting one particular business goal. As the authors say, try asking your Waze app if your lunchtime cream cheese bagel fits in with your current diet.

General AI is what drives the Singularity pundits, and can be seen in movies like Her and Ex Machina. It is the pursuit of a machine that has the general intelligence of a human. Thus, it can discuss politics, tell a joke, and drive a golf ball within seconds of each other. One researcher puts it this way:

“Go into an average American house and figure out how to make coffee, including identifying the coffee machine, figuring out what the buttons do, finding the coffee in the cabinet, etc.”

This is easy for any adult to do, but far beyond the capabilities of machines at the moment — it is insanely difficult for them.

Super AI is the stuff from which dystopian nightmares come. It is the level of intelligence that gives rise to movies that show robots taking control. Think I, Robot and the Terminator series. It is where machines can build themselves into entities with IQs measured it the tens of thousands. But all of the researchers working in the field — those who are actually building the machines — are saying that we are highly unlikely to witness such things in our lifetime. Says Andrew Ng, Chief Scientist at Baidu Research:

“Worrying about [general or super] AI is like worrying about overpopulation on Mars, before we’ve even set foot on it”

Conclusion

Ignore the hype. Artificial Intelligence, as we know it in our working world, is the narrow focus of machines on particular tasks or goals. It is mildly disruptive at the moment, but will become more so as businesses come to understand it better, and derive more value from it.

It is displacing small numbers of people at the moment, but if those people are prepared to retrain and reskill to meet the requirements of this new industrial revolution, smart organisations will find work for them.