Welcome back to The Immersive Lens! In this episode, we explore the intersection of artificial intelligence and everyday decision-making, taking a hard look at how large language models are becoming an unavoidable part of our cultural fabric. Whether it is Anthropic funding AI fellowships for nonprofits, the unsettling biases uncovered in a recent Stanford grading study, or the manipulation of multi-million dollar prediction markets, the central theme is clear: AI is holding up a mirror to society's best intentions and biggest flaws.
With Dave Ghidiu on a working vacation, Jenn Carney steps in to join Paul Engin for a lively, head-to-head comparison of ChatGPT and Claude. They test these platforms with real-world summer prompts, ranging from weekend relaxation schedules and custom road trip playlists to spontaneous budget getaways. Ultimately, the verdict is that while AI tools offer incredible speed and organization, maximizing their utility requires precise prompting, clear constraints, and a watchful eye for ingrained biases.
Key Topics
The Rise of AI Philanthropy: Anthropic recently announced a fellowship program that embeds young professionals into nonprofits to help them leverage AI tools. By offering a generous salary to candidates with limited full-time experience, the initiative aims to combat the potential for AI to exacerbate income and power inequalities by empowering smaller organizations.
Algorithmic Bias in Education: A recent Stanford study revealed that AI models provide vastly different feedback on the exact same middle school essay depending on the demographic tags assigned to the student. This sobering reality highlights that large language models absorb our human biases, requiring educators to be highly thoughtful and explicit in their prompting when utilizing AI for personalized learning.
The Vulnerability of Prediction Markets: Platform bets on Polymarket are meant to act as an uncorruptible truth machine by pooling human knowledge, but big money often finds a way to exploit the system. From manipulated votes by cryptocurrency holders resolving niche gaming bets to politicians leveraging insider knowledge, decentralized prediction markets are currently facing complex regulatory and ethical hurdles.
Mastering AI Itinerary Planning: When comparing ChatGPT and Claude for trip planning, providing personalized context and hard boundaries yields drastically better results. To get the most out of an AI assistant, users should outline strict budgets, explicitly state what they want to avoid, ask for multiple contingency options, and consistently iterate on the generated schedule.
Mentioned in the Episode
Prompts Used
- I have six hours free on a Saturday. I want to relax, but also feel productive, build me a day.
- I need a playlist for a road trip this summer. I want something that feels like summer is actually happening.
- Can you export this playlist to something I can import into Spotify or Apple Music?
- I'm planning a spontaneous long weekend trip for two people with no destination in mind. The budget is reasonable, but not extravagant. Keep it around five hundred dollars. We want interesting food, something unexpected and minimal time in a car, starting from Canada, where the podcast studio is located. Give us three completely different options and make a case for each one.
- Let's choose Pittsburgh. Give me a more detailed itinerary.
Transcript
Click here to view transcript
Paul Engin And keep that one. I want to use the Jeff as the error for the intro. I think that'll be a good one. Jenn Carney No, it died before. Oh, man. Jeff Kidd Is he still there? Paul Engin Welcome to the Immersive Lens, the podcast exploring the technologies reshaping how we live, work and learn. Jenn Carney From AI and virtual reality to creative media and design, we're diving into the tools and ideas shaping our connected world. Paul Engin My name is Paul Engin, join us as we uncover the people and ideas driving the next wave of interactive experiences. Jenn Carney And I'm Jennifer Carney, standing in for the delinquent Dave Ghidiu, who apparently thinks that summer vacation is a thing. This is The Immersive Lens. Paul Engin Yeah, what is up with this Dave character, huh? Jenn Carney Yeah, I know. It's not right that somebody who works, um, you know, 23 out of every 24 hours should actually take a break. Unbelievable. Unbelievable. Yeah. But he's, um, the most productive, uh, person on vacation I've ever seen. I'm getting emails from him. He's handling stuff. Paul Engin Yeah, he was sending a video for, we were originally going to have another guest on talking about universal design and he created a video while he's on vacation for me to ask. Yeah, so this is my Jenn Carney question is it cuz he's just enthusiastic which we know he is or is it really about control does he need to be in control I think that's a topic for a future episode yes yes well Paul Engin have to do a psychoanalysis of day-to-day. Jenn Carney I bet you Claude could do it based on his previous chat history. We'll have to, oh, that'd be