Transcript: AI and Older Adults

This is the transcript for the Rethinking Aging Club Podcast episode recorded September 24, 2026.

Episode Title: AI and Older Adults
Recorded September 24, 2026 / Published
Guests:
Kyle Shannon, Co-Founder, AI Salon; CEO and Co-Founder, Storyvine
Benjamin Surmi, Social Gerontologist; VP of Senior Living, Arvo
Co-hosts:
Linda Sherman, Managing Editor Boomer Tech Talk, Rethinking Aging Club podcasts founder and executive producer
Ray Gordon, architect, urban planner, futurist, former Pratt Institute grad school professor and avid science reader.
Debbie Howard, caregiver advocate, co-founder, Tokyo-based AgeTech platform Living Best.
AI and Older Adults show notes

In This Conversation

Introduction: AI and Older Adults

Linda Sherman 00:05

Welcome to the Rethinking Aging Club show. I am Linda Sherman. Each week, we host presentations and discussion with special guest experts on an important topic in the field of rethinking aging. We’ll be introducing this week’s co-hosts and guests during the show. We’re happy you’ve joined us.

Today is September 24, 2026, and our Rethinking Aging Club podcast theme today is AI and Older Adults.

I want to start by explaining what we mean by older adults because it’s a very broad term. Depending on who’s using it, it can mean people over 50, 60, or 65, and it can include people who are active, working, and completely independent, as well as people who are receiving care. For today’s conversation, we’re looking at that whole spectrum. We’re interested in how AI can help us directly as we age in our work, our creativity, our productivity, our ability to learn and simply in making our lives easier.

I am particularly interested in whether AI can actually be a leveler for older adults in the workplace. We’ve spent years hearing about the advantage of digital natives, but being quick on Instagram doesn’t replace decades of experience, judgment, and wisdom. Learning to use AI effectively may give experienced people a whole new set of tools to put that knowledge to work.

And then there is another very important side of this conversation: caregiving. How is AI being used to support people receiving care, their families and friends, and the professionals who care for them, and increasingly AI is being built into AgeTech products and services in ways we may not even see.

We’re going to talk about the opportunities, but we’re not going to pretend AI is perfect. It makes mistakes. It can produce mediocre work. It needs human judgment, and there are legitimate questions about privacy, trust, and how these technologies should be used.

So, rather than approaching AI with either fear or hype, today we want to explore a more useful question: How can we use these extraordinary new tools well to help us live, work, and care better as we age? And I have two terrific people here to explore that with us today.

Before I introduce Kyle Shannon and Benjamin Surmi, I’m going to introduce my co-hosts. My co-hosts are Debbie Howard and Ray Gordon.

Debbie Howard is a marketing and communications strategist with over 40 years of experience, including decades of work in Japan. She co-founded the Tokyo-based AgeTech platform Living Best, which uses consumer insights to guide innovation and go-to-market execution. She is also a caregiver-friendly workplace strategist and author of two books on caregiving. Through both Living Best and her company Aging Matters International, she participates in the AgeTech Collaborative from AARP.

Ray Gordon is an architect, urban planner, futurist, and prior graduate school professor at both Pratt Institute and NYU. Ray brings a unique perspective to scientific aging-related topics, and Ray is also a co-host and reporter on our weekly news for the ages by Rethinking Aging Club, as is Debbie.

Ray Gordon 04:06

And I’d like to introduce Linda Sherman. She’s the founder and executive producer of the News for the Ages podcast and the Rethinking Aging Club podcast, which is a participant in the AgeTech Collaborative from AARP. She is managing editor of Boomer Tech Talk, home to the club’s podcasts. Linda previously served as CEO and CMO for multinational subsidiaries in Japan and now serves as chief longevity officer for LA-based Miracle of the Sea. She helps people understand the best research products and conversations in aging, longevity, caregiving, and age tech. Linda is also a role model for healthy longevity, staying physically fit while embracing aging with curiosity, purpose, and community. Take it away, Linda.

Linda Sherman 04:52

Thank you very much, Ray. So I’m going to introduce Kyle and Benjamin, and I’m going to ask them each to speak for two minutes, just telling us about what they’re doing in the field of AI, and then we’re going to have our discussion, and we’ll circle back to learn more about what they’re doing during that discussion as well.

I met Kyle Shannon at Mike Stelzner’s Social Media Marketing World and AI Business World this past April and joined Kyle’s AI Salon in May. Kyle Shannon is a digital pioneer, entrepreneur, storyteller, and AI community builder with more than 30 years at the intersection of technology and creativity. He co-founded Agency.com, one of the early giants of digital advertising, and is now CEO and co-founder of Storyvine, an automated video storytelling platform. Kyle also co-founded AI Salon and created the AI Learning Lab, helping everyday people explore generative AI through curiosity, experimentation, and play. His latest project, *The Great Repurpose*, explores how people can rethink work, creativity, and human potential in the age of AI. A longtime tech optimist, Kyle focuses on making emerging technology approachable, practical, and empowering for creators, businesses, and communities navigating rapid change.

Benjamin Surmi is a social gerontologist and vice president of senior living at an innovative AI-supported training company called Arvo, which is at Arvo dot care. We are delighted to feature Benjamin Weekly on our News for the Ages by Rethinking Aging Club, where he is a regular reporter with exceptional perspective on both senior living and the products that serve them. Benjamin has spent nearly two decades exploring how we can better support older adults and the people who care for them. His career has taken him from frontline caregiving and community operations to national leadership and around the world in search of promising innovations in aging and care, including helping bring the internationally recognized Humanitude approach to the United States. Today his focus is on what comes next, reimagining how we upskill the senior living workforce, moving beyond traditional training toward practical, in-the-moment learning that equips people to do extraordinary work.

