Could AI companions comfort people with dementia in times of distress? Advocate Gwenda Darling and researcher Jill Bennett discuss groundbreaking new technology developed at University of New South Wales.
Category
🤖
TechTranscript
00:00Could these artificial intelligence powered companions developed by
00:04researchers from the University of New South Wales offer comfort to people with
00:08dementia? Advocate Gwenda Darling thinks so.
00:13I can see how useful it would be, particularly the night, and particularly if you aren't sleeping, or you're sundowning, or you're going through the stage where you truly believe somebody's taken something and you're pulling everything out of the wardrobe.
00:29And perhaps you could vent to the AI and you know that you're not going to be
00:34deemed as stupid. And I, having talked to it, I talked to Donna, the name of the
00:42character, I talked to Donna and said to her, has she pulled her sweet peas out
00:47because it's time to change the sweet peas over for petunias. And she said yes,
00:52she thought she'd grow some herbs, some basil and some mint, and that led me into
00:56the conversation with her about cooking lamb. Did she use mint or rosemary on her lamb? And she talked about the flavours, and then I thought I'm actually going to talk about something interesting, and I said my mother always cooked the lamb in the fuel stove, and Donna talked about how technology's changing, and we discussed air fryers and the difference, and then I thought, yeah, but I'll make a mess in the kitchen, and she talked about some options of focusing, perhaps playing some music.
01:24Professor Jill Bennett, head of the University's Big Anxiety Research Centre, explains that these companions are far from your average chatbot.
01:34Our companions are distinctive because they're all made with people with lived experience of dementia, mental health issues, real people who tell us what exactly they want from a chatbot.
01:51One of the main distinctions with our characters is that whereas most of the chatbots that you experience are reactive, and so they just respond to what you say in the moment and try to prolong the conversation, pretty much agreeing with everything you say.
02:10Ours are a little bit more considered than that. We've developed them so that they can, first of all, recognise people's moods, and the other thing that we're attempting to develop is their capacity for actually planning with people.
02:30This is very important because it's fine if a chatbot agrees with you most of the time because it will make you feel good and validate it, but if your mood's very low or you're depressed and very pessimistic about life, what you probably need or benefit from is someone who gently challenges those beliefs.