Professor Miranda Wolpert

And we know that not all AI is built to be sort of therapeutic or helpful.

On the other hand, we also know that humans, therapists, are also not always helpful and may also sometimes worsen conditions for people.

So I think you need to take that in the round.

What I also think is really exciting is we're seeing completely different forms of therapy come forward

And I think we'll see more of those as we go forward that are not based on a model of two human beings sitting down and talking about the nature of the difficulties.

They may be based on completely different things.

So, for example, there's a set of researchers we're funding who are looking at work for people with psychosis who hear these very distressing voices.

and are learning ways to challenge those voices.

And the way they're training people to challenge those voices is to have a sort of avatar of that person that they can sort of talk back to and learn how to speak back to.

At the moment, that involves a human therapist talking them through.

But again, in the future, if that could be supported by AI, there may be ways of training people on how to address these thoughts, how to address these very distressing thought voices that allow them to get on with their lives.

So I think there are going to be a whole range of different interventions

that are going to be supported because we're going to be able to escape from a sort of assumption that of the way that what therapy means is going to mean something very different within the next few years.