Hello and welcome to the UKITEL podcast. I'm Sarah Hall, Deputy Director at UK-Nacheting Europe.

And today I'm delighted to be joined by Louisa Moore, who's a professor of political geography at Durham University.

She's particularly interested in how

forms of data and algorithmic analysis, and we'll explain what that is in a minute, are changing

ideas around state security and what it means to live in society today. Her most recent book

is called Cloud Ethics and was published by Duke University Press,

and Louise is currently halfway through a major five-year project, funded by the European

Research Council, which examines these themes. Welcome, Louise. So let's start really simply.

Can you explain to me what machine learning technologies are? So

these are totally central to your research and what risks they might pose. Yes, certainly. It seems

like a very simple question, doesn't it? But actually, of course, the discussion around machine

learning as a concept goes back as far as the 1950s.

And actually, we can find computer scientists in the 1950s, Arthur Samuel, for example,

describing machine learning as giving computers the ability to learn without explicitly being programmed.

And I think there's something about those debates in the 1950s, even Alan Turing was involved

in this question of whether it was possible for a machine to learn something in excess of

direct programming that had been controlled by a human.

And so, though we often think of this as a kind of new issue that we're talking about

in contemporary society, in fact, this goes back nearly a hundred years, this kind of central

question around whether it's possible for a machine learning technology to discover something

in data beyond the things that have explicitly been programmed. So when we think about machine

learning now, we're most commonly talking about things called neural networks. And the neural network

...