Can AI Provide Artificial Care?

[00:00:02] Speaker A: The benefit is the long term of later, when older adults actually really need these systems help because when they're really healthy. Early on, we found in our research that they resist adoption because they can still do a lot for themselves. And then for the folks who are experiencing cognitive decline, they actually do want to adopt these systems and they wish they would have done it earlier. [00:00:38] Speaker B: Welcome to the Age of Aging, a show about living well with an aging brain, produced by the Penn Memory center in the Michael Nadoff Communications Hub. I'm Terrence Casey. [00:00:47] Speaker C: And I'm Jake Johnson. How you doing today, Terrence? [00:00:51] Speaker B: You know, not great. I don't know if you were familiar with it, but this morning there was this massive outage across huge portions of the Internet. They called it the cloudflare outage and it was impacting a bunch of individual websites and ChatGPT and so on. But where it really got me, everyone on our team is, is familiar with my dependence on third party project management software and I lost access to that this morning because of this outage. [00:01:24] Speaker C: So what is this, this, this AI, this software that you're using, what does it do? [00:01:30] Speaker B: So I input all of my tasks and I include if there's a deadline or how much of a priority this is and what window of time I should be working on it, whether this is a work project or a home project or so on. And so it's helpful with things like making sure I get few edits on our latest episode of the podcast, but I also have used it to maintain a cleaning schedule in my home and washing my kids sheets, for example. [00:01:59] Speaker C: Well, thank God you're the boss, because I don't, I, I, I do not do any such thing. And besides scribbling down something in my, in my notebook in the morning in terms of organization, but that is very relevant to the topic today, which is AI and dementia care. This was a topic that we knew that we wanted to do something about at some point because it is such a huge thing that people are talking about. You kind of see it everywhere. It feels like every company has some kind new AI. So we wanted to capture a little bit of the, the many companies and projects and tech that's now being developed for older adults in relationship to AI. You know what I didn't realize when we started making this episode was actually just how vast the tech was going to be. All of the projects that are being worked on, and this is a topic that will definitely continue to come up again and again, maybe in a little bit more of a focused way, but this is more of a broad look. So in this episode, you're going to hear a lot of different voices in this space, all kind of discussing different AI systems that can benefit people in their homes as well as a hospital setting, helping nurses and clinicians do their job better. Then we dive into a specific company, New Day's AI. And you know, there's a lot of companies in this space and we chose New Days not for any particular reason, but we included this as an example of a company that's in this space now that is working on tech and AI for the care of older adults. And then we end the episode with a lot of ethical questions that I think we all are going to need to take into consideration if AI is going to become a part of dementia care and the care generally of older adults. [00:03:59] Speaker B: When you hear people talking about AI and some of the concerns, you often will read about some of the environmental concerns about it. I think we don't really get into that at all in this episode, but certainly we encourage people to learn more about their own environmental concerns of using AI, both for the care of older adults and for managing their children's laundry schedule. So I appreciate that recap. I'm really looking forward to it. But first, a word from our sponsor. [00:04:29] Speaker D: Understanding Longevity can better prepare individuals to make thoughtful decisions not only about their finances, but about their health and lifestyle, too. Matching lifespan with healthspan and pairing both with financial readiness has become a defining challenge. [00:04:46] Speaker A: The TIAA Institute wants to help build. [00:04:49] Speaker D: Longevity literacy to better support outcomes for all. [00:04:56] Speaker C: I probably don't have to tell you that in 2025, AI is everywhere. The term pops up so much in media that it's almost lost all meaning to me. So I googled the definition of artificial intelligence in researching this episode, and NASA defines AI as computer systems that can perform complex tasks normally done by human reasoning, decision making, creating, et cetera. Now, that's a pretty broad definition, and as you'll see today, just in the care of older adults, AI has a vast amount of applications. However, for all the problems AI seems to solve in elder care and dementia care, an equal amount of questions and considerations arise with how we use these tools in the most beneficial way possible. [00:05:44] Speaker E: I'm Alex Mihailidis. I'm a professor at the University of Toronto, and I'm also the former scientific director of the Ageworld Network here in Canada. [00:05:53] Speaker C: Dr. Alex Mihailidis has spent the last 25 years thinking about how we can use new technology to assist older adults. He said AI's ability to learn from and cater to individuals was the first way he and other researchers saw its potential with older populations. [00:06:11] Speaker E: As I've