Looking to the Future of Caregiving: Artifical Intelligence

Good caregivers are few and far between. It's a tough sector to keep staff for several reasons. So what's the answer for supporting an aging population or people with disabilities? Artificial Intelligence is already starting to play a role. We caught up with Mercer Gary from Drexel University in Pennsylvania about the positives and some ethical concerns when it comes to AI and caregiving.

 

TRANSCRIPT

 

Looking to the Future of Caregiving: Artificial Intelligence

00:03

Welcome once again to DDA's podcast, encouraging abilities. I'm your host, DDA communications manager, Evan Kelly. Now I just want a little caveat at the top of this, um, because we are talking about caregiving now, just to understand that developmental disabilities association, DDA is a community living agency that provides over 50 community based programs and services to children and adults with developmental disabilities and their families here in Vancouver and Richmond, British Columbia. So today we are talking about a topic that

 

00:33

It's kind of front and center a lot in the media these days and we're wondering about it how it might affect our lives That is artificial intelligence with the advent of online AI such as chat GPT and other applications for Video and music you name it AI seems to be paving the way forward in many respects But what are the limitations is it good for us is it bad? So today I'm joined by Mercer Gary she is a poss

 

01:00

postdoctoral fellow at the Hastings Center, assistant professor of philosophy at Drexel University in Pennsylvania. She has a PhD in philosophy and women's gender and sexuality studies from the Pennsylvania State University. She looks at conceptual questions and feminist ethics around normative significance of relationships in order to strengthen applied interventions in bioethics and ethics of technology. So,

 

01:29

That's a lot to take in to see with that. I'm not even sure what I read, but definitely some high level thinking there. Thinking there. So thank you for joining me today, Mercer. Great to be here. Thanks for having me. Now I wanted to talk to you because DDA, as I said, is a community living agency. We employ many people in the caregiving field, like in the hundreds, to be clear. And it's a field that is often tough to keep staffed, whether it's lower wages, the battle for the high cost of living here in twenty three, twenty

 

01:59

particularly here in the lower mainland of British Columbia where housing costs are crazy. Also, with the caregiving field, schedules are all over the map. I mean, we have care that is 24-7. Also, with other options, maybe the caregiving field is simply not as attractive as a job. So right now, we're running the risk of not having enough people to look after the most vulnerable. So that's where artificial intelligence...

 

02:28

may come into play, can AI at some point be used to help care for seniors or those with disabilities? Now many companies are developing technology to assist with caregiving and even DDA is working with a local team called Three Spheres of Reciprocity to develop a robot to assist our caregivers. They've actually developed two different kinds of robots at the moment. We're a long way I think from implementing things, but at least it's being looked at. So Mercer, let's get right into it. What...

 

02:56

What got you interested in the line of study that you do? Sure. So my expertise is in feminist care theory, which is kind of an umbrella term that spans some different approaches, but generally is interested in care as both a source of ethical value and a kind of relationship and as a marginalized kind of labor.

 

03:25

So it's really important to me to take into consideration both the side of the person receiving care and the need for receiving care that is dignifying, that is respectful, and that is meeting their needs, while also taking very seriously the way that care work is devalued socially and economically in our society.

 

03:51

and has been historically and currently delegated to some of the most marginalized people, predominantly women of color today. And so my first interest in AI and care came with the advent of social robots that you're already referring to, robots with some kind of capacity for learning, machine learning typically. And this led me to two major questions that have grown in my research.

 

04:20

One is a more abstract one within feminist ethics and the other is a more practical, concrete one in kind of feminist sociology, feminist economics of care. So in feminist ethics, the

 

04:38

The framework of care ethics is grounded in intersubjective relationships and says that those relationships are themselves valuable in a way that's often overlooked and also can give rise to ethical dispositions and ways of thinking about ethics that have not been emphasized in the history of moral philosophy.

 

05:06

So if we're focusing on that central caregiving relationship, it's a pretty big question, who can be party to that relationship? What does that relationship look like? And so when I first started hearing about these social robots, it proved to be a pretty interesting test case of kind of how far does this notion of the caring relationship extend and could it include something like...

