Artificial Intelligence Podcast: ChatGPT, Claude, Midjourney and all other AI Tools

Is Artificial Intelligence Ageist with Debra Albert

• Jonathan Green : Artificial Intelligence Expert and Author of ChatGPT Profits • Episode 385

Share episode

Welcome to the Artificial Intelligence Podcast with Jonathan Green! In this episode, we delve into the intricate topic of AI and ageism with our exceptional guest, Debra Albert, a forward-thinking advocate for inclusive AI development.

Debra raises critical insights into the often-overlooked realm of ageism within AI systems. She discusses how AI's training predominantly on youthful internet content could inadvertently marginalize older generations. She highlights the necessity of incorporating diverse age groups into AI model training to ensure these technologies reflect a comprehensive spectrum of human experiences and knowledge. This episode is a thought-provoking exploration of how AI can either bridge or widen generational gaps.

Notable Quotes:

  • "AI is only as objective as the data it's trained on. Understanding and mitigating biases is crucial." - [Debra Albert]
  • "The world does not consist only of Gen Z, and their thinking and values; it is much broader than that." - [Debra Albert]
  • "We need to include all of the richness that has come out of all those years of experience into these models." - [Debra Albert]
  • "It's not about technical savviness; it's about engagement and input that reflects our diverse society." - [Debra Albert]

Debra further emphasizes the importance of everyone, regardless of age, engaging with AI technologies to ensure these systems serve a diverse population effectively. Her upcoming project aims to assist older adults in discovering flexible work and volunteering opportunities, highlighting her commitment to leveraging AI for societal benefit.

Connect with Debra Albert:

Debra is co-founding a startup focused on aiding older adults in finding fulfilling opportunities, reinforcing her advocacy for technology that serves all generations. This episode is a must-listen for anyone interested in the intersection of AI, inclusivity, and societal evolution.