So, I want to ask Kyle first to give us his two minutes and then Benjamin. Thank you very much.

Kyle Shannon: Discovering AI and Its Creative Potential

Kyle Shannon 07:53

Beautiful. Thank you for having me, everybody. I’m Kyle Shannon, and thank you, Linda, Debbie, and Ray, for having me.

As Linda said, I’m a storyteller by training. I have a degree in the arts and pursued life as an actor and a writer in New York City for a bunch of years, and then discovered the World Wide Web and ended up starting one of the first online magazines, an art culture magazine called Urban Desires, and co-founded Agency. com, and really got to be there at the beginning of a technology: the World Wide Web. In this case, that really ended up changing everything. And I remember feeling, you know, how lucky I was to be part of something that I knew was going to change the world and had an experience that did that. And I didn’t think that I’d see that again in my lifetime.

And in the summer of 2022, I was playing around with some image generation tools. Stable Diffusion was the name of one of them, and I thought, oh, this AI stuff’s pretty interesting. I’d been following AI for years, but you know, it was pretty clear you needed to be a mathematician to be able to be in the game, you know, or a researcher, you know, someone deep in the science side of it. And so I was paying pretty close attention to AI because, you know, the image generation stuff in 2022 was really interesting.

And then November 30th, 2022 happens. ChatGPT comes out, and I start interacting with this thing and realize pretty quickly that the Turing test has just been passed, which is the Alan Turing in the 50s said: you know, at the point at which you interact with the machine, and if you can’t tell if there’s a person or the machine on the other side, you know we’ve hit artificial intelligence, and so it seemed pretty clear that was there for me.

And I started playing around with asking it to write me some Python code, and before I knew it, within 90 minutes, I had written a fully functional Python application that was making API calls, and I’ve never written a line of Python in my life, and I just had this massive epiphany that the world has just changed, but no one knows it yet. And so I immediately started the AI salon and the AI learning lab and thought, I am going to, you know, commit whatever free time I have to understanding and learning everything I can about this remarkable tool that can be a tool for self-expression, and it really is that and then some.

Linda Sherman 10:07

Thank you so much. And there’s obviously so much more to say there. We’ll circle back very soon. I want to let Benjamin just give us some better background on what he’s doing with AI as well. Benjamin.

Benjamin Surmi: The Future of Caregiver Training

Benjamin Surmi 10:21

So cool to meet you, Kyle. Thank you, Linda. I love your focus, Kyle, on storytelling. That is massively a huge part of what I really want to bring to senior living and training, because so much of it really needs better storytelling for us to actually make a difference in this field.

Yeah, I’ve been working in the senior living field for almost my whole career. Started when I was maybe 15 years old, and I’m now that many decades old. And it has been a huge part of my life is working with seniors in all of these assisted living, memory care, skilled nursing settings, home care. And as I went through my career and became leaders of different settings, one thing became very clear to me, and that was all the best research we have, all the great books that have been written, speakers that have spoken on quality care for elders, everything they say depends on supporting the frontline worker in actually enjoying their work and having the skills they need to do their work. If they don’t have the resources and tools to do it, you can say everything you want about what should be, could be. You can regulate the system to death. It’s not going to make a difference unless you can literally give the skills to the people doing the work.

And so my career took a turn recently, and joined a startup that’s mission is exactly that. We are working to rethink how we upskill frontline workers from diverse backgrounds, diverse language backgrounds, diverse educational backgrounds, in diverse settings, how do we upskill them? How do we scale that upskilling across the world, starting here in the United States, so that you’re not dependent on gurus who can do some in-person training and maybe only help so many people at a time? How can we actually scale the skill-building needed right now when the United States needs excellent workers, people who know what they’re doing? So that’s what I’m working on, and obviously AI is going to play a big part of that.

Linda Sherman 12:32

Great, thank you so much, Benjamin. I want to go back to Kyle to let him tell us specifically about the Great Repurpose, which is his latest project, Kyle.

The Great Repurpose: Rethinking Work and Human Value

Kyle Shannon 12:46

Absolutely. So, if you go to thegreatrepurpose.com, you can read about what we’re up to there. The Great Repurpose really came out of for the past 18 months or so. I had been just thinking things are going to get weird. Things are going to get weird, and I didn’t quite know what that looked like, and I knew it had to do with AI adoption. And in December of 2025, a new model came out from Anthropic called Claude Opus 4.6, which all of a sudden was good enough that engineers and programmers kind of shifted from a position of like, well, you know, AI is okay, but it still requires a human in the loop. All of a sudden, the narrative started being more like, “Ha, I think this thing can actually do my job now.”

And pretty much two months after that, I started seeing posts on X where programmers were talking about, “I hate my job now because I love programming, and I no longer program.” And you know, I’m babysitting AI agents that are doing the programming. And the thing that I was good at, and the thing that I’ve spent, you know, decades perfecting, is no longer necessary. It’s no longer valuable.

And you know, a lot of people had been talking about AI is going to come and take the jobs, and there’s going to be job loss. But what struck me in those comments about the engineers is, oh, even people that keep their jobs, their job is going to fundamentally transform. And you know, this is happening right now with programmers, just because that’s the nature of where the Frontier Labs focused. But everyone’s job, everyone’s job, is about to transform massively.