 

05:33

a social robot or do we want to draw a line before we include technology in that relationship? So that question became especially interesting because people do seem to develop intense attachments to many social robots and especially ones designed to function as companions.

 

05:58

But a lot of people have some squeamishness and hesitation about the kinds of connections people feel, users feel. And I think there are really philosophically interesting questions about whether that kind of connection between a human user and a robot is capable of generating the same kinds of connections.

 

06:23

ethical value that we typically think of when we think of care. So when we think of AI looking after people, what is the actual potential in your mind?

 

06:37

Yeah, so I think that what's most interesting are the ways that AI might be able to assist us in supporting and valorizing human caregiving work. So I think both because of technological limitations and because of ethical and economic concerns about putting AI in the position of being the sole caregiver, we instead really wanna be thinking about AI in the supporting role.

 

07:06

And so what can artificial intelligence, whether that comes in the form of an AI-driven social robot, or whether it's an algorithm in an electronic chart or a monitoring system in a person's home, how can we use these technologies? And there's such a range of them under this broad umbrella of AI.

 

07:35

to support the real human caregivers, many of whom are unpaid and drastically overburdened, and the people in need of care, many of whom don't have enough say in the direction of their care and enough ability of self-determination within care processes, regardless of their ability. How can AI contribute to making...

 

08:04

the giving and receiving of care better for the people involved. Yeah, exactly. And now, as I mentioned, DDA is working on creating this this robot called Aether. And at the moment, the design is really to have this sort of robot roam around like a group home, for example, looking for spills, looking for, you know, if someone who's fallen and needs help.

 

08:31

you know, maybe identifying an open door or something like that. And even getting to, you know, hoping to get it to sort of recognize people's faces and be able to interact, monitor, you know, medicine timing for medicine, medicine intake, that kind of thing. We're still a little bit a little bit away from that, but I have seen it sort of in practice. What in your experience, what have you seen in terms of practice in the care setting? What are some success stories?

 

09:00

What are some failures or things you might be concerned about? So I've definitely seen robots like the kind you're describing, and they're used in all sorts of things. I mean, they're using customer service. They kind of roam some grocery stores now in different parts of the US and Europe, especially. And there have been...

 

09:27

some significant successes with those kinds of technologies. I think another one that has had some success is a larger robot designed to assist caregiving staff in moving patients, helping someone, pick someone up to both protect the

 

09:54

bodies of the care workers from injury and to make sure that people are, you know, being moved in the ways that they need to in order to not develop sores, etc. So that, I think, has been a significant success in some places. When we think about AI, we think about the

 

10:15

generative AI like chat GPT in the caregiving sector, I think that there's real potential for using it to synthesize research and get kind of a leg up on what is a really vast body of academic research that has the potential to contribute to important policy and practice

 

10:41

but can take a single researcher much longer to do analog by themselves. So those are some different useful ways of employing AI in the care sector. Do you feel this is moving at a... are we starting to embrace this too quickly?

 

11:09

So I think that what is definitely already happening is that these technologies are being driven by commercial demands, market demands. And so what's getting developed and what's getting rolled out?

 

11:34

is heavily influenced by what's going to be most profitable. And so I think what we may be doing too quickly is kind of settling in.

 

11:46

into certain grooves of technological development according to what's going to sell and not adequately exploring and developing other avenues that might actually be benefiting caregivers and care receivers more. Mm-hmm. Now, back to sort of talking about care. Care can be a pretty broad term, I guess. To me, it means comfort. Like, it's one thing to have...

 

12:15

something roam around monitoring falls and spills or medication distribution, but how Do you think AI could bridge that emotional gap?

 

12:25

Yeah, so I definitely agree that care is a broad term and people mean a lot of different things by it. And for a lot of scholars of care ethics, the emotional or relational component and often a kind of connection of empathy between parties and a caregiving relationship is really central. And so,

 

12:53

And within that framework, the ability to give care requires moral agency. It requires people to be, it requires someone who is giving care to be an ethical subject, capable of making decisions, acting on them, and responding with empathy. And so there are.