good. Yeah, that'd be fun. Yeah, yeah. Paul Engin to bring someone from psychology in and we'll have them psychoanalyze. That'll be perfect. Jenn Carney Not pseudoscience-y at all. No, no, not at all. Paul Engin So Jen, you've been on before, you know we start with our hot takes and I wanted to see if you wanted to start off our hot takes. Jenn Carney Sure. So I wanted to talk about a program that Anthropic announced and their announcement was probably like a month ago now, but I don't think you've talked about it on the show yet, but they rolled out a program that they're calling Claude Core, which is clearly a nod at something like the Peace Corps or AmeriCorps. It's, it sounds amazing. Claude is, Anthropic is sponsoring the cost of fellowships to embed young career professionals into nonprofits. Anthropic's first going to train them in all the different Claude tools and then give them a year-long paid fellowship experience where they're embedded in a nonprofit and using those AI tools to help the nonprofit achieve its mission. So I think it's amazing. And they're paying like a decent salary. The people who get chosen for this, Anthropic's gonna pay them $80,000 for the year to be on fellowship doing this work. And it's only open, sadly. You can't apply, Paul. I can't apply either. It's aimed at people who are early career, so you have to be 18 and you cannot have worked for more than two years full-time at that point. Paul Engin between 18 and a certain age, or is it 18? No, just... Jenn Carney You gotta be 18 and you can't have worked for more than two years. So, you know, if you were like that 55 year old person who's still in grad school and just goes from degree to degree, I guess you'd be eligible. Yeah, right. But so I think they're really sort of aiming at that, like that Peace Corps type audience, right? Those young, inspired, energetic people just coming out of school with tons of knowledge and a desire to improve the world. Paul Engin Yeah, that's really interesting. So it's even before they go to college. This is like right after high school. Yeah, or it could be Jenn Carney college graduates, right? I mean, or it could be somebody still in college. Paul Engin Okay, so it can be over 18. They can be over 18. Jenn Carney They have to be at least 18. At least 18, but- And they can't have worked for more than two years. Two years. Paul Engin Okay, I see what you mean. All right. Jenn Carney Um, so I think it's pretty fantastic and applications are open now, um, for their, their, they've committed to a thousand positions over several years. Um, but the first hundred are going to start in, um, October and applications are open right now, and then there'll be another cohort that starts in January. And another one, I think they said next August. Um, so yeah, so people should, uh, apply. Um, I encouraged my son to apply. So he looked at the application and there's actually some online courses you have to take before you can apply. Um, which is interesting. Um, so, uh, it looks like a great program. Paul Engin It's a great cause. I think that it's something that is really needed from big corporations sometimes for a greater good objective. Jenn Carney Yeah. And I mean, that's anthropic has been trying to convince us all along that it really is sincere in its desire to see AI help our world and not harm it. And, uh, clearly AI has the power to exacerbate income inequality and power inequality, and I think this is Claude or anthropic's way of trying to fight against that and, um, empower the small nonprofits. Paul Engin Yeah, I think that Anthropic is really, at least its core values, they're following through with them, regardless if it hurts them financially in some way. And I think that it's really, it makes me as a consumer feel more comfortable using it because I know that they're going to be, I guess, safe with my information or with stuff I'm using that I know that it's going to be used in a good way and I don't have to worry about that malicious aspect of. Jenn Carney Yeah. I mean, it's always hard to, to trust a large corporation, right? Um, so, but, um, I mean, I trust them more than I trust some of the others. That's true. Paul Engin Me too, me too. And my hot tag is actually with Anthropic as well. I saw Chloe Lebinski, who is, I don't know her official title at Anthropic, but she did a speech for the Alliance of Responsible Citizenship Arc. And it was interesting because she kind of talked about what artificial intelligence is. And when she spoke about it, it was interesting because she said that, you know, all of these models are trained and they build human-like characteristics based on what they're trained on. So, you know, the logic is, you know, you could have a malicious characteristic and all of the content and training around that could be centered around that concept, that characteristic, just like you could try to build a model around someone who is thoughtful, someone who is mindful of, you know, others. And then the responses and the training gets built around that because the training