Artificial intelligence is more than just talk – Google’s top inventor

14 Jan 2010

Before joining Google, the internet giant’s head of research Peter Norvig was NASA’s highest-ranking computer scientist. He was recently in Ireland to give the Boole Lecture at University College Cork.

You have been instrumental in creating the Google we know and use today as well as the development of the NASA Mars Exploration Rover. What inspires you to create?

I look at the opportunities for how you can do something better with the technologies that are available. It takes a lot of effort to keep somebody alive in space and if you want to send them to someplace like Mars it costs an extra amount because you have to bring them back.

If you send a robot, it doesn’t cost so much if the rocket blows up. So if you could just make robots capable enough we could explore a lot more for the same budget and wouldn’t risk lives. It seemed like a good thing and the technology was just getting to the point where that would be feasible.

I guess it’s the same thing at Google: if you had the choice, you would go to an expert, ask a question and get a reasoned response, but maybe technology could do almost as good a job and do it faster.

You wrote a book on artificial intelligence and are currently looking at the future of search. Do you think we’ll soon be talking to our computers?

Yep, that is happening to a degree now, but we’re not using the same language. We’re talking to a search engine in keywords rather than in whole sentences and it doesn’t quite understand us as well as a person would.

But on the other hand it is giving us answers that a person wouldn’t, so it has its strengths and weaknesses.

Where do you see technology such as the internet going? For example, the next 1 billion people to come on the internet will do so via mobile phones.

Mobile is going to be extremely important. Our interaction will be different – we won’t be typing as much, we’ll be talking more and speech recognition is going to be increasingly important and display a lot smaller.

The advantages with mobile are that if you’re in a specific location and you ask a specific query then – because of GPS – there’s going to be an answer that’s appropriate to the location.

What makes Google researchers get up in the morning? 

Some of it is need, in terms of what is it I want to do and tried but I can’t. Or we look at search logs and say here’s somebody seeking an answer and we couldn’t help them. Some of it is driven by opportunity, by saying: “I was working on something, I got this solution and I think some of this would relate to a related area” and figure out if there’s a need for that.

How do you co-ordinate so many people who are inventing stuff all the time, resulting in a product?

It’s a little bit of chaos. People have ideas and what we try to do is make it easy to test the ideas out. It’s a small investment and doesn’t have to be a big strategy decision to say we’re going to try something out.

So one or two people should be able to say: “I saw something that might work and I can build it and test and make a demo, try it out.” Maybe it’s not the final product, but it would give other people an idea if it is going to work or not. Other people see it and may then get behind it. Only at that stage do we come in and say yes, we’ll make it an official product.    

By John Kennedy

Photo: Google’s head of research, Peter Norvig

John Kennedy is a journalist who served as editor of Silicon Republic for 17 years