been often reminded by older adults themselves is that when you've seen one older person, you've seen one older person. A one fits all approach often doesn't work, especially if they have something like dementia or Alzheimer's or physical or sensory impairment. The role of artificial intelligence really was helping to learn about the person, learn about their needs, learn about changes in their abilities or disabilities, and then to adapt the way that the technology was able to operate and the types of assistance that it provided to the to the individual. [00:06:48] Speaker C: Dr. Mihalides said that these kinds of AI tools operate by tracking people's vitals, their level of activity, and their general health patterns through wearable devices and home sensors. While many of us have smart home systems that turn on our air conditioning automatically or turn on and off the lights at certain times, these new smart home systems would offer another level of assistance for older adults by introducing interventions based on the data it collects about the individual. [00:07:18] Speaker E: When we hear the term smart homes, we probably instantly go to things like, well, you know, my thermostat adjusts the temperature for me automatically. You know, we're talking about going a few steps further in the care of older people. So homes, again, that have sensors throughout them that can monitor again, their movement patterns and then use AI to, to adapt things as needed within the home or to call for help to the person or whatever it may be. We're looking at using sensors that are built into common pieces of furniture like your bed, that can automatically collect your heart rate and your breathing rate and your blood pressure. And so that way we're getting an overall picture of these physiological states as well of a person. Then again, adapt and change as needed. [00:08:02] Speaker C: Dr. George DeMuris is a professor at the University of Pennsylvania School of Nursing and School of Medicine and is the director for the Penn Artificial Intelligence and Technology Collaboratory for Healthy Aging, also known as Penn AI Tech. Penn AI Tech funds up to 12 projects focused on using AI to better the lives of older adults. [00:08:26] Speaker F: The advantage of passive sensing is that you don't require the person to learn to operate any new hardware or software. Instead, the technology is there in the background. So think of motion sensors that can detect motion, bed sensors that can detect restlessness at night, bathroom visits at night, or pulse and respiration. We have stove sensors that can distinguish between having left the stove on and walked away versus preparing a meal, door and window sensors that can tell you how much time you're spending inside versus outside the home. [00:08:59] Speaker C: Smart home AI systems could then use the data it collects from these sensors and contact a doctor if someone's blood pressure changes, or for example, a person is going to the bathroom more than usual. Or the AI could call an ambulance in an emergency situation, like if someone falls. [00:09:17] Speaker F: There's a lot of data that are being generated, and obviously the idea is not to have a human at the other end watch those data points come in in real time, but instead develop intelligent algorithms that can detect certain patterns and more importantly, notify us when there's a deviation from what would be the normal pattern for that person. [00:09:39] Speaker C: Dr. Mihalidi said that more current AI systems are being developed around prediction. [00:09:45] Speaker E: You know, it's great that we can detect the fall and get the person help, but the holy grail here, obviously is to prevent the fall from happening, right, or prevent the cognitive impairment from getting worse, or prevent that chronic disease, whether it's diabetes or something else, from getting worse. For example, we did a project where we took that kind of sensor data from a person's home and ran it through some predictive models, and we were able to predict, you know, who may develop worsened cognitive impairment over the next year with fairly good accuracy. It was like 90% and higher in terms of the accuracy. There's other projects that are doing things such as looking at predicting who may develop a urinary tract infection, for example, again, just by looking at their patterns of living and how much are they drinking, how much are they going to the bathroom, things like that. [00:10:36] Speaker F: More specifically, with fall prevention, we have a project using depth sensors that provide silhouette extractions. So we use those silhouette extractions to calculate stride length, gait speed, balance, and other gait characteristics that ultimately are used to calculate a fall risk score for that individual. So if your likelihood of experiencing a fall increases, there's things that we can do to prevent the fall from happening. For example, more tailored exercise programs, environmental modifications, making low cost fixes to keep people more safe in their home, removing loose rags, improving lighting. And so a lot of those projects really are targeting a proactive approach. You're trying to intervene before the adverse event happens. [00:11:25] Speaker C: And now a word from our sponsor, Rothkoff Law Group. [00:11:29] Speaker B: Caring for an aging loved one isn't easy, but you don't have to do it alone. At Rothkoff Law Group, we guide families throughout New Jersey and Pennsylvania along every stage of your aging journey. Our team of elder care attorneys, care