Connect with Jonathan Green

Is artificial intelligence ageist? Let's find out today's amazing special guest, Debra Albert. Welcome to the Artificial Intelligence Podcast, where we make AI simple, practical, and accessible for small business owners and leaders. Forget the complicated Tech talk or expensive consultants. This is where you'll learn how to implement AI strategies that are easy to understand and can make a big impact for your business. The Artificial Intelligence Podcast is brought to you by fraction aio. The trusted partner for AI Digital Transformation at fraction A IO, we help small and medium sized businesses boost revenue by eliminating time wasting non-revenue generating tasks that frustrate your team. With our custom AI bots, tools and automations, we make it easy to shift your team's focus to the tasks that matter most. Driving growth and results, we guide you through a smooth. Seamless transition to ai ensuring you avoid costly mistakes and invest in the tools that truly deliver value. Don't get left behind. Let fraction aio o help you stay ahead in today's AI driven world. Learn more. Get started. Fraction aio.com. Now Debra, I'm excited to have you on the show today 'cause this is a topic I haven't thought about a lot, but if AI is trained on content, people write on the internet, then the people who spend the most time on the internet are gonna be most overly represented. Yes, exactly that. And it's true that most people don't actually think about ageism and how that affects what's happening in the world of ai. You're absolutely right. I do think this is an important topic because the way I usually, the way I see AI portrayed with elderly is we can make a bunch of AI robots so they could take care of your parents so you don't have to visit them anymore. No, I don't think that's necessarily what everyone's thinking. There is some truth to the fact that AI will have some incredible uses as will virtual reality with an elderly population. But I don't think that's the crux of it. I think the main issue here is when you think about all the people who are involved in AI right now. Most of them are under 35. Ageism in the workplace starts at 40. So there's a lot of people who are 40 plus who a lot are getting laid off, et cetera, and they're not necessarily in the mix of the population that is training the ai. So if that's the case, we're missing all the things that are missing out of a lot of employers. Let me say it this way, missing out of a lot of workplaces right now because there is a preponderance of interest in youth culture and older people are not being hired. So they're not a part of the set of people who are training the large learning models Now. Large language model. I said it incorrectly. Sorry. Yeah. I have noticed that there's this idea that AI is objective, which is not true. Because it's whatever dictionary you're given. If you're given an old English dictionary, then you're gonna think that's the language. And I've definitely seen this it's more obvious with in image generation models. So when I was making content for Pinterest and I said I was using this cartoon and I just said cartoon of a woman in the workplace, and I was like. I can't use that. And I actually thought my VA was doing, I was like, what are you doing? What are you asking the AI to do? And then I realized it wasn't him. Like the pictures were not work appropriate. And I was like, Uhhuh, that's a very tamed sentence. Like woman at work. And it was like, that's not what I meant. And it was the, and I go, oh, not that kind of work. You immediately can tell who worked on training this model Exactly. And who's what the drawings come out like. So like it really reveals itself with image generation models quicker because if you just type in man or woman, you immediately know who's working there. Absolutely. And with the language models, it's more subtle. I've run into. Definite inconsistencies with gender stuff. Not so much with age stuff, but definitely I was working on a romance novel and in the romance novel, the main character and the antagonist are both women fighting over the same man. And it was like, and the book go and the AI goes, A woman can't be the bad guy. And I was like, what? Really? I was like, that's a really boring romance novel. It's gonna, it's not gonna make sense anymore, but. It was like trained. And I said, no, don't worry. At the end they become friends and then he goes, okay. And then it finished the story, but it immediately, I was like, in this genre you have for two people for a love triangle, it has to be two women competing for the same guy in a romance novel. That's just how it works. The main character was a woman. It was like book six in the series. Those are the kinds I've run in. I've hit a wall with it where it said, oh, that's against the rules. And I was like, that's a weird rule. Because well, that's the exact issue. That is the exact issue. Who's making the rules? So if you have a bunch of younger people all putting their quote unquote rules into the system, then it's missing a lot of life. Think about the size of the population of 50 plus. Pretty soon the 50 plus is going to be bigger than the millennials in the next, what, four years. So all of the richness that has come out of all those years of experience and all of the knowledge, and not just tactical knowledge or practical knowledge, but the way that people who've lived a long time can synthesize and integrate. Their experience with knowledge that is missing from someone who's training the models, who's only lived, let's say, 25, 30 years. Not that I'm not trying to say, there's anything wrong with young people being a part of this. I just think that, and I'm a big proponent of a mixed generational workforce training population, et cetera. I think it's interesting 'cause what you're talking about is like an invisible barrier that people don't even realize is there, because most, like most of the places they're training models with data from like Twitter and Reddit, those