Hello, my name is David Ronsman and this is Talking Politics. Today's Talking Politics

Guide is with Jennifer Cobb, who is leading a big research project in Cambridge on trustworthy

technology and she is going to be talking to us about machine learning.

These Talking Politics Guides are brought to you as ever in partnership with the London

Review of Books, who's summer sale with the Paris Review, two subscriptions for one

low price, is open to Talking Politics listeners. Head to lrb.co.uk forward slash guides for

more information along with the usual lists of further readings from the lrb archive.

Let's start with just some basic definitions so we know what we're talking about. Can you tell

us what we should understand by artificial intelligence?

Okay, so artificial intelligence, what people mean when they're talking about that is using

machine learning. It's not robots that are coming to kill everybody in webachimality,

it's just machine learning, which is basically a process of getting a machine on computer

to train itself, to spot patterns and correlations in big data sets, really big data sets of

millions and millions of records. Without having to actually specifically program it to

do it itself, it can adjust itself and change its processes and its algorithms.

As a form of intelligence, is it anything like human intelligence yet?

Oh no, no, there's a big difference between artificial intelligence as it sort of talked

about now, and artificial general intelligence, which is the kind of more human-like intelligence.

But there's also no reason why even if we got artificial general intelligence that it

would be human-like, it could be a whole other form of intelligence that we don't currently

recognize, you know, there are other forms of intelligence.

So the machines we have now that are drawing through these vast troves of data looking

...