keeps iterating and it iterates around those concepts that you're training on. So what she was saying is that you will notice, and I feel like I've noticed and I know in talking with Dave that he's noticed, I'm sure you might have noticed as well, if you use different LLMs, they have a different way of responding. They have a different way of giving you similar, it might be similar information, but slightly in a different way or positioned in a different way, if that makes sense. Jenn Carney Yeah, well, oh my gosh, I have so many thoughts on that, because I think it ties into a couple things. I think the term, correct me if I'm wrong, is it called a hyperstition? This idea that what we give AI can sort of come to be, right? So what we're scared of, what we're nervous about, we can be creating that future reality. So for instance, we have a lot of dystopian literature where the robots take over, right? Well, the AI has read all that literature, right? We don't have a lot of literature where the AIs exist and are helpful and everything goes great. And so we are creating a scary future. The other thing is that there was a new study that came out of Stanford just this week. And actually, it's a little bit sobering and terrifying to me. They used AI, they took middle school essays and they used it, different AI models, but they tagged the essays, they used the same essay, but with different tags and said, you know, this is from a girl, this is from a boy, this is from somebody who is Latino, this is from somebody who is white. And the feedback that the AI gave on the exact same essay varied dramatically, depending on the characteristics that they thought the writer had. So for instance, if it was from a girl, the AI was more likely to say, I love what you've done there. If it was from a white male, the AI was more likely to give substantive feedback about the argument that the person was making in the essay. If it was from traditionally underrepresented minority, the AI was more likely to give feedback about things like spelling errors. So what this, I mean, this is terrifying. What this means is in our attempt to build personalized tutoring, what we've done is taken all of our human stereotypes and biases and taught AI that that's the good thing to do, to respond in a biased way. So we have to be really careful in the way that we're using these AI tools and even in this idea of personalized learning. So I think all that ties into what you're talking about here, that, you know, AI is a product of its training. And what we train it on is flawed. It's human and it's full of inequities and biases and stories. And and we have to be really thoughtful about that. Paul Engin Yeah. And I think that's why it's important that, you know, as you have that human element in there to make sure that what you're seeing is accurate or correct. And unfortunately, we can't control the training of these big language models. But it's really, that's really shocking about the biases. Oh, it's terrifying. Jenn Carney And for me as an educator, one of the things I'm doing all the time is telling my students, hey, you know, you can use AI to give you feedback and write, be a personalized learning support for you so you can show it your writing and say, hey, how could I improve it? But what that is actually saying is that students in my class might get very different experiences because the AI is making judgments about them based on their name, their previous context that it has about them. And it's gonna give different feedback. And interestingly, that study used four different large language models and saw the same issues across multiple large language models. Paul Engin We'll have to put that, that study in the show notes for sure. Um, that, that is really amazing. Cause you would think at this point we talk about biases and you would think these language models would already, okay, we're talking about it. They obviously are talking about it and I don't even know because of their training, how they're going to untrain. Right. Right. Cause now it's already in one facet of the neural network and you know, how are they, can they clean it up? I don't know. I don't know. It's really interesting. Jenn Carney Well, and some of these are unintended consequences, right? So if we label this essay as from somebody who's an English language learner, right? And it gives more feedback on grammar. That's the AI trying to respond in a way that's appropriate based on our tagging. But if we don't for that, you know, middle school writer also help them develop in that, in the strength of their argument and their overall composition skills and we're doing a real disservice, right? So we're turning these students into one-dimensional characteristics of themselves. Yeah, or I would... Paul Engin wonder if we have to start really focusing on better prompting. So we say, you know, we say, don't look at this as a, you know, African American male, look at it in an unbiased manner. I want you and kind of have to verbalize it in the prompt. Yeah, give it more