And what also struck me is, you know, we live in a country where we highly identify our self-worth with our job, with the tasks that we do. Right? It’s a holdover from the Industrial Revolution, where time equals money, and if we meet someone, Linda, if I meet you at a party, you know, we often say, “What do you do? Right? We ask, “What do you do?” We ask, “What are the tasks that you do that you exchange for money?” And those tasks are about to be democratized and devalued.

And so, what struck me and the whole idea behind the Great Repurpose is your value has always been more than your task. But we live in a world where we talk about the tasks, and so really looking at what is your value beyond the tasks, and then from there saying, okay, if that’s my value, then you know what’s my purpose? What do I want to do in life? Who do I care about? And then how can I then use AI, the same tool that maybe is, you know, redefining all of what we do? How can I use that same tool to redefine and repurpose who I am and what I do in the world?

So basically, all starts with putting yourself at the center, understanding your value independent of your outputs, and then start using that as a foundation to say, now I’ve got this remarkable capability of AI. How can I use that to amplify who I am rather than try to compete with this thing that is increasingly powerful?

Linda Sherman 15:50

Thank you, Kyle. And I know that one of the first things that, when we’re effectively using AI, that we want to do is to make sure that the AI that we’re working with as our assistant it understands exactly who we are, and they can help us figure that out with questions and answers and so forth. But that also allows all of the LLMs and bots coming through to be able to capture who we are and to develop our expertise so that it’s seen online, so that we’re seen for what we want to be seen for online, right?

Kyle Shannon 16:34

Right. Really being clear on who you are, what you value, what you want, and, you know, in the case that you’re describing, you know, creating foundational files for agentic systems, like some of the new stuff that’s coming out, like OpenAI’s codecs and ChatGPT work. You know, being able to concisely tell those systems who you are and what you value is a big part of how you make them powerful for yourself.

But even just using ChatGPT, rather than thinking of ChatGPT as something that’s going to give you answers, where you know you just sort of treat it more like a vending machine, shifting that mindset to say, “Wait a minute, I’m in the driver’s seat here, and I’m going to use AI to amplify me. So being clear on who you are and what you want, what your ideas are, and then use AI to amplify those ideas.

That’s a big part of what we do in the AI salon, where we talk about this cycle of AI readiness: play first, create excellence, generously lead. But you’re always at the center of it, driving the system based on the things you want to accomplish, as opposed to trying to compete with this machine that is going to go off and do lots of other things.

Linda Sherman 17:46

Excellent, Benjamin. Would you like to comment on that aspect as well?

AI, Creativity and the Advantages of Life Experience

Benjamin Surmi 17:52

Yeah, I think, Kyle, I love your perspective on the idea of really getting clear on who we are and what we’re trying to accomplish so that these tools can amplify and even open doors for things we are really important to us, but we were looking at the toolset before us. And none of those tools would help us accomplish the things that really were on our hearts to do. Right.

So I think I hear you also saying, right, there’s this whole new toolset that’s open to you. I think, for instance, you have storytellers who may not ever have the budget to create a movie. They never will be able to set up a podcast. They will never be able to learn art or graphic design as well as someone who’s been doing it for years. And now we’re at a time where these people who have got stories in them, right, can literally write a podcast, and 11 Labs will generate it for them with music and sound effects. They can write, and a video is created, and they can create their own mini-series, right?

Like we’re at a time where those who are creative but could never afford the tools now all of a sudden get a chance to create in whole new ways. Same with people who maybe never would have the infrastructure to do deep data analysis in their business or in their personal life or in their research, and now all of a sudden you’re putting supercomputer at their fingertips, or let’s say they’ve always wanted to study a certain aspect of medical care or healing arts or something related to building ecological homes. All of a sudden, a person where that’s their passion, that’s their vision, just has a whole new tool. I think that’s part of what you’re getting at. Am I right?

Kyle Shannon 19:39

Yeah, it’s very much what it’s about. And you know, it’s funny. There’s a term that I don’t know if I came up with it. I stumbled upon it, where, you know, in life you get good at certain things, right? You have a passion for it, whatever it is, and there’s things you’re really good at, and then there’s other things that you’re not good at, and you know you’re not good at them, and maybe some of those you’re not good at because you’re actually not good at them and don’t enjoy them, but maybe some are things that you really wanted to do, but maybe Step Daddy told you you weren’t talented, and you believed him. And you know, you went through your life just always sort of wanting to do that. I realized that there are these skills that we don’t have that we would really love to be good at, right? And I call them “aching gaps.” We’ve got gaps, and then we’ve got these aching gaps—the things that you really wish you were good at. And what AI does is it all of a sudden puts in immediate reach the ability to do those things that you never thought possible.

And one of the things that I’m seeing, you know, this is a conversation about aging. One of the things that I’m seeing, I hear on the news that, “Oh, young people are good at AI and older people are not good at AI.” That’s not been my experience at all. Like within the AI salon, there’s a lot of Gen Xers, there’s a lot of Boomers that are absolutely killing it with AI because they’ve got such life experience, and they can look at the output of AI and say, “Oh, that’s good. That’s not good,” and they’re discovering how to fill these aching gaps.