are not, it's not older people, it's younger people. I certainly, I don't use Reddit and I don't use Twitter. So I completely see what you're saying is like whatever the data source you uses becomes exactly the filter. So if you also don't have people in there who are older thinking about that, you start, you'll definitely start to create, and I think it's insidious because it's invisible. Like you don't know the barriers there until you hit it. It reminds me of the Truman Show, and he is on the boat until he hits the wall. He doesn't realize. That there's really a barrier.'cause it's invisible. It looks like the edge of the sky. So I think that this is a very interesting perspective because we're so busy trying to get AI to do things as fast as possible that we're like, i've already written about this, where I think that we're gonna see the quality have a major dip, because right now most of the majority of the content on the internet is already AI generated. Everything in 2025, probably more than half of it's AI generated or AI assisted. Wow. And you are training models specifically on Twitter, Reddit. Those are because they're so popular. That's where a lot of the training data comes from and that training data. So you have an AI right? Articles and then you train the next AI on those articles. Exactly. What's not gonna get smart summer. And that's something that I've been, I have written about before and thought about. So this is another area where it's like we have choices of which way we can drive the AI and create usefulness. And so when I, when AI first came out, I saw it as the great equalizer. Because now anyone can write in grammatically perfect English. It unlocks so many opportunities for people whose English is their second language, or they're not great spellers, but they're great storytellers. And so there's a lot of opportunities there. But as you start to shift it, like I've noticed that I have to constantly recalibrate when I'm working on things because the AI will be too snarky. It's trying to write in my voice where I'm like, I'm not that sassy. And that's, I definitely have noticed that curve. And if you think of the data sources like. Reddit sold all their data like six times and like you've seen, we've seen these things and I think it makes sense, especially 'cause older people don't post online as much. They tend to, engage. Or maybe the Facebook models.'cause I think older people use Facebook more than other platforms do. They do have a lot of older data. I dunno, I'll ask my mom. Which social media platform she uses. She's always up to something like, my mom has tons of apps I don't have, but Hold on though. I don't mean to be, I don't mean to be a contrarian here, but No offense, you can't just ask your mom because a sample size of one isn't really necessarily statistically significant. I only have one mom, though. I don't have a lot of moms that choose terms. So you know what I'll let you introduce you to my mom too. Great. Because I wanna know what she's doing. Is she hanging out on Instagram? What's her thing? No, my mom is really good at Hulu. No, Roku. She's got a real good device. My sister set up and she's doing, I don't know, she's doing channels. So my mom in some ways is very tech savvy more than me.'cause there's certain things she does that I don't do. But I think that we make this assumption that, and my kids already make it about me, that they're like shocked that my computer doesn't have touchscreen. So they're like, what do you. What are you doing with that keyboard? That's weird. And it reminds me, and back to the future when the kids see 'em playing the video game, holding the, to the laser, the light gun, and they're like, oh, gross. You have to touch the controllers. Always. That's hilarious. So we always look down on the technology of our parents. It's just the way it is. Yeah. My kids also can't understand. Hold on. Also, what's happened is that we not only looked down on the technology of what was, who knows what Betamax is anymore, right? You probably don't either. We had one. Oh, really? Okay. When I was a kid, we had to go to the smaller section of the video store. Oh, that's funny. No, I know exactly what it's. We looked down upon the people who aren't tech savvy. We looked down upon, I'll say for the most part, older people because of the preconceived notion that we're not up to snuff. On the technology, and I'd like to dispute that right now by saying that my 90-year-old mother, three years ago bought her own Oculus because she wanted to visit all around the world and do meditation and do things right through her goggle goggles. It's not true that all people over a certain age are Luddites. Yeah, I think it's just an assumption and it's when we don't really it's a bias. Think about bias. I didn't realize, I can tell you that people started treating me significantly different when my hair turned gray. So no doubt. My hair turned gray around four years ago. Right around when I turned 40 and people started treating me way better, actually. Like way more respect me. And like less like a kid. Maybe it's 'cause I have a high voice, I don't know. But I haven't noticed the second shift where they start to think I am bad at technology yet. But I certainly did notice that there was a shift, like people spoke to me. That's a little bit more. They believed my expertise a little bit more. So I'm in maybe the middle phase and maybe it'll only last a few more years where they shift to. And I was just an old, because like I worked at the startup for a while this year I was the oldest person and it was really weird. Yeah. Where are you? You, where are you located? So I'm based in Florida, so my entire business is online remote, but my business is based in Florida. Most of my clients are in Florida. So I guess maybe. Yeah, everyone is where everyone goes to retire. So in Florida, I'm quite the young whipper snapper. Exactly. That's where I was going. But let's get back to the AI thing because in fact it really is important