coordinators, public benefit specialists, and elder care advocates ensure your loved one's well being. And your peace of mind. Rothkof Law Group, your partner in elder care advocacy and senior care planning every step of the way. Visit rothkofflaw.com for more information. That's R O-T-H-K-O-F-F L A W.com One. [00:12:06] Speaker C: Way AI is being used in preventative care is through targeted exercises that promote brain health. The company New Day's AI has introduced what it calls a new cognitive treatment based on clinical research. Co founders Babak Parviz and Daniel Kelly looked at the latest research in dementia care to find interventions that might help build resistance to dementia. One study they found was called the Internet Based Conversational Engagement clinical trial or iConnect study. This clinical trial found that adults who receive a 30 minute phone call four times per week about an open ended topic had higher cognitive test scores than the group that didn't. Dr. Parviz and Kelly took iConnect's methodology and utilized their AI avatar named Sunny to lead these conversations. Here's Dr. Parvis when someone comes already. [00:13:00] Speaker G: They would like to use use our service. They will see a clinician once every two weeks or once a month. So with some frequency they see a human clinician. That interaction involves assessment, involves feedback and involves therapy. So they see a human clinician. The interaction is not very frequent but in the in in between those human to human visits the our users would interact with our AI system. The nickname is Sunny nearly on a daily basis in the form of a long form conversation. So these conversations could be half an hour long or even even longer. And that's how sort of the cognitive exercises are implemented. [00:13:39] Speaker C: Dr. Parvez explained that these cognitive exercises shouldn't feel like taking a test. Talking to Sonny is meant to feel like having an open ended conversation with a friend about a topic like impressionist art. However, within these conversations, Sunny encourages the individual to reminisce about their own experiences related to the topic, use certain vocabulary and apply their own reasoning to the subject. Sunny then learns from these conversations and personalizes future conversations to the individual. [00:14:10] Speaker G: The other thing that Sunny does is that it has a self reflecting memory. So once it conducts a bunch of conversations it reflects on those and tries to surface things that are important and publishes those in the form of a report to our clinicians. So the clinicians have a view of actually what's going on in these conversations, which is really a fundamentally new capability because typically in between therapy clinicians don't have access to patients and don't know what's really going on there. So this really gives them a full view of nearly on a daily basis what's really happening to the individual. And we've also given our clinicians the ability to give guidance back to the AI system of what to do by using AI. [00:14:56] Speaker C: Newdays also doesn't have to train professionals to run these conversations, hopefully making it more accessible. Full 24. 7 access to Sunny and New Day's cognitive exercises currently costs $99 per month, on top of the cost of clinical visits that can be covered by insurance. New Day's AI is available in Washington, California, Texas and Florida, with hopes to expand across the country. A note here that the Age of Aging and the Penn Memory center do not endorse New Day's AI as a product, but highlight it as an example of one of many companies now utilizing artificial intelligence in the dementia care space in a hospital setting. AI tools also have the potential to be incredibly useful, particularly when it comes to predicting patient outcomes. Here's Dr. Demiris again. [00:15:49] Speaker F: The use of AI can also help us with efforts to predict how people may actually respond to certain treatments in the hospital. So by using valid algorithms, this can be a very powerful tool for clinicians who can then better plan and organize services and and place emphasis on different subgroups of patients based on those predictive algorithms. [00:16:16] Speaker C: So obviously, AI in the care of older adults is a massive space for innovation and has the potential to completely reshape our healthcare system. But while AI has the potential to improve various aspects of healthcare, it also raises a lot of ethical questions that researchers, providers and patients will have to take into consideration. Dr. Mailee Chang is a research scientist specializing in human AI collaboration. She said if older adults are going to allow AI into their homes, building acceptance and trust will be a challenge. [00:16:50] Speaker A: I mean, these AI systems are going to be in people's homes for the long term, right? So what does it take, first of all, for them to allow these systems to enter their homes and then to stay and continue using? And the benefit is the long term of later, when older adults actually really need these systems help, because when they're really healthy. Early on, we found in our research that they resist adoption because they can still do a lot for themselves. And then for the folks who are experiencing cognitive decline, they actually do want to adopt these systems and they wish they would have done it earlier. And so a critical question here is like, what is the timing, a good time to adopt? Because you don't want to introduce these systems when, you know, later on when they're already