And you know, one of the things I see consistently is people who’ve had stories they’ve wanted to tell all their lives. Going back to storytelling, Benjamin, is they’ve had these stories they’ve wanted to tell all their lives, and all of a sudden they now have the capacity to do that. And the work that they’re producing is really quite remarkable. And so, you know, I’m witnessing people coming out of retirement and changing careers and being completely inspired by this, even though the narrative out in the world right now is the robots are going to kill us. I’m sort of seeing the opposite side of that, where these intelligent machines are allowing us to self-express ourselves in ways that were never possible.

Linda Sherman 21:37

Excellent. Ray. Go ahead.

Human Judgment, Taste and Trust in an AI World

Ray Gordon 21:38

Just sort of interesting. I always had this philosophy that the value in your professional life ultimately is what you know and not what you do. And I always felt that as you progressed professionally, you would figure out, you know, what you are an expert at, and you would have other people that you might be able to offload the actual doing to so that those people, for instance, with an architect, you could be the creative force, the designer, and have someone draft them and draft up your ideas. And I thought that that was really how you differentiated somebody who had experience and knowledge in that field versus someone who’s starting out and still learning.

So I was wondering if that’s true any longer; I think it is. I think that the goal would be to have that value be marrying your experience to be able to analyze what your AI output is doing, whether it’s good or bad, or how you can improve it.

Kyle Shannon 22:34

Well, so I’ve got a really specific opinion on this. I actually think that you can choose to work with people that have expertise, but I think expertise is the thing that’s being democratized. And the AI systems are getting better and better and better and better.

So, one of my favorite human beings on the planet is a guy named Rick Rubin. And he’s a music producer, and he produced a lot of the early hip hop and the Red Hot Chili Peppers and Adele, and he’s won lots of Grammys. And the interesting thing about Rick Rubin is he doesn’t know how to do anything musically. He doesn’t know how to play an instrument. He doesn’t write songs or lyrics. He doesn’t know how to work the recording console. He’s hired because he’s got taste, and you know his job is to basically have taste.

And so, where I think this is ultimately headed is that it is going to be our job to have an idea and then allow the tools to do the actual execution of it, where single-solo operators can absolutely pull off things that, you know, historically would have taken teams of, you know, 10s or dozens or hundreds. They’re just going to be able to do it. And so our job in the future is really just going to have ideas and maintain the fidelity of the idea using AI. So I think we are in the middle of a transition right now from valuing specialists at the top of the value pyramid to putting generalists at the top of the value pyramid because the doing is going to be democratized.

Benjamin Surmi 24:04

That’s a really good way of looking at it. I think there’s another layer, though, that is the people skills, because I think yeah, you’ll have solo folks doing cool stuff, but it’ll be interesting to see how many solo folks actually, you know, really make massive businesses, right? Really make massive change. It’ll be interesting just to see how that works out.

Because what I’m still seeing is those who are doing just incredible stuff with AI. Like it’s people who know how to work together, how to build trust, how to build relationships, how to enter into someone else’s world and understand their problems and solve their problems, and how to build that relationship of trust, where they’re willing to hand you a check for $2 million to support their business goals. I think that’s the other thing that’s not going away anytime soon, and that is learning how to network, learning how to have solid relationships with other human beings. Like, those are the people who are going to ultimately be really successful.

You’ll have people who will trend on Instagram for a little bit here or there. You’ll have people who will do cool stuff, and it’ll be a flash in the pan. And then a big corporation will take it, buy it, spend them a lot of money. But it’s really all about relationships, as far as scaling things, getting lots of people to use it, letting lots of people to really benefit from it. From what I’m seeing so far.

Kyle Shannon 25:16

I think that’s right. The foundation of the AI Salon is about that. It’s about being in a community of people who you know and trust. And I think that we’re exiting an era where I would maybe work with someone because they had skills, right? And a lot of times, how many times you’ve been in a situation where you work with someone and you’re like, they’re just an absolute jerk, but like they’re really good at whatever Photoshop or that thing, and we would hire them for their skills?

And I think we’re absolutely moving into an era where we’re going to hire people. We’re going to work with people because we trust them. I know that you know Jane is going to show up, and she always cares about the job and making sure they get it right. And Sally’s going to show up because you know she wants to make sure that everyone’s having a good time. And why we interact with people is that trust that you’re talking about.

Yeah, I’m not suggesting that we’re moving to full solo entrepreneurship. Although, you know, there’s a bet in Silicon Valley of who’s going to be the first single-person billion-dollar company, and the New York Times wrote about that person about six months ago. So that’s already happened. So that’s a possibility. But I think you’re right. I think why we work with people is going to be about trust and relationships primarily.

Linda Sherman 26:24

And I like the fact that you said also a person who has taste, because when Ray was talking about architecture, maybe an AI can design a building, but there is taste to make judgment about how to tweak that and how to give direction to what you want, right, Ray?

Ray Gordon 26:48

Well, yeah. I mean, I think that Kyle’s position about or thought about having a generalist at the top of the pyramid is actually very valid. I think that as you get more and more experience, certainly in creative fields, you have a broader view of all the possibilities, so you have an expanded list of things and options that you might want to consider. And those kinds of things can be used to apply to an AI or to people who are helping you, who are technicians at doing a very good job at a specific skill.