context on what not to look for and what to look for. Jenn Carney So you got to be like anti-racist in your prompt explicitly. Yeah. I mean, maybe it's interesting. And I think it's revealing in what the AI has learned about society is revealing about society itself. Paul Engin Yeah, that is and now you have another hot take Jenn Carney Yeah, do we have time for it? What is it? So I'm interested. So, um, I read an interesting article yesterday and I know this episode won't air for a couple of weeks, probably. So, um, the article was in the New York times on June 28th and it was called the donking of a truth machine. And it was a, about a prediction market bet, a poly market bet about whether the word donk would be said during a seven hour e-sports broadcast, the PGL Bucharest grand final, final. And like anything in poly markets, sometimes, uh, these are silly bets about inconsequential things, but with a lot of money at stake. So there was about $5.7 million in betting on whether the word donk would be said during a seven hour broadcast. And interestingly, there is an argument about whether donk was set or not. Um, and donk, by the way, is the name, I guess, of a particular gamer. Um, but it really doesn't matter. It's just that it's a unique word. And, um, one of the commentators bumbled the word don't and it sounded like donk to many listeners. Um, so there's a debate about whether that fulfills the poly market that or not. So it turns out when there is a debatable outcome in poly market, they have a way to solve this problem, to decide what the truth is. And they call it the optimistic Oracle. And the optimistic Oracle is essentially set up to be an arbiter of truth. And what it is, it's kind of like its own little mini prediction market, like its own little meta prediction market. The only people who get to vote in the optimistic Oracle are people who own cryptocurrency called UMA, U-M-A. And, um, the idea is that if they vote on the side that actually wins, they get a little, um, bonus. And if they bet on the, if they, uh, vote on the side that loses, then they lose a little bit. So it's meant to be an uncorruptible system because they're going to want to be right and win their money. But like any other uncorruptible system, when there's big money at stake, uh, humans find a way to corrupt it within just a few months, right? So there is, um, in this particular case, uh, there is a, there's a company that helps UMA owners know which way to vote. Um, and it takes a cut and it can then essentially swing the vote on these debatable poly market things. So instead of being, instead of the prediction market, taking all of the knowledge of the human race and figuring out, right? That's what, what prediction markets are supposed to do. Instead, because of the money involved, we have once again, um, push the decision-making to just a few people who happen to have a significant amount of money already, um, in the state. Paul Engin It almost feels like lobbyists. Like they got lobbyists now. Jenn Carney It's like crazy. It's so insane. And it's just kind of mind blowing. So poly markets are such an interesting real world experiment. But we're finding all the ways they can be corrupted. And I mean, I'm sure you've heard the George Santos a news story over the last few weeks. No, what is that one? Okay, so remember George Santos, our famous politician who lies about everything, right? Yeah, so there was a prediction market bet. I can't remember which market it was on. I don't know if it was poly market or not, but a prediction market bet about whether he would be in attendance at Trump's inauguration. And he had made a statement in public that he was planning to attend. So of course it swung the market strongly in that way. And then he bet heavily that he wouldn't attend and he didn't attend. And so he won a bunch of money, right? And so now he is being charged. Okay, well, that's what I wanted to ask you. Cause I was- Paul Engin thinking about it, I was like, wait, this whole dunking, like, if I was one of the announcers for that, and I saw that, I could bet that I'm going to say it. Jenn Carney and then you could say it. And I could say it, right? Yes. What would be, how is that illegal? And so it's so interesting because they're trying, so I learned recently that the prediction markets don't fall under the normal SEC. They follow under a separate regulatory body because sort of the traditional rules about insider trading, it's unclear if they should apply to prediction markets. Because prediction markets, the whole idea is that we are pooling all the best knowledge of all the people in the world. And if somebody has insider knowledge, wouldn't we want that knowledge to be part of the prediction market? But should they be able to bet on it? Right, that's right. Paul Engin Right? I mean, because then it's like, I don't know. Jenn Carney complex new world we are living in. I don't know, I've never been on a prediction market, if you call. Paul Engin I have not. I have not. I know that it's becoming a bigger, bigger