 

13:18

questions, some of them are kind of technical questions of what AI could do or how we could develop it, but some are much more fundamental about is it possible to generate the kind of consciousness, self-awareness, and emotional connection that is characteristic of caregiving in a...

 

13:42

technological artifact and most people are going to say no to that. It's possible. It might be possible. Who knows? I mean, the way the speed at which this is all sort of developing, to me, like having something that can recognize empathy might not be too far away, but it's hard to say at the moment. Yeah. So, I mean, it's...

 

14:07

It's a really interesting, difficult question because it brings us to what empathy is. If empathy is kind of a shared feeling based in shared understanding, shared vulnerability as a number of kind of philosophers will define it.

 

14:31

the lack of vulnerability on the part of a robot means that it cannot have the capacity for empathy in the way that humans experience it. We can think of even, you know...

 

14:48

I'm sitting here with my dog right now and I know that when I'm with my dog and she sees that I'm crying, she will, you know, reach out and lick my face or there are, you know, signs we see at least what we take to be signs of empathy, perhaps not identical to human empathy, but like it in other animals. But there is.

 

15:14

there's a shared condition of vulnerability there. And that is not something that a robot and a human user would share. Yeah, no, that's very true. Some of the, you know, often when we think of care, we think of people I think typically go to to seniors and people who are, you know, aging. But we deal with people with cognitive impairments and developmental disabilities. What concerns would you have

 

15:43

with AI surrounding care with these people, some who might be nonverbal, who just might not be able to communicate as well. So how do we, I'm gonna have this question a lot, how do we bridge that gap with AI? Yeah, absolutely. So I, disability studies is,

 

16:07

also part of my background and something I'm interested in and committed to. And I do think that especially the question of communication with someone who's nonverbal and the kind of really nuanced and, you know, multi-sensory communication that is

 

16:29

necessary to develop over a period of time between a human caregiver and a person receiving care who's nonverbal. That's a level of kind of sophisticated response that is going, seems to me to be difficult to approximate. So that would certainly be a worry. I think

 

16:55

this kind of broader question of what the integration of technology into caregiving does to the social value of care work.

 

17:05

is a big concern for thinking about care in disabled communities as well as in aging populations. So we currently don't, socially, as a North American society, does not value caregiving very much. We don't pay it well. We don't give good working conditions for it. We don't treat it as a prestigious.

 

17:33

or meaningful job, and we expect it to be something that, again, mostly women do on top of other work, and it's their participation in it is assumed and often not even really confronted to. So that we're integrating.

 

17:59

AI, robots and other AI-driven technologies into caregiving, is that saying that the work is so of such little value that we don't even need a human to provide it?

 

18:20

And if that's the case, then what levels of, what kinds of quality control of the care are we willing to put up with, especially for people who are vulnerable? And so I think that maintaining high quality care requires valuing it both economically and socially. Yeah, absolutely. And that's one of the problems. I mean,

 

18:49

You've hit the nail on the head is that we don't value that work which also you mean you could take it one step further Maybe we don't value seniors or we don't value people with developmental disabilities and it's about Shifting that whole thing. So I mean we can spend billions and billions and billions of dollars in developing AI But it might just be easier to pay people more, you know and create that that social shift to get people to understand that

 

19:17

you know, seniors and people with developmental disabilities, which is going to be all of us at some point, we need to treat, treat them like the human beings they are. But that also sort of, you know, pushes my mind into a whole other sort of sphere of looking at like here in British Columbia, we've worked hard over the last 70 years to get rid of institutions. DDA was very instrumental in creating community living and getting people back into the community. If AI advances that much,

 

19:46

I almost see like a step back into an institution in a way where we've got these people housed in this area where it's just covered by AI and nobody else. I mean that's a sort of a nihilistic way of looking at it, but I mean to me the potential's there. Yeah, I definitely think you're not alone in that.