experiencing so much changes and adaption that it's another thing for them to try to figure out. Another really important piece of the adoption is trust. And with older adults care network there are multiple stakeholders. You have, you know, their adult children, their friends, professionals such as clinicians and lawyers. And so the AI will most likely need to report information to others in the care network that the older adult does not want shared. So there may be these moments where the older adult feels that they got betrayed by the AI. And so how do you design for acceptance in the long term where you can recover from these moments of potentially broken trust? [00:18:21] Speaker C: Who controls these AI smart home systems is another question Dr. Chang and her team are trying to parse out as it will likely change as a person gets older. [00:18:31] Speaker A: We saw that older adults define control in two ways. One is like who controls the AI system? And then the other is like who makes decisions for the older adult. So what's really interesting that we found is that for the control of the AI that it will change depending on the older adult's health. So when they're healthy, they want to control the agent. They want the agent to just listen to them be like passive in the background. But then later, if they experience cognitive decline, they want the agent to assert control on their behalf to advocate for them, because they recognize that their ability to advocate for themselves may decrease. And so they want the agent now to be more proactive and to have more decision making authority. And so this means that the agent's role will need to be dynamic. Our research also shows that older adults expect who the agent works for to change over time. And so when they're healthy at the beginning, they want the agent to work for them. And then later, when they experience cognitive decline, they want the agent to transition to work for their informal caregivers, such as their adult child. So this control element is really important when we talk about adoption and acceptance for the long term. [00:19:46] Speaker C: Consent is another question when thinking about AI in older adults. When a person consents to having these sensors in their home, providers will have to make sure the individual is aware of the degree of monitoring they're signing up for. AI avatars like New Day Sunny can also be tricky, Dr. Demaris explained. [00:20:06] Speaker F: So for a person who's experiencing cognitive limitations, it can be challenging sometimes to distinguish between what is the human response versus an AI response. AI tools that tend to be more anthropomorphic, like a virtual friend, a virtual companion. We need to make sure that we fully understand what the implications are when those are being used by people who might be experiencing cognitive decline over time and in some cases rapid cognitive decline. [00:20:36] Speaker C: In a hospital setting where AI could potentially summarize patients charts for providers or use patient data to predict potential health outcomes, patients will also need to be made aware and give consent to the fact that their information is being used. Here's eunji oh, a PhD student at the University of Pennsylvania School of Nursing whose research focuses on end of life care as well as how AI might assist sick patients with dementia and their families. [00:21:04] Speaker D: AI requires a lot of data, and in terms of palliative care, the discussions and documentations about end of life or death can be really personal and the patients during that stage can be very emotionally vulnerable. And when it comes to the secondary use of EHR data, it typically doesn't require additional consent from the patients. So the patients may not even be aware that they're the that the documentations that contain their conversations are being used for future algorithms. [00:21:38] Speaker C: Oh said that while she believes AI can do a lot of good for patients and their families, replacing human healthcare providers with AI, especially in an of life or palliative care setting, has a lot of implications. First, if the training data these AI systems use is in any way biased, then their care will also be biased. Additionally, there are a lot of subjective considerations in healthcare that O said AI just might not be able to navigate in the same way a human would. [00:22:08] Speaker D: Another concern is that palliative care focuses on distress and symptoms, which can be very personal and subjective. Even though two people are experiencing the same things, one person could be very distressed and one person may be okay. For example, like when we use structured questionnaires to measure anxiety and depressive symptoms, so we choose the level of severities for each symptoms. But if there is an NLP system that analyzes the clinical notes and flags, oh, this person has depressive symptoms, this person have anxiety symptoms, will that really align with what the patients perceive in terms of their own emotional state? [00:22:48] Speaker C: Ultimately, O said, we don't want people to lose trust in the healthcare system or become more distrustful than they already are. [00:22:56] Speaker D: I feel like the humanistic touch should never be replaced by AI. The one thing I worry about is the patient's trust in the healthcare system. Like if we replace certain things with AI, then is that going to impact the trust in the system? Especially when it comes to, like, historically marginalized populations, there is already a mistrust in the system and if