thing. But I might look into I watch it sometimes to see, you know, what trends are happening. Yeah, yeah. But it's, I don't know, I feel like with something like that, it can be swayed a lot more, right? Where it's like, you know, will so and so fall during this concert? You know, I don't know. It's when there's something that can be that manipulated so easily. I don't know if it's a good betting. Jenn Carney Yeah, I don't actually, this is just my lack of understanding of economics and big business. I don't understand how the markets make money, right? Like how is this worth it for poly market when it can clearly be manipulated in so many ways? But I don't know, the house always wins, right? I don't know. I don't quite get it all. I'm sure they get a percentage of any transaction that happens, so it doesn't matter, win or lose. Paul Engin They get a you know, you bet a you bet a dollar they get 10 cents. No matter what. Yeah, you win or lose it Doesn't matter. They get 10 times. Yeah. All right. All right. So why don't we get into our deep dive today? Um, I I think this will be a fun one This deep dive is called AI planning and we're gonna gear it a little bit more toward like our summer fun because it is The summer. Yeah Jenn Carney Absolutely. I got a road trip this week, a 13 hour road trip and then a couple days later I got to take that 13 hour road trip on the way back. So I'm excited that this is going to help me plan. Paul Engin Oh, awesome, awesome. So I was thinking what we would do is we would use two LLMs. We'll use Claude and chat GPT. And you'll use Claude, and I'll use the chat GPT. Does that sound good? That's good. Jenn Carney sounds great. I am seeing Paul's laptop for the first time and he has like 1,000 icons along the bottom. Paul Engin productive tools down here. I don't know how he's finding anything. It's microscopic. It is. All right, so we'll start a new chat for each one of these so it doesn't borrow from anything we've asked it before. All right, sounds good. Sound good? Yeah. All right, so the first one we'll do is the I'll read the prompt and then we'll see what we get from it. So the first one is going to be I have six hours free on a Saturday. I want to relax, but also feel productive, build me a day. Jenn Carney All right. And my prediction is that it's going to struggle with this because there's not enough contacts. What do you think? Paul Engin I'm thinking the same thing because it doesn't, unless it's piggybacking off of things that I've asked in the past. Jenn Carney Yeah, I mean, Claude knows a lot about me, I'm not gonna lie. Paul Engin We'll see what happens here, all right. Jenn Carney All right, Claude has titled this chat balancing relaxation and productivity. Paul Engin Yeah, and I have a slow start. I have I have it broken up in the sections and I have it's called slow start. Mine's done, by the way. Are you minus. Jenn Carney Still going here on Claude, yeah. Paul Engin are you on low? Okay. All right. I'm just your model. I just wanted to see if you were Jenn Carney Yeah, listen, I got to save my tokens for some work this week. So we're going to stay on low. Paul Engin Uh, all right. Jenn Carney All right, so what did chat GPT plan for you, Paul? Paul Engin So I have a slow start for the first 30 minutes of my day. I'm gonna skip the urge to jump right into tasks and I'm gonna make my favorite drink, whether it's coffee or tea, so it doesn't know what I prefer. And then it is coffee. So sit somewhere comfortable, read a chapter of a book simply to enjoy a quiet morning without my phone. The goal is to start the day feeling intentional rather than rushed. And then it gives me an hour and a half block after that of a meaningful project to do. And I can choose one, organize a room, work on a hobby, finish a personal admin task, learn something that I've been wanting to learn, and then keep it to one part, but it's saying limit it to one of those. And then- Jenn Carney Not specifying, but it's saying, here's a block of time and you should do one thing productively. Yes. Paul Engin tell me to get outside so visit a park do something outdoors that I enjoy just take a walk um oh this is interesting it says leave headphones off so I can enjoy the spatial because I see that a lot people are always having their headphones on um and then um I have a little lunch lunch break uh and uh watch an episode of a favorite show or play a game or read um and then reset for uh my future self so this is spend an hour doing a few small tasks so tidy up for 20 minutes prep a meal do a load of laundry um and look over the calendar for the following week and then I end my day with uh something enjoyable with uh something that feels rewarding to me so something like uh cook a nice dinner or go meet with friends or watch the sunset or work on the hobby that I started so I can Jenn Carney So it looks to me like it gave you way more than six hours, right? The prompt said six hours, but it, what time did it start your day? It's ending your day at 6pm. Paul Engin Oh, no, no, it's it's our