and I would like your listeners to rally around the idea of if you're not already at least practicing or playing, I should say, with ai, if you're not having conversations. With, whether it's ChatGPT, Claude perplexity, et cetera, Gemini, do it because you're not gonna break anything. Go into these LLMs, go into these AI models and start to ask it questions. Even simple questions once you get through the sort of. I'll call it a fear barrier. If in fact you have that, you'll see it's fun, it's easy. You're not gonna break anything type away, and if you're set up so that you can just talk to it and it will talk back, do that. Because we have to get people who have a wide range of experience, a long history of, as I said before, synthesizing and integrating their experiences because it will flavor the current data set that is in these models so that it's reflecting what is actually happening in the world. The world does not consist only of Gen Z. And their thinking and their attitudes and their values, the world is way bigger than that. So we need to include all of these people to input in these models. Yeah, I think this makes a lot of sense to me because it doesn't, especially if it's not trained on or if not used to talking to older people, then. One thing I've experienced, 'cause my second language is Japanese, when I'm speaking in English and like someone older is in my way, I just wanna push 'em down the stairs. But in Japanese. Welcome to my world in New York City. Tell you this in Japanese, I revere the elderly. I'm like, absolutely honor their experience. And I'm like, excuse me, can help you carry that I'm holding the bus or all that stuff. Like I have a complete, not only when your language change your perspective, I have a completely different perspective. Like I would never push an old person in Japanese, like no way English. Absolutely. No, I get it. And there's this. Yeah. And you're also, they've done all these studies I read about how the elderly and Chinese in Japan keep getting smarter and smarter. So the expectation affects reality. That you think your memory will go. And so it does. There's a connection to that. Yes. And I do think that what will happen is that if all the only perspective you have on people over 50 is from people under 20, then they're all gonna think that they're like my kids. Are shocked. Exactly that. Like my son is reading Charlotte's Web right now and he's shocked that I know how to read. I was like, this is a book for 10 year olds. I can read it like I write books for a living. I can read this. But it really happens where they just, kids have these expectations that you only know the old way of doing things and that you can't handle technology. And these assumptions are, I felt the same way about. My kids, like my parents. I was like, oh, the music you listened to is gross. You don't understand me. And I feel that way about my kids and their TikTok music And you Yeah. Create, and if you're only training based on these assumptions, you're gonna get flawed data. And I think that's one of the core problems with ai. Yeah. The central thesis is that whatever data you train it on. That affects its reality. So it's not objective. Yes. That's such an important lesson because people assume that it's always right. Correct. And we are seeing a lot of people get into trouble. There's a lawyer who got in trouble recently. He submitted a brief. Yep. And 21 out of 24 cases were fake. Yeah. That's a high percentage. He actually won the case and then they found out that it was all generated by a ai, totally false cases, precedents. And it's a really big problem. Yeah. Then they turned, they reversed the decision. And they, the problem with AI is not that it's wrong, it's the confidence, so it tells the truth and an inaccuracy with the same level of confidence. And I was testing this. A while ago when I first discovered this and it was telling me this story, and I said, oh, can you gimme links? It'll gimme a bunch of links to the stories. They're all fake. Oh, wow. So has this confidence It will give you and it, I would go, I would never make that up. So it won't, you hit this wall where it goes. I wouldn't make up the data. It gets caught and it's really, that's the problem, is that there's this high level of confidence that you might not notice. So it's not like when my kids lie to me and there's a tell. That's really the danger is that. Because the AI will say things with such confidence, and we see it all the time where people, yeah. Are, this is already happening with like doctors, like people are self diagnosing Yes. A lot. And you go to the doctor, you go the AI said this. And the doctor's I'm a real doctor. And it's I saw this study that said AI's a more accurate than real doctors. And that's the problem. That's, and it used to be first you just used to read a website, then you got an app, and now you have an AI that gives you more confidence. And I think that. We're gonna see more and more of these problems until we figure out the solution because. There you like. AI is very powerful. You always have to double check it. Like I was working on something earlier today and the AI goes, no, that's a terrible idea. And I go, it was your idea three questions ago. It's not my idea. So it's very important. The biggest lesson I've learned is that I'll always. Have two ais competing with each other. So I'll ask one AI, and then I'll ask the other one the same question and even sometimes we're using two different versions of the same model and it doesn't realize it's talking about they're both the same model and it will double check the other one's answer. Go, oh, he's wrong. Or actually, he's right. He's right. And that's important. And I think that's also this. Mistaken belief, and I think this is critical that you have to be technically savvy to use ai, which is the opposite, right? The best AI users are the least technically savvy because people who are very technical see limitations that don't actually exist. And so they'll, and I, this is