we start identifying patients that can receive palliative care using an algorithm instead of a provider doing all of that, then will that worsen the trust? So I think there's a lot of guardrails that need to be set up and we need to highlight that it is not the AI that's replacing the providers. It's just a tool that can support the provider's decision making and the provider should definitely have the final say with. [00:23:48] Speaker C: All of those considerations. I don't want to paint AI technology as doomed to fail in the care of older adults. Obviously, this technology is still in its infancy and lots of decisions will have to be made going forward as to how it will be implemented in the most beneficial way possible for all parties. As we've talked about many times before on the show, healthcare and particularly dementia care already comes with a lot of ethical questions around consent, privacy and individual dignity. Adding AI into the equation might help with some of these problems, but will also inevitably bring about more complications. Some existing systems, like advanced Directives, may help with future questions around AI, but also come with their own complications. For Dr. Chang and Dr. Demiris, a multidisciplinary approach is essential to integrating AI into dementia care. Most importantly, though, these systems and their implementation must include the perspective of the older adults being cared for. [00:24:50] Speaker A: I think that a really true solution is that it will require a interdisciplinary approach that includes, of course, AI researchers and developers, clinicians, social workers and elder law attorneys. [00:25:02] Speaker F: Additionally, for many of our older adult participants in these projects, we want them to feel engaged and be part of the project as active decision makers. So not only being told we're monitoring something about you and you don't need to know why, but instead ask people to look at their own data. So I'm hoping that as technology continues to advance, we will be able to establish the right frameworks where the question is not which specific hardware or software are we using, but rather how do we ensure that whatever it is we're using is being used in a transparent, fair and appropriate way within healthcare. [00:25:47] Speaker B: Welcome back, Jake. Thanks again for that reporting. There is so much to unpack here. I just want to highlight a couple of things that stood out to me and then we can talk a little bit more about the episode. First off, I think the potential here with AI to personalize care for older adults is remarkable. I love the one point that was made early on that when you've seen one older person, you've seen one older person. I think that really applies not just to AI in care of older adults, but care of older adults, full stop. And it really resonates with what we're doing here at the Penn Memory Center. Dementia care and geriatrics is Never really one size fits all. And using AI to learn in individuals patterns through sensors in the home or conversational tools, that could really be transformative. The second is the predictive power of these systems. It's not about reacting to a fall or a health crisis, it's about preventing it. The idea that we can use these algorithms to analyze gate patterns or bathroom patterns, for example, can help prevent emergencies. It's proactive care and it could save lives and reduce stress for caregivers. But the biggest concern, of course, you brought it up at the beginning here is the ethical concerns. And they're just as big as technological ones. We have issues of trust and consent and control. And they're not abstract issues, they're deeply personal and unique. Who owns the data and who decides when an AI system can intervene? And how do we make older adults feel empowered and not monitored or babied? For me, what it really comes down to is that AI cannot replace humans human care, but it can augment it. But we have to get the ethics right or we risk losing the trust that makes care possible. And that really is going to need to be front and center as these tools evolve. [00:28:04] Speaker C: Yeah, and I think one thing we didn't address in this episode is the actual title, Can AI Provide Artificial Care? And I think that's because we don't really know yet. All of this stuff is still so new and developing as we speak. So I think we, we primarily wanted to propose the question to the listener and give them the information that we have and to really think about this question because I know we have a lot of caregivers in our audience and this is becoming a real question for older adults and for people that have an older adult in their life, like whether or not they're going to use these tools to help with their caregiving. And I'm curious for yourself, Terence, maybe a two part question first. Do you think that you would ever, you know, in 25, 30 years, I don't want to assume I know how old you are. Would you ever consider putting these, an AI smart home system in your home? And do you think that your parents would be accepting of a, of an AI system in their home? [00:29:07] Speaker B: That would be a yes and a no in order. I could not imagine my parents being comfortable with this as, as they age, period. I may ask them that at the Thanksgiving table this year just to see how much of a laugh I get out of it. I. However, I know what the concerns are and the risks and maybe I'm being too laissez faire. About it, but I just, I'm fine with it. I, I, I, you know, I, I kind of grew up with the Internet for better or