block. It's it's not time framed. It's saying the first 30 minutes, the next hour and a half. And so and it just blocks out the six hours. I see. I see. I thought that was time. Got it. And then it broke down to, you know, 25 percent relax, 25 percent meaningful accomplishments, 25 percent movement and recovery, 25 percent future planning. So that's interesting. Jenn Carney It feels like you need an even split. All right, so I'm going to say Claude gave me a checklist. Like it's formatted really nicely in a way that I can check things off. And it's much more personal than yours, Paul, but it's also wrong in some of its assumptions about me. So it has me starting again, like yours, a slow start with coffee, no fun, or no phone, I'm sorry, not no fun. I was doing nothing useful, journaling, sitting outside, watching the yard, letting my brain boot on its own terms. And it assigned that to a 9am time in particular. And then it has me spending 45 minutes doing banjo practice. Now, I don't own a banjo. I don't play banjo. I did talk to Claude about if I wanted to learn banjo, how would I do that? So it is assuming that I've progressed in that life plan, which by the way, is not really a life plan. I never knew. I know is really, I just had a whim one day and thought, well, it'd be fun to learn banjo. And I talked to Claude about it. And now it thinks I'm committed. And it does not realize that that idea has come and gone. And then it has me doing one task with a clear finish line. And it's giving me some very specific things from my recent to-do list. And so it's giving me some ideas that are very specific to me for 45 minutes. Then it's suggesting I take a walk-a-bike ride or time in the garden. It's referencing the Finger Lakes so it knows where I'm at, cook something real for lunch. Then it has me doing a craft project for 45 minutes. And it's referencing specific crafts that I actually do like fiber art. And then winding down 30 minutes of reading or sitting or maybe playing the banjo again for no reason. Paul Engin So I think you must have memory enabled on mine. I have it disabled. Jenn Carney I do have memory enabled on mine so that's Paul Engin That's what it's borrowing. That's interesting. So Banjo, that's awesome. I never knew. Yeah, it's not going to happen. I do like the way it formatted it with a checklist, and it's easier to read yours than I think it is to read mine. Yeah, and it's giving me... Jenn Carney an option to swap a block at the end for anything different or to shift the timing so and it seems to you know have created a way for me to easily do that. Paul Engin So I wonder if for you, you would need to add a context that said something like don't reference my previous. Jenn Carney I mean, but why would I want to not reference it? If this was real planning for me and not a podcast experiment, I'd want it to have that practice. I might need to go back and update the memory and let it know that I am not purchasing a banjo. Or working on a. Paul Engin a banjo project of any type. Jenn Carney I know, unless someone wants to teach me banjo, I'm open. Paul Engin All right, so what do you think, outcome-wise, are you leaning one, it seems like they're somewhat similar in how they structured it with like a ease into the day. Did you have any future planning? Jenn Carney it did not do any future planning. Yeah. And so yours was more split, I would say 50-50 productivity and relaxation. And Claude said mine was split more 60 relaxation and 40 productivity. Paul Engin Oh, Claude's lazy. Jenn Carney I mean, we didn't give it much context. Maybe Claude knows that I'm lazy. Maybe it's based on previous memory. Paul Engin Maybe it's this whole Dave thing. Maybe Dave is like seeping through Claude. Dave is not there. Jenn Carney All right. How about, are you ready for a new one? Let's do this one because this applies to me this week. I need a playlist for a road trip this summer. I want something that feels like summer is actually happening. A mix of decades, some sing-alongs, nothing that requires explanation to enjoy. Give me a 20 song playlist with a sentence on why each track earns its spot. Paul Engin All right, so mine knows nothing about me mine thinks I like So you did a new chat I did Jenn Carney Yeah. Paul Engin Hey, you might get a playlist from this now, right? Jenn Carney Once again, chat GPT is way faster than... Yeah, I already got my 20 and you're still thinking. Paul Engin but maybe yours is more thought out or it's going to be structured. Jenn Carney Right. I mean, maybe mine is just more personalized and yours is, uh, um, you know, a quick Reddit, uh, list. Paul Engin I got, while you're still thinking, I got Mr. Blue Sky, Electric Light Orchestra, what if it's practically sounds like a sunshine arriving, making it one of the best possible opening songs for my trip. I got a little Brian Adams Summer of 69. Jenn Carney One of my all-time favorites. Dream. Paul Engin Fleetwood Mac, Fast Car, Tracy, oh my gosh, Tracy Chapman, that's old school. Yeah, yeah. Jenn Carney Yeah, rather than the