 

20:07

that concern, that's often this kind of institutionalized AI robot run care facility is a real kind of dystopian scenario that people have been warning others about for as long as these robots have been in development. And I think that some of that is real, especially with the price tag of a lot of this technology,

 

20:36

individuals, families, people living in community, but really has to be bought by institutions. But I think there are a couple of things that might actually work in the other direction. And one of those is the possibility of AI-driven technology to support people living in community or living independently longer.

 

21:01

And the kind of development of that technology has been focused on aging populations, especially as we're confronting real growth of the older population in the U.S. and Canada, as some other countries are already experiencing, and a diminishing number of people capable of caring for them.

 

21:30

there are ways that building AI technology into homes can support independent living. And that might come in the form of, as you were discussing earlier, some monitoring systems that would detect whether a fall has taken place or whether someone hasn't moved in a while, and report that through appropriate channels.

 

22:00

So I think there's potential there, but the biggest question is going to be the price tag. Yeah. And I think, you know, the one thing with seniors and people with disabilities or disability community sort of in general is that as loneliness can be a huge factor in their existence. And it would be nice to see if AI, maybe they don't need help getting into bed or something like that. But, you know, having artificial intelligence around.

 

22:29

where you can have a conversation might dispel some of that loneliness in an artificial way, I guess, but it's, you know, in some cases, it might be better than nothing. Yeah, definitely. I think that those kinds of empirical studies which are being conducted and have been for a number of years now are really interesting. People, it, it,

 

22:55

Some of these socially assisted robots, companion robots especially, even with pretty low levels of artificial intelligence integrated, have been shown to act as kind of conversation pieces in a group living environment and in ways that go beyond kind of initial novelty phases, but really kind of sparking conversation between people about what this object even is. Is it alive?

 

23:23

I really like it, I don't know why, that sort of thing. People do form attachments to these robots and name them and have conversations with them and that happens both with people who experience cognitive impairment but also with people who do not have a cognitive impairment and are very, very clear that this is...

 

23:50

You know, this is this is a robot. This is not a living thing but still express affection for it and You know, yeah as you said there there may be situations where it's better than nothing and although we certainly shouldn't lose sight of the Social conditions that are leaving us with nothing if not other than a social robot. No

 

24:17

This AI is obviously probably creating a whole bunch of new laws that we haven't even thought of. I mean, what if there's AI is in the caregiving sector to some level and there's a mistake and someone dies? Who gets the blame for that? Yeah, this is a huge question and figuring out how to ensure

 

24:46

in the, you know, still heavily involved in reviewing, assessing, auditing the tasks accomplished by the AI in order to have that accountability.

 

25:08

absolutely critical. Where there are algorithms being used in hospitals to make decisions about increasing in medication, something of that nature. There's a recent article in the Washington Post, I believe.

 

25:30

about nurses being, human nurses being overridden by artificial intelligence algorithms on clinical decisions like these and that, you know, becoming a serious problem. And so figuring out how we have these checks built into our AI systems that are going to allow humans to assess what they're doing.

 

25:59

and are going to provide enough transparency for that kind of detailed human assessment is going to be crucial if we're going to use AI in the care sector in these ways.

 

26:23

It's such an important aspect for our lives and almost everything we do. You know, we look at technology in cars these days. There's semi-autonomous driving, or some cars are claiming that they can drive by themselves, but it's still such a difficult thing for humans to actually let go of the wheel and let something else control it, even though statistically it might be better.

 

26:51

for, say, fatalities on the road. And so in care, I think that would be the same thing. Like, how do we as humans let go of that? I mean, not let go of trust, but trust something else that isn't human, even though it might actually make a better decision.