worse. And so I, I, I think I have accepted that everything is out there already and maybe I, I don't know, I'm like a technological nihilist I guess at this point, you know, I, I, and, and I'd like to think too that by the time I'm an older adult, you said 25, 30, let's say 40, 45 years from now, that there will be many more safeguards and policies in place to protect individuals. But yeah, I think at this point, knowing the risks and accepting the concerns, yeah, I would be down with it. The impression I get is that you're not quite so nihilistic with your personal data. [00:30:32] Speaker C: You know, it's interesting. I think that I'm definitely skeptical of AI and I'm kind of just skeptical of new technology maybe in general at this point. And it's kind of ironic, I think people assume that people my age are very embracing of new technology. But I'm in my mid late twenties now, the beginning of my life. I didn't have any, you know, Internet or technology midway through my life. When I was about in high school, iPhones came out and then, you know, I've lived with social media kind of ever since. So I think we've, we've seen how unhappy sometimes social media and technology can make us. And I think that, okay, I'm speaking for my whole generation and I'm not, I don't mean to, but I'll speak for myself and I'll say that it's given me a kind of healthy skepticism I would think of new technologies like, like artificial intelligence. I will say I also think that I do see the benefits and I do see that for an older population who really need a lot of care and assistance and a lot of the times that requires a lot of people, people giving up their jobs, their, their day to day to help care for these people. If we can build in technology that can do some of that caregiving, I think that's pretty incredible and would be incredibly beneficial. You know, it's funny to try and think about what I would want when I'm an older adult and I agree with you. I think that hopefully there's going to be a lot more legislation and protections around that. But in terms of myself, I, I get obsessed with like data too. Like I kind of can't wear like Fitbits or Apple watches. Like I'll just obsess over every little thing. So I could see myself doing that with the AI. If it's like monitoring my heart rate when I'm watching TV or like measuring my gait or how many times I go to the bathroom, I'm gonna be like, really stressed about that. So, yeah, I feel complicated about. Was interesting to do this episode and really try and be fair about the pros and, you know, some of the questions that also arise about it now. [00:32:51] Speaker B: I find it not at all surprising that you and many of your peers reject it, if only because it's a tale as old as time that what the previous generation has done is the least cool version for the next generation. But when it comes to AI, we talked about the beginning. It is everywhere. And I think the biggest concern is that when we take things out of human control, we remove the human nature from it as well. I mean, you talked about potentially becoming obsessed with your data and hyper focused on that. And you can imagine when a machine is running entirely on data inputs that it's going to be a lot harder to manage, sort of an emotional adaptation to it. I think we see that a lot in AI and use of caregiving. Caregiving is not a one size fits all type of task. And AI would need to become much more adaptive to an individual based on their history to create an environment that avoids what we talk about as the uncanny valley. If you have AI, assisted caregiving or storytelling and so on, that is pretty real, but not quite real. It can actually be deeply unsettling for people rather than the heartwarming, I don't know. Do you ever watch snl, by the way? [00:34:26] Speaker G: Yeah. [00:34:27] Speaker B: So this. This past weekend, the Saturday Night Live did a skit on AI generated video based on photos. Had you seen this one? [00:34:38] Speaker C: No, I haven't. [00:34:39] Speaker B: I recommend checking it out. We could look to it in the show notes too. Obviously it's Saturday Night Live, so, you know, maybe not general audience entirely, but it's a skit about grandchildren visiting their grandmother in a nursing home. And they showed her an app that had scanned in photos from her house and then brought them to life. But in this skit, things you wouldn't expect or want to happen were happening with the images. Some were a little crude and crass, but some of them were just bizarre. Like a family dog that was just out of frame was picked up and it had two tails and no head, and it just freaked the grandmother out. And you and I both know that's sort of standard practice with AI. We're going to have these issues. And the kids are totally okay with this because they know they're familiar with AI, they're familiar with these different artifacts that come out of it. But it was deeply unsettling for the grandmother in the skit. It gets a little bit off the rails pretty quickly, as you can imagine. But I think that's a really. It's a legitimate concern that what's meant to be this comforting, caregiving moment can be kind of horrific pretty quickly. [00:35:50] Speaker C: Right? Yeah. When Dr. Demaris, I think, was talking a bit about how, you know, it's tricky with the AI avatars, like the Sunny is the one that New Days has. And how do you, you know, make sure that people understand that they're still talking to an AI system, you know, even when they're in cognitive decline? That sounds very complicated. And you don't want to