redo that hit the last year or that was so popular. Paul Engin Yeah, these are all old school. Yeah, now we asked for a mix. Jenn Carney a decade. Is it a mix of decades? Do you think? Paul Engin We got Weezer's, not yet. Weezer. I know. Shut Up and Dance, Walk the Moon, Katrina Eagles. These are, I don't, Outcast, Amy Winehouse. Jenn Carney All right, so give us another song and the context that it the the argument that it made for it. Paul Engin Well, I'll do cruel summer Taylor Swift modern pop at its most exhilarating with a chorus built for shouting in the car. Jenn Carney Well, that's true. I can get behind that. Yeah. Paul Engin I would say that 90% of my songs are older songs, right? Outside of the Taylor Swift and I have a Dua Lipa, everything else on my list is like 10 years or older. Jenn Carney have context on how old you are, do you think? Do you think that's based on you? I mean, we did ask it to be a mix of decades. Paul Engin Oh, you know what? I I guess I'll read the top here. It says mixes classic 90s to 2000s favorites. Hmm. I guess that's Jenn Carney That's not a lot of decades. Paul Engin That's not a lot of decades. No. What did you get? Jenn Carney All right. So here comes the sun from the Beatles from the 70s. It says, opens every window in the car, three notes in, and you're already somewhere better. Life is a Highway by Tom Cochran, Walking on Sunshine. I got that one too. Did you? Cruel summer. So we got some overlap here. Mr. Brightside, did you have Mr. Brightside? Paul Engin Uh, Mr. Brightside, I do not have. Jenn Carney I got a little queen, don't stop me now. Ooh, I have Dua Lipa too, levitating. Is that what you have for Dua Lipa? Paul Engin uh let's see and i got queen don't stop me now either i got that too yeah Jenn Carney interesting. I got Dancing in the Dark, Bruce Springsteen. No Scrubs, TLC. Now that I would not have put on my list, but that was a good one. Watermelon Sugar, Harry Styles. Okay. Take Me Home Country Roads, Don Denver back in the 70s. So I got a lot broader mix than you do. I got Earth, Wind and Fire also from the 70s. I have that as well. Oh yeah. Paul Engin September yep Jenn Carney Yeah, I got, I mean a few more, but Dream's Fleetwood Mac. So we've got some overlap here. So what's interesting to know where their sources are. I'm gonna ask it. Paul Engin I was gonna ask, yeah. Jenn Carney All right, and Claude says, my own knowledge, no search, no external source. That playlist came entirely from what I know about these songs, artists, and why they work for road trips. Paul Engin Oh, that's interesting. So I said, where did you compile this list from? It's very similar to the one generated by Claude. And it says, um, I did, yeah, I didn't compile it from Claude or copy it from existing playlist. The similarities are unsurprising because there's a handful of songs that consistently show up when people ask for a summer, Oh, interesting. Jenn Carney Oh, that is interesting. Paul Engin The ones that are consistent are Mr. Blue Sky, Summer of 69, Take it Easy, Life is a Highway, Walking on Sunshine. Jenn Carney Hmm. So, um, Clark got a little snarky with me. He said, honestly, nothing on that list is obscure enough to need a fact check. Just trust me, dude. I know. It's your problem. I know. Paul Engin Please. Come on, Claude. All right. All right. Jenn Carney What I'm interested in, though, is can they export this playlist to something that's useful for me? Paul Engin Oh, yeah. Let's see. So is that what you're asking? Yeah. What should I ask it? Can you export this playlist to something I can import into my into Spotify or Apple Music? Jenn Carney Oh, that's a good one. While you're on Spotify, you should like and subscribe to the immersive lens. That's right. Thank you. Oh, wow. You're good at that. See, yeah, I've been listening to Dave. Paul Engin All right, let's see, it might tell me to add a plugin. Yes. Jenn Carney It's telling me to use a tool called Bot Lister or Tune My Music or, I don't know how you would say this, Sound IIZ for Apple Music and that it can export me a clean plain text list of songs that I could import into those tools. Paul Engin Okay. Yeah. And mine actually says, absolutely. I've created the playlist in two formats compatible with many playlist import tools. Yeah. Although they don't tell me. Oh, yep. And then sound is tune my music for your music. All right. So. Jenn Carney It's pointing us actually to the same plugin tools and giving us the same output. Paul Engin Yeah. And it looks like you can actually export this SS or CSV file. And then you can import that directly into Spotify. Huh, something I'd have to try. We'd have to keep asking it questions on how to do this, but this would be something interesting to play with a little bit more as far as that goes. All right. So let's do one more prompt. And then I'm thinking, why don't we do the spontaneous this prompt and then kind of go through some best practices as if you're planning using AI to plan a future trip. So the one that we're going to use here is going to be called