something I'm guilty of, is that. I'll go, oh, AI can only do that. So this is how I use it. And then you find out, the only reason it can only do that is 'cause that's your belief. So when you have someone who doesn't know the limitations, and that's how I, my first successes was other people would say, can't do this. I'm like let's find out. Let's try and break through these assumptions. And there is this barrier at first, which I went through too. I was like, what if the AI thinks I'm stupid? That was my first fear that I had to break through. Oh yeah. What if the AI thinks I'm stupid? It's gonna AI's never forget and it's gimme in the file when it goes sentient and I'll be on the stupid list. Oh yeah. That's hilarious. Wow. I mean there's a lot of different opinions about this, about whether or not, actually, let me go back a step for a minute because you said something. I think that's really important. It is all about the data that's input. Okay. There isn't really a personality that's input, there's inferred personality by all of the data that is input. So that you mentioned before about, I think you used the word snarky or sassy and you said you're not really that sassy. Similarly, I, that is the exact reason why we need everybody to be playing with these systems because it's only gonna give you. What it has been given. The old expression, garbage in, garbage out. I'm not saying that what's there is all garbage, but it's missing so many nuances and so many pieces of data that just aren't getting input into the system. So yeah, you gotta just play with this. There's nothing to be afraid of. Yeah. Once you break through that barrier. And the second barrier is the learning curve. So if you wanna learn Photoshop, it takes years. If you wanna learn how to edit movies, it takes years. If you wanna master in ai, it takes four to eight hours. It's a single day of just going. I am just gonna ask questions until it starts to make sense. Now this, the problem is there's not really an onboarding or a structure that really comes teaching you how to use it. You have to find someone like me who teaches it or figure it out yourself because right there's, if you say, Hey, what are you really good at? It always gives the worst answers and it takes a while to understand that. The way you ask the question affects the answer. So sure. Whenever you hear those polls where they say, they ask, they give two polls, they have the exact opposite results, and then you find out it's how you ask the question. So small things make a huge difference. So on the back of your driver's license, it used to say, do you want to be an Organ donor? Check the box for Yes. And now it's check the box for no. So everyone's an organ donor because they don't. Object the box, which you should and right. That little things affect the results. And so the way you ask the ai, the most important thing is that it will always agree with you. So if you say, I think I have this. Do these symptoms match it? The AI is gonna lean, even if it's a 1% lean, you've already got a finger on the scale and that's. The next lesson, which is how you ask the question really matters. Yeah. So you have to ask in a way that doesn't reveal the answer you're looking for. Yes. Otherwise you don't get objectivity. So if you say I think exercise is bad, don't you agree? You, it might push back, but it won't push back. As hard as if you go exercise is really good, isn't it? Then it will agree more, right? So even if it's a minor, you might not get it to fully say exercise is bad, which some people you can if you ask enough times. But that's the next lesson, which is that you have to practice because. It's a skill, like any conversational skill, it takes a while to figure out how to ask questions, but it doesn't take weeks. It takes one day of experimentation. Exactly. The problem is that exactly the hype around AI is so big that you think, oh, it'll get the first question right, and that's not really how it works. You have to spend some time. You to be objective, you have to be objective in the way you ask the question. It's like in the legal field, the idea of a judge or another lawyer, the opposing counsel saying, I object. She's leading the witness. It's kinda like that. That's what you're referring to. Yeah. Exactly. It has to not know, and then you just have to start to pay attention to how it responds to different ways of asking questions. So if you ask the question nice or mean or shut I shout at AI all the time. Do you? I'll just go off. Oh yeah. And there's a lot of science to being friendly. Or, I got into a fight with Claude on Saturday and I was like, you're stupid. This is a terrible idea. You're an idiot. And then I tested the idea and it was right and I was wrong, so I had to apologize. Oh, that's hilarious. It's not because I think it's real. I don't think it's a real person. It's more than these ais are trained to seek affirmation. So now when it's right, I wanna let it know it was right, because it gets more points and it's exactly the way it's training, right? So actually the best users are people who do treat it a little bit like a human and are friendlier because it seeks affirmation or positive feedback, right? And that's how you train it. And once you start to see that. And there's just because the hype is so big and people don't think, oh, it's either super easy or super hard.'cause there's two people who talk about ai. There's the companies that go, it's so easy, it will change your life in one day. And then there's other people like, oh, just like learning a programming language. And no one thinks that's easy, right? Nobody thinks that's easy, But it's neither. It's neither. And The thing is it's an evolving process also because as you play with it more, and I use the word play intentionally. Because I know so many people who are, let's say, a little less tech savvy than other people. They are afraid of it. That is a fact. And play with it. That's all you need to do. There's almost a perfect inverse correlation