Julia Shaw

And so that was the research of Elizabeth Loftus and Saul Kasson.

So false confessions was Saul Kasson and false memories was Elizabeth Loftus.

And I was just doing them both at the same time.

And the question was, could you get people to believe that they committed a crime that never happened and confess to it?

And not just that, but

believe that it actually happened.

So remember it.

And the answer to that in short is yes, you can, especially using specific leading and suggestive interview techniques.

And so the procedural learning from that, which is what I'm most interested in, I don't like that's sort of a party trick to be able to actually do it.

And that's just so that you can then take that and go, okay, well, how do we prevent this?

And so I've since trained police, lawyers, I've trained people at the ICC, the International Criminal Court, who deal with collective memory.

So they deal with hundreds of witnesses at a time in war crimes.

And the question is, how do we try to preserve as original as possible memory without contaminating it?

Because, well, or at least without contaminating any more than it already is.

And that's where social psychology, I think, excels, is that we have done lots of research on how social settings change what people say, and to some extent, what people believe.

And I think that's also where, actually, the leap to things like AI, I think, is not far, because ultimately, the way that we're engaging with large language models is that, and generative AI in general, is that it's structured as a social interaction.

It's structured as a conversation most of the time now.

And that is...

what we do.

That is literally what I train the police on doing is how to make sure that you don't distort people's memories in the process and how to ask good questions.