So your job is to say, okay. I think this is the overarching idea. These are the principles I want to approach the project with and apply those not only to just the roof but also the walls, the windows, the finishes, and have that somehow be cohesive. And I think communicating that, or being able to communicate that to those—whether it’s a person or an AI actually doing the work—is the perfect marrying of those two kinds of skill sets. That you know you’re going to have people or expertise or AI that’s an expert at doing something and have them take off on something, but make sure that you can give them some guidance so that they, you know, give you a product that you are feeling is in the sweet spot of what your original vision was.

Kyle Shannon 28:04

I think that’s right, Ray. I think it ultimately becomes if you think about, you know, how does something rise above the noise, right? When you start playing with AI, one of the remarkable things that you discover is, “Oh, this thing I wasn’t good at, all of a sudden I can be good at,” and you make a song, or you make a picture or whatever it is, and you share it with the world, and no one cares. And you’re like, “Well, I’ll make another song,” and you make 10 songs, and then no one cares, and no one cares. And why is that? Well, everyone is now making content like that, right?

And now that we’re moving into an agentic era, where I can just give an AI system a goal, and that goal might be go start me a TikTok channel and create all the content and, you know, just blast all the content out there. We’re about to enter a world where the amount of content produced, you know, significantly outpaces anyone’s ability to take it in.

And so what’s going to rise above the noise if all of the AI systems are all learning from one another and they’re going to be copying one another? What’s going to rise above the noise? And I think what’s going to rise above the noise of “Oh, Linda did that thing, and she told that story about, you know, her cousin, and that turned into that thing and her idea her specific idea that put just enough twist on, you know, the same capability to make it stand out in a unique and interesting way.” I think that people that have that skill, the ability to take the same technology that you know everyone else is using and now they’re using at scale, and be able to twist it just enough to have something rise above the noise. That’s what I think the new skill is going to be.

Debbie Howard 29:41

I think that’s really, really interesting, Kyle and Ray and Benjamin. All of these trends converging together. I also think that the thing we’re talking about right this minute speaks to what Linda was talking about early on about AI being possibly a leveler in the workforce for older workers.

I know that I am biased, being 72. But the way I look at what, let’s say, ChatGPT or any of the different AI modes pumps out, the way I look at that is really, really different from my team members who are 20 years younger. I have a different perspective, and I obviously think it’s better, you know, and more experienced. You know, I’ve got a lot that I’m bringing in when I look at something from AI.

And I actually also noticed that in the last couple of years, my own assignments as a writer and a commentator have moved more towards, you know, how do you use and expand using the AI like we’ve been talking about, but also rehumanizing the AI and targeting it for consumption because there’s a lot of garbage out there now. To your point, Kyle, with everybody pumping this stuff out, and I think one of the things that we can bring to the party, as let’s say experienced practitioners, is that perspective that maybe makes it a little more useful and hopefully makes it stand out a little more.

Linda Sherman 31:20

Thank you, Debbie.

Benjamin, I wanted you to have a chance to talk more about what you’re doing at Arvo, which is very AI-centric and using it at a lot of levels. Can you describe that? And you had told me the other day about what you’re doing with 11 Labs, which is fascinating.

Arvo: AI-Powered Training for Senior Living Caregivers

Benjamin Surmi 31:42

So, imagine you have to go work at a restaurant or hospital or factory. Most states have decided you really should have some training before you do that. And so, across the United States, there’s hundreds and hundreds of companies. It’s a, you know, multi-billion-dollar market for learning and development, and companies for the past 40 years have been making a lot of money off selling really cheesy videos, PowerPoint slides with voiceovers, with little quizzes, maybe little cheesy games that you play along the way, and nobody learns anything. Most of the time, people are on their phones watching TikTok when they’re going through their training, and it created a sense of compliance training is not really important.

And yet, having worked in a very risky field, caring for seniors and their most vulnerable. I see how absolutely vital it is that a lot of what’s covered in those trainings becomes internalized and acted upon. I mean, we’re talking about knowing what to do in a fire, in an earthquake, when someone gets poked with a needle. We’re talking about being able to help someone who’s refused showers for two weeks, and you know they really, really need a shower, right? We’re talking about life-and-death situations, and yet because we have scaled training so often with online training that is not effective, does not capture attention, does not actually give anybody any sort of upskilling, it’s literally just checking the box on mandated training. People look at that training and look at that knowledge as unimportant and then end up making poor decisions, doing things that affect their lives. They go to jail. They have a criminal trial because they did something that, if they had just known, if they had just practiced, they might have been different.

But we also have group training and in-person training happening all over the world, right? And you could argue that’s a much better training, and I would agree. However, even with in-person training, you often are in a room with 100 people. You’re taught a skill. Maybe there’s a role-play, like a really good training. There’s some role play. There’s group discussion, etc. Right? Most people are just talking in the back when the role play is happening. They’re not actually role-playing. There’s no one there to coach them on whether they did their role play right or not. Most group trainings in the United States, like the best ones we’ve been to, we just had a good time. We laughed. We heard great stories. We cried. We were inspired. Did we actually change our behavior when we went back to work the next day? I would say 90% of the time, no. We maybe told people about it. It was a really great one, and we shared about it, but we really didn’t change our behavior.

And so, part of our thesis is that one-on-one coaching of team members in any kind of workplace is essential. And whether you’re delivering some content in person or delivering content online, it must be paired with one-on-one coaching, where people are practicing their skills and getting immediate feedback on how well they did. And what Arvo is doing is focusing on one specific industry: the senior living industry. And focusing not on building just a generic product that, like, hey, any company could, let’s say, implement this one-on-one mentorship, scalable AI-driven coaching, but we can actually track it back to every state regulation, so that when this employee is doing that coaching and that mentoring process online with AI avatars and mentors who are walking them through building skills, they’re getting the credit for having done that with their state.