be deceiving people. But then again, that tool is meant to help people. You know, it's meant to help with social isolation and also help people with their memory. So there's a lot going on there that is going to be very hard to figure out, I think. [00:36:33] Speaker B: So you just mentioned about letting people know whether they're. When they're talking to AI. And right after the episode, I was talking about how for AI to be successful, it will have to augment human care and not replace it. And I think one of the case studies we're seeing right now for that is in the use of AI in summarizing patient visits. Having worked with a bunch of clinicians and having grown up around clinicians, I know how much time they spend just reviewing notes from individual patient visits. And one thing we're seeing that's sort of behind the scenes more is a recording of a patient visit that is then input to AI and sort of helps the clinician confirm a diagnosis or confirm the best course of care and so on. And what I'll be interested in seeing is not just the ones that are blatantly obvious. You're on a chatbot, on a customer support section of a website, or, you know, the Penn Medicine homepage. But should your doctor let you know at the beginning of your annual checkup that this visit's being recorded and the notes are being driven by AI? When you get an imaging scan read with the assistance of AI, should. Should you be alerted to that, or does a patient assume that human eyes are looking at all of their scans exclusively? And where we draw that line is what I'm most interested in. Obviously you have the big picture concerns of what happens if all my Data is put into this computer and disseminated to a global computer system computing system. But those one on one engagements, the small bits of communication in everyday life between clinician and patient, or research participant, or podcast host and audience, that makes you wonder just how do we go about that disclosure? So in the interest of full disclosure, when it comes to AI, what role does it play in the production of the Age of Aging podcast? Do we have any AI tools at our disposal or in our workflow here? [00:38:50] Speaker C: Yeah, I mean, it's pretty crazy the amount of AI tools that are out there now. I use a program called Descript to edit the podcast. It's a really cool software in that it actually creates a transcript of the interviews so that you can edit the audio itself, but you can also edit the words directly and that will change the actual audio of the podcast. And there's a few features on the script that incorporate AI and one of them is the regenerate tool. So if somebody messes up a word, you can highlight it and click regenerate and it'll use AI and your voice to redo the word. Which is kind of crazy and makes me feel a little weird, but also is very helpful when the audio is kind of bad. So you know that there's that. That juggle between the. The good and the bad with AI. But regenerate tool, there's a few other things. I mean, it can literally make an AI version of your voice. I've accidentally done that before and it's very freaky and it makes you sound very monotone and kind of sad. But it does sound like my voice. I had a lot of fun playing that for my friends after that happened. Cause it was just so ridiculous. Yeah, those are the main tools that I use with AI, but I know that there's actually a ton in here that I could be using more for the podcast. And I'm sure will very much influence video and audio production going forward into the future. [00:40:27] Speaker B: But for now, we want the Age of aging to maintain its human touch. [00:40:31] Speaker C: Exactly. That's what we're always striving for. So as little as we can. [00:40:37] Speaker B: All right, well, I think that's all the time that we have for today, but thank you, Jake, for your extensive reporting on this episode. And it sounds like we're going to be maybe revisiting this topic in the future. [00:40:49] Speaker C: Yes, definitely. I think this will be one that we come back to again and again. But yeah, if people have ideas for future episodes or have specific projects or technologies they want to learn more about or know more about in this space. Yeah, definitely reach out. You know, we're always looking for for more story ideas for, you know, future episodes of the Age of Aging. [00:41:12] Speaker B: Well, looking forward to that next conversation then. Thanks, Jake. [00:41:16] Speaker C: Thanks Terence. [00:41:21] Speaker B: Thanks for listening to this episode of the Age of Aging. This show is made possible by generous support from the Michael Nadoth Communications Hub Fund and our sponsor, the TIAA Institute Institute. The Age of Aging is produced by the Penn Memory center in partnership with the Institute on Aging and the Penn FTD Center. Our team includes Dalia El Said, Jake Johnson, Jason Karlewish, and myself, Terrence Casey. Contributors include Nicolette Calcarecchia, Emily Largen, Allison Lynn, and Megan Sharp. [00:41:52] Speaker C: Special thanks this episode to Dr. Alex Mihailides, Dr. George Demiris, Dr. Bhavak Parviz, Dr. Mailee Chang, and Eunji oh. If you enjoyed this episode, please consider subscribing to the podcast, leaving a review or giving us a like. These types of things really help others find the show. And if you know someone who might be interested in these conversations, share this episode with them. We also love hearing from our listeners. If you'd like to reach out, our contact information is in the show. [00:42:22] Speaker B: Notes. [00:42:24] Speaker D: SA.