 

27:13

Yeah, and I guess I think that for the development of, we can't just increase trust without increasing trustworthiness in the parties that we're trying to trust, right? So we need to make sure that AI is actually trustworthy, that we know its capabilities and they're consistent and they're,

 

27:43

They're consistent, they're transparent, and we know what to expect from them. Obviously trust isn't, trust requires something of a leap of faith and of going beyond what you explicitly know and can control. But there are, I think both technical,

 

28:11

for technical development and and development of human systems of accountability that will be necessary to have uh... a i dot it's trustworthy and so before we can talk about how to increase human trust in a i think focused on making a i'd trustworthy

 

28:37

And that also covers this idea, not even an idea, this fact that AI, like chat GPT for example, has sometimes made things up. I've read stories where lawyers have tried to drop cases and the chat GPT, or one of them, I'm not sure if it was chat GPT, actually made up cases to back up their argument and they were disbarred by using that. So we, it's, you know.

 

29:06

if it makes something up, but we still need to make sure there's some sort of failsafe, that that's not going to happen, particularly in the care or even the medical field. That seems like vitally important.

 

29:17

Yes, absolutely. Yeah, so again, we have to make sure that there are places for humans to be tracking and auditing what an AI system is doing and capable of catching these hallucinations, as they're called. Your, all of your education and expertise is in

 

29:46

largely in ethics. Are we going down the right path when it comes to caregiving? Um, I'm not confident that we are, especially if we're focused on developing caregiving technology that is going to save and make money, rather than thinking about how we're going to

 

30:16

the really disenfranchised people who are both doing the direct giving of care and the direct receiving of care. So I think that a first step is the meaningful integration of direct care workers and people receiving care, especially disabled people, especially older people.

 

30:42

into the development and implementation processes of AI-driven technologies in this sector. So ultimately, I guess, the ideal thing, unless AI was an absolute perfect thing and had, you know, autonomous robots to do things, that a hybrid model is still really the only best model in a sense. Yes.

 

31:11

a way of combining AI-driven systems with human expertise, human connection, and human accountability processes.

 

31:28

So where would people prefer to receive AI care as opposed to human care and vice versa? Yeah, so you know, because caregiving is caregiving in the experience of dependency, interdependency is so stigmatized in a lot of...

 

31:52

North American culture, especially there are a lot of people who say that, you know, they actually would prefer a robot or an AI driven operating system to respond to them in states of need than another human. So this might especially occur with intimate tasks of bathing,

 

32:17

But it also might occur in moments of emotional vulnerability where a person is, you know, disclosing their own emotional psychological difficulties and has feelings of either

 

32:44

shame in showing that to another human being or an increased feeling of dignity in having some kind of distance from the person or object providing for that need. And so this has come up.

 

33:09

both in disabled communities and also in service members, people in the military who have had more difficulty opening up to, say, a human therapist in the room than a distant

 

33:37

technology driven therapy chat bot. Interesting. And so there are both kind of situational and cultural differences around when we feel comfortable receiving care. And in those cases, some people do seem to prefer

 

34:07

assistive technology instead of humans. There are also questions about how people socialize and how people get their social needs met, and that may not be the same for everyone. There are socially assisted robots that have been designed and tested, especially for children on the autism spectrum, and there's been some significant success there.

 

34:33

with these child users feeling much more socially engaged by this technology than they are able to find at least with their peers at a given age, or some peers at a given age. So I think that there are non-trivial connections that AI systems can make possible.

 

35:03

which of these connections are made possible for people, which ones are developed by companies and which ones are affordable and accessible, and who's involved in the design and implementation process. But there are some potential ways that AI-driven technology might be preferable.

 

35:33

True human person people and that's very interesting. So it looks like the future I don't know if it's bright or dark, but it's definitely interesting as we move forward and into the care field and and AI in general Well, you have been listening to DDA encouraging abilities podcast My guest today has been Mercer Gary Paws postdoctoral fellow at the Hastings Center and assistant professor of philosophy at Drexel University in Pennsylvania We're talking about the future of caregiving

 

36:00

through the potential use of artificial intelligence. It's a fascinating discussion, and certainly something to keep our eyes on. Mercer, thank you for joining me today. Thanks so much for having me. Thanks for listening.