You know, California requires 12 hours of dementia care training. Most people just sit people in front of a computer for 12 hours. They watch mindlessly a bunch of videos, and then they are told to go to work. We believe they should be practicing how to talk to a person living with a brain change, how to negotiate a shower, how to handle a situation where someone is asking to leave in the middle of the night, and you’ve only got one other caregiver on duty with you, and how do you manage that situation? They deserve the opportunity to practice that privately, confidentially, and be able to get the coaching they need to keep practicing until they get it right. Like that’s what the future of learning needs to be. People need to get credit for doing it, and they need to be able to practice until they can do it really well.

So we’re using AI tools to create human-like experiences where people can practice their skills and be able to get the personalized coaching they need to keep up leveling themselves and get credit while doing so.

Linda Sherman 36:27

Fascinating. That sounds really cool. So I would interact in a situation where somebody wants to leave in the middle of the night and see if I can be successful or not with interaction going back and forth. That’s very real on the other side with a lot of resistance, right?

Benjamin Surmi 36:46

Exactly. That’s the goal. Like right now, we are still—this is such a new field, right? How do you pull this off? There’s been some companies that have pulled off, you know, kind of a standalone tool. You don’t get any credit for doing it, but you can practice in other fields like HR—like how do you fire someone correctly, right? And typically, you’ll have the employee sitting in front of you virtually, and you’ll interact with that employee, and you’re trying to fire them without them getting angry with you, and then you get like a written performance evaluation on how well you did.

We’re focused on that frontline workforce, right? English language learners, people who speak multiple languages, people who may or may not speak English very well. We’re working with people who maybe are illiterate or read very little. People who may only have an eighth-grade education, if that. You know, we’re working with a diverse workforce, and so we can’t give them a page-long LLM-created profile of all the things they could try better. We actually have to create an experience where they’re coached verbally in person by someone that they can trust and feels and looks a lot like them, right?

And so, at the moment, the avatar experiences are with a peer or a supervisor. In other words, let’s say another caregiver or a nurse that you would report to, right? And they talk like that particular persona, and so this brings up the storytelling that Kyle was talking about, creating these characters that they get to interact with, and then that character is kind of walking them through a case study or saying, “Okay, what would you say in this situation?” And then, “Okay, well, here’s what’s happening: Mary’s resisting you.” Right? Because what we haven’t innovated yet, but our plan is how, if they’re interacting with an older person and they’re in this kind of dramatic experience, how do they get the feedback afterwards. Right, because obviously Mary can’t break character and be like, “Okay, so here’s what worked really well. Here’s the three things you should try next time. Want to go again?” Right, we can’t quite do that.

So to answer your question, our future vision is to get to that place where they can actually interact with someone who feels like an older person living with a brain change, or an older person who’s irate because their bill was incorrect, or whatever the situation is, and be able to then have a coach step in, be like, “Okay, here’s what happened. This is what I see happen in this situation. Why don’t you try this again? Try going back in.” That’s kind of the direction we’re headed.

Linda Sherman 38:58

Awesome. That is so awesome. I’m really happy to learn more about what’s actually happening at Arvo. Thank you so much, Benjamin. Can we talk a bit about the production of products thanks to AI? Even in the AI Salon, we’re seeing people play with creating tools that are specific to helping with Parkinson’s and so forth. Benjamin, can you help us a little bit with that on generally what’s happening with AI and tools to help older adults with caregiving and their families?

AI Tools for Caregiving and Senior Living

Benjamin Surmi 39:35

Yeah, I mean, I think you can divide them into a couple categories. One category is tools that families go to for help in figuring out what to do. You have tools that help a family monitor or track the well-being of their loved one. And then you have tools that the older person uses themselves to make life better in some way. And then you have tools that senior living communities, hospitals, skilled nursing facilities use to protect their residents, know their residents’ whereabouts, understand the care they’re giving their residents, track what must be done for residents and whether it got done or not, right? You have kind of those different categories.

So, like, for instance, in the senior living space, right, you’re seeing EHR companies, companies who manage all the data about what medicine you’re on. Did you fall today? What kind of care are you supposed to get at 4 o’clock p.m.? You know, those companies are starting to really do creative things. Like August Health is one of the premier players here, and you know they’ve just launched intelligent service planning, where in the past nurses have to write out a full service plan. Well, now you have the potential where the AI tools in the background are reading what the nurse has assessed. The nurse has said this person has this issue, this issue. They have trouble with this. They have trouble with that. Here’s what their concerns are, and the AI does a first pass on suggesting, “Here’s what we think might be the best plan.” So then that nurse can quickly review, add, delete, etc.

They’ve also launched a tool where you are getting caregivers. They come to work. They’re brand new. There’s 100 residents to learn. How do you do that? You can’t read through every page in the EHR. You can’t scan through dozens and dozens of pages for 100 people and learn anything. So they’re creating these small, tiny, bite-sized pieces of information that a new caregiver could quickly get up to speed on who this person is they’re caring for. Right? Those are examples of AI at that space, or Teton AI with their sensors that they put throughout senior living. And those sensors are using all kinds of data analysis and AI predictions to tell you, “Hey, this person’s at risk of falling out a window. This person’s at risk of hurting another resident. This person’s at risk of falling in the next two minutes.” Right, like that’s the kind of things that senior living is looking for: is how can we prevent people from getting hurt in these settings where you have so many vulnerable people being taken care of? But then if you look at, like, the other spectrum, family members needing information, right?