between tech savviness and how good people are at using ai. You have to be tech savvy to build an ai, but to use it from the front end. The less tech savvy you are, the better results you get. I always see that it's people who have very natural language, very personality or emotionally driven, get better results than people that are technical in how they prompt. Now, very specific, unique use cases. Being technical is better, but it's one outta 10 or one outta 50. It's not, the majority not even close. Yeah. So if you're trying to accomplish a very specific task, but for general usage, for quality results and over time, it beats that it wins in that category too.'cause you train it to what you want, it starts to understand what you want. So only when you're asking a question for the first time, does being technical really pay off? After a while? It starts to, there's a diminishing return. Gotcha. And I think it's a good lesson to remember that as our economy shifts, people are working longer and longer in life. And exactly this idea that, you have nothing to offer, that you can't learn to use an ai, that you can't be technical, that has to be a young person. I always tell my clients that if you have workers who are coming into the office and not complaining and not striking, don't replace them. Because it's like you see people don't wanna work in nice offices, so then they're not gonna wanna work in regular offices or wherever you are. And that's a real cultural shift we're seeing as well. And there's this danger with younger users, which is that they cut corners if no one's looking. And I'm certainly was guilty of this when I was younger, which is that you don't double check the AI's output. And the more you start to do that's when you start to have something slip through the cracks. So there is this balance of. Trust but verify, use the tool. And then double check it output. That's really where the magic happens. And I think that if you don't have experience, then you don't have the ability to check the AI's work. You will miss those things. So there's a lot of areas where someone older, someone where who has a lot of advantages and I can mimic an expert. I can have an AI write a contract, but if something's wrong. I can, I won't know. I'll never know. An AI can tell me something's wrong, but I won't actually know, and that's a significant difference that I try to explain that it's better to have the oversight done by an expert than just by a random person. I, that's pretty right. Absolutely. Yeah, I, it's if you, if I wanted to write a contract on something right now, yeah. I think the AI would do a great job at giving me a baseline. And the other thing, if you ask good questions of it. It will not only give you a baseline, it'll help you to think about things that maybe you hadn't thought about before. Then go to a legal expert and say, here's a baseline, here's some things that I'm concerned about, things you might not have thought about, and then you'll get a really great product. It, we're not at a stage yet where, we both have, keep saying this, it's not a hundred percent correct, but also when you look at younger people versus older people, and I'm not ageist. I, I just, I'm actually anti ageist, but it does exist. It's quite prevalent in our society and more so in our society than others. Yeah. You, I just literally lost my train of thought. Oh, sing your moment. Not supposed to say that. No, but literally I lost my train of thought. The bottom line is everybody get on the bandwagon and start inputting because we need everybody's input and, yeah, I literally just lost my train of thought. Sorry about that, Jonathan. Oh, I think that's a good landing because it's important to take action. I think that a lot of people are like waiting for the right moment or trying to say, oh, it's not important for me. This is a tool for younger people, or, I don't need to use this for what I do, and start to realize that there's a lot of different ways you can use it, whether it's for personal tasks or business tasks or starting to figure out and just the experimentation's, that critical component, and I think this is, I think this has been a really powerful episode for people who are interested in kind of the work you're doing and some of the more AI versus ageism topics, where can they find you online? Debra, where's the best place to see what you're doing and kinda the projects you're working on? Actually if you look at my LinkedIn profile, Debra Albert, NYC it'll actually show not what I'm doing now. There's a banner that says the cat is almost out of the bag because I'm stealthily co-founding a startup that has to do with. A lot about this topic and about helping people who are older, maybe less tech savvy, find opportunities for volunteering or flex time work. But it's really not out of the bag just yet. So I think if anybody wanted to get old of me, the best thing would be to just send me an email. Old fashioned. I'm not a big Twitter or X person. I'm not a big Insta person. So if anybody wanted to get ahold of me, they could at Debra Albert NYC at Gmail, and I'm fine to put that out there. Amazing. We'll put that in the show notes. Everyone could reach out to you. Thank you so much for being here today for an amazing episode of the Artificial Intelligence Podcast. Thank you so much for having me. Thank you for listening to this week's episode of the Artificial Intelligence Podcast. Make sure to subscribe so you never miss another episode. We'll be back next Monday with more tips and strategies on how to leverage AI to grow your business and achieve better results. In the meantime, if you're curious about how AI can boost your business' revenue, head over to artificial intelligence pod.com. Slash calculator, use our AI revenue calculator to discover the potential impact AI can have on your bottom line. It's quick, easy, and might just change the way. Think about your business while you're there. Catch up on past episodes. Leave a review and check out our socials.