There’s several new apps where they’re trained on best dementia care practices, and so a family member can pull up that app and turn it on and be like, it’s basically ChatGPT, but it’s trained by experts, right? And so you can ask, “Hey, I’m in the shower. Mom doesn’t want to shower. What do I do? Hey, oh my gosh, I can’t get any sleep. Mom’s pacing all night. What do I do?” Right. And there’s tools that are being put out by experts right now that focus on like how to connect with resources in your local community and how to get the information you need in a timely way. So those are examples on two different ends of the spectrum.

Ray Gordon 42:44

Kyle, I have a sort of a question here. What are the processes for training the AI that are training? How would that work to refine the AI so it can be better off at doing the kinds of things that Benjamin is discussing?

Personalizing AI: Context, Prompting and Voice

Kyle Shannon 42:58

Sure. I mean, there’s lots of ways to think about training. I feel like the term training gets tossed around by researchers in these labs as this ivory tower kind of thing. Training can be as simple as providing the AI that you’re chatting with more context, right? So if you say, “Tell me a story,” ChatGPT is going to tell you a story. But if you turn on the microphone and say, “Hey, you know, I grew up in the 1930s, and my dad worked for the railroad, and I know he worked in the line from Texas to Chicago, and you give AI all this context, and then you say, “Now tell me a story,” it will tell you, right? “Tell me a story about that time.” It will tell you a story that’s much more relevant to you. So, at its most simple, training these AI are giving them specific data that are relevant to you.

When you’re talking about agentic systems, there are a couple of very specific files. There’s a file called user.md, and the user is you. And so what goes in your user.md file is like who am I? You know, I’m an artist innovator and I’m a storyteller and I started these companies and I’m passionate about these things and I care about these people. So you put into this document a little bit about who you are, and then there’s one called agent.md and that’s the AI. And you basically tell it what its job is. Your job is to support Kyle, and you know, collaborate with him, and help refine ideas. And then there’s a memory file, memory dot idea, that as you work with this agent, it remembers what you talk about.

So just understanding that there are systems now where you can provide them very simple. These are just text files. There’s nothing fancy about them. You can actually have ChatGPT if you’ve been interacting with it for a long time. It knows a lot about you. If you’ve been talking with ChatGPT for, you know, more than a couple of months, it knows everything you’ve talked about. So you can literally say to it. Based on what you know about me, create a user.md file and create an agent.md file so I can use it in one of these systems. And then you just give those text files to that system and say, you know, use these files to get better at doing things for me, and it will, and it will know who you are. And when you, you know, every time you start it up, it looks at those files and it knows who it’s talking to, and it understands what you’ve been working on, and it really is quite remarkable.

So the thing about AI that I think does require a little leap of faith of understanding is that if you don’t know anything about it, your assumption is very often, ” Oh, I need to be a mathematician. I need to be an expert at computers. I need to be super technical.” No, you actually don’t. You just need to understand what you want and be able to ask for it and kind of understand that there are a couple little tricks along the way that they’re going to get easier and easier and easier over the future. But you really just give it more information about who you are, and it gets really, really good at serving you and amplifying all of that stuff.

Ray Gordon 46:04

So the human skill that is involved here is being able to communicate with the AI specifically in order to get a better response and better output from the AI.

Kyle Shannon 46:14

Yeah, communicating who you are, what you value, and what you want. And I think that similarly, you know, we’re moving into this era that Benjamin was talking earlier about trust and who do you trust and who you are. Getting good at articulating who you are as a human, like what are you good at, what are you passionate about. You know, if you join the project, what are you bringing to the party?

Like in my case, I’m an artist innovator, and I get really excited about stuff, and I tend to inspire people to want to try new things, and it’s like my ability to articulate that, you know, lets someone else go, “Oh, okay, I could use that kind of energy. You’re like, “Ooh, no, I need someone who’s more serious, and I need a planner. And Kyle’s not a planner; he’s chaos,” right?

That thing that is the humanness of who you are, like what are your superpowers? Being able to communicate that in person, but also figuring out how to translate that to be able to tell that to AI, I think becomes kind of the superpower moving forward.

Benjamin Surmi 47:12

Kyle, I’m curious if you’ve had the same experience I’ve had. There’s all these experts who tell you, “Here’s how to prompt,” and there’s all these new tools that are coming out—they’re like, “Hey, we’ll adjust your prompt for you, or we’ll write a prompt for you to make that image, or to make that video, or to make whatever.” Right? Like, we’ll be sure to write the prompt for you. You give us an idea, and we’ll write the prompt.

And what I’ve experienced is I get the absolute best results by just being completely off the cuff. I just talk into my computer and I tell it what I’m looking for, and I give it my context as much as I want in my own words. I always get the best results from that every time I try to do the prompting. Was like, well, make sure you use these keywords, and make sure you prompt in this way, and make sure you give it this scenario. And I never get the results I want. I’m just curious if you’ve had similar experiences, or have you found like really strict prompting rules actually help you?

Kyle Shannon 48:04

1,000% with you. Here’s my overarching rule right now: anyone who tells you they’re an expert at AI, run away from them as quickly as possible. Anyone who says they’ve got the answer on how stuff’s going to work—nobody knows anything because these technologies are literally changing. It was every three months. Now it’s about every three weeks. Like, what works in one, three weeks later it won’t work as well. And so what you just said, Benjamin, just be yourself. Articulate. Just being able to have a point of view as a human and say, “Here’s what I want to do, and I know what good looks like, and I’m going to ask the AI to give me something.”

And if the AI gives you back crap, you know, you mentioned Debbie about being on the planet for a while, and your perspective, your POV, is valuable. It’s really valuable because if you’ve been on the planet for a while, if AI gives you back crap, you can look at it, and you go, “Oh, that’s crap,” and then you figure out, “Well, what do I need to tell it so that it gives me something that’s less crappy from my perspective,” right? And then you say, “Oh, no. When you gave me that, you know, you assumed I meant this. I meant that.” And you tell the AI. You just kind of guide it.

And as you said, Benjamin, in your own words, you know, however you articulate your views of the world, you’re the lens that you see the world through. AI is really good at interpreting that and processing that, and then giving you back some other result. And so, rather than thinking about AI like a vending machine where you sort of put in a prompt and get out an answer and expect that to be brilliant, no, think of it more like a conversation where you are the creative director. You’re the Rick Rubin producer who’s got the idea, and then AI is just your partner, man. You’re just jamming with it back and forth. So absolutely, don’t spend money on prompt libraries. You don’t need to. You don’t need to prompt anymore.

Linda Sherman 50:00
I think that the whole prompting thing has just changed dramatically in the last six months, three months, even to be so much more human-based than when we were being very specific about prompting. And in terms of the UX on ChatGPT, which, in the AI Salon we were watching evolved over a few-week period where it was completely different every day until they got it sorted out, and just today I just put into the show notes something that I did today which is generations at a glance, just to remind us what’s in each generation, like the silent generation versus boomers and so forth.

And I was talking to my ChatGPT, and when I said I want you to make this into a Google Sheet that I can then produce a PDF with, she said we need to go to work, and it did a seconds count, and it’s the first time I’ve seen this. It gave a seconds countdown, and let me click that to flip from ChatGPT on the web, like ChatGPT.com, to the work tool within ChatGPT, because that’s where I needed to be. And I think it’s going to continue to progress in that way. Yes, we do need to know what we want and to express it and so forth, but the user interfaces are going to get better and better and better. What do you think?

Kyle Shannon 51:35

They’re going to get much better, and they’re going to if you haven’t played with the voice mode of ChatGPT. If you launch like the ChatGPT app, there’s a little icon that looks like sound waves. Click on that and just start having a conversation with your ChatGPT. It is remarkably human, and you can literally—like, one of the things I’ll do is I’ll put that mode on in my car, and I’ll have it on Bluetooth or whatever. And on my drive to work. I’ll be brainstorming about things I want to do at work, and by the time I get to work, I’ve got a whole concept articulated and written out that was really just like a jam session, you know, talking with this thing.

So, I think the interfaces will get increasingly simple, and a lot of the fancy technology is just going to drift into the background. If you’re a geek and you want to nerd out? That’ll always be there for you, but I think where this stuff goes is it gets very, very intuitive and just, you know, kind of magical from where we are today.

Linda Sherman 52:28

And the wonderful thing about voice is that it keeps a transcript of everything that got said.

Kyle Shannon 52:33

Yep.

AI Storytelling and the Future of Creative Work

Benjamin Surmi 52:34

Ask you a question, Kyle. So you’re talking about storytelling and how that’s such a huge part of your passion. And I’m curious. Like one of the things that we’re looking for at Arvo is people who can make incredible short-form video series that can be embedded throughout a learning experience or be able to weave in, you know, podcasts or radio shows or TikTok-style videos that help connect with a whole generation of learners. Where would someone like me go to find those storytellers who are really adept at the new tools and who could do that kind of stuff?

Kyle Shannon 53:12

I would say just pay attention to if you’re not on X, pay attention to X. Follow creators there because you definitely want storytellers who understand how to use the tools, but also start paying attention to the tools.

Some of these agentic platforms, like I’m noticing in the past week with the advent of ChatGPT 6 and Claude 5.5—I think it is Opus 5.5. Those models are creating full-on animation stories, videos, just from the models themselves, just out of code.

So start paying attention to what the systems themselves are producing, because you might be able to do hyper-personalized content series for individuals rather than having to think about it like mass media. But I would start following creators on X that are using AI, and then really start following the technology because I think we’re probably within a year of the tech being good enough to produce high-quality personalized content.

Benjamin Surmi 54:10

What terms would you use to find that content on X?

Kyle Shannon 54:14

Just go start finding artists. There’s a woman named Kelly Boesh, B-O-E-S-C-H, who does these really remarkable music videos. Find artists like that, and then start following who they follow. There’s also a guy named Robert Scoble who has lists of AI artists, AI founders, things like that. And go look at his lists. And if you basically follow one of his lists and then go through those and find people you like and follow who they follow. Your algorithm will shape up in about two hours.

Benjamin Surmi 54:47

Love it.

Linda Sherman 54:48
All right, this was wonderful. Thank you so much to Benjamin and Kyle for joining us today, and thank you to Ray and Debbie for being here and to our audience; we appreciate you.

Thank you for joining us today. This is Linda Sherman, founder of the Rethinking Aging Club. We appreciate your supporting the Rethinking Aging Club podcast by subscribing to our channel. See you next time.