planning spontaneous long weekend trip for two people with no destination in mind. Budget is reasonable, but not extravagant. Keep it around five hundred dollars. We want interesting food, something unexpected and minimal time in a car, starting from Canada, where the podcast studio is located. Give us three completely different options and make a case for each one. This seems like a pretty good prompt. It seems to have some specifics. It also has things that we want to stay away from. I think this will give us a pretty good result. Jenn Carney And just to give credit where credit is due, that was a prompt generated from Claude itself that we edited a little bit. Paul Engin it. Speaker 3 Oh, yours is already. Well, I started early because. Jenn Carney is so much faster and once again, so much faster in its output. Claude is still thinking. Paul Engin I got images, I got everything, mine's done. Jenn Carney Now you and I are actually hooked into two different Wi-Fi networks here in this room. So I wonder if there's a difference in the speed of our Wi-Fi or if it really is its thinking. But I now have three options. So Claude has produced three options with actually quite a nice looking output with maps and itineraries with a specific picture. So option one is Hudson, New York, the antique town on the rails. Option two is Toronto, the city that eats the world. And option three is Pittsburgh, the underdog city that'll surprise you. And then there's detailed lists of what to do in each one. Paul Engin I got my option one is train adventure to Buffalo. Jenn Carney And what is more exciting than Buffalo, New York? That's right. Paul Engin Instead of driving, you make it a journey by making it a train trip. Jenn Carney Well, I'm assuming this Hudson one is the train ride also since it mentioned the rails. Yeah. Yup. Using, uh, Amtrak, uh, Rochester to Hudson five hours, about $80 a person each way. It says. Paul Engin I think that's to make it interesting, make it different. Yeah. So, and then it gives some different things you can do there. It gives a rough price on the train tickets, the hotel, the food that it would cost and gives you a rough budget. My next one was Finger Lakes Wine Country without the winery focus. So that's interesting, because a lot of times we have great wines in the Finger Lakes. So there's a lot of wine tours, but something unconventional would be just walking downtown, do farm-to-table restaurants, something that's a little bit more unconventional. Downtown where? You know, let's see, what does it say? Oh, this is interesting. It's not specific. Oh, Canandaigua. Oh, okay. It is saying Canandaigua, Geneva. Geneva. Geneva. Jenn Carney That makes sense. Cause we did ask for a road trip from Canandaigua, but I'm not sure I counted as much of a road trip when it's 20 minutes down the road. Paul Engin I did say it's only about 30 minutes from Canandaigua and then it gives a list of things like running a kayak in, you know, Seneca Lake and it gives a rough cost. And then my third option, it's Pittsburgh. Really? Jenn Carney All right, so let's talk about the differences in Pittsburgh. So in Pittsburgh, it has me. Let's see. Oh, it has some fun things in Pittsburgh. It has us going to the strip district on Saturday morning for Pennsylvania, Macaroni, Holy's Fish Market, La Prima Espresso. It has us going to the Mattress Factory, which is a really unique and contemporary art museum. It has us riding the Duquesne incline, which is a classic touristy thing to do in Pittsburgh, where you get a great view of the city. Got to do that once. Yeah, yeah. So what does, what's your Pittsburgh itinerary include? Paul Engin Well, you know, what's interesting is you have a lot more detail around each of your elements and you have maps. I have small thumbnail images for each of the destinations, but I don't have a lot of detail. All it tells me is that eat your way through the strip district, explore restaurants and breweries in the Lawrenceville neighborhood, write historic incline and then visit Phipps Conservatory and Botanical Garden. But it doesn't give me there's links so I can find out more. But it's literally just a bullet list and it gives me a cost. Does yours have costs associated? Jenn Carney Yes, it does. So my cost says, for this option, if, because there's two people, if we drive and split the gas cost, it's about $60 round-trip and fuel. Budget hotel on the North Shore, $100 to $130 per night. I've never found hotels that cheap in Pittsburgh, and then food about $150. So it's assuming I'm going to be able to find a really cheap hotel. Paul Engin Yeah, mine seems a little more accurate. It's $220 for hotel, $180 for food, $70 for gas, and then parking and ride share, $30. So I have a total of around $500. Yeah, minus. Jenn Carney saying 450 to 500 is the total. So let's, do we have time for a follow-up? Let's say, let's choose Pittsburgh. Give me a more detailed itinerary. Let's see what it can do and how long it takes there. Paul Engin All right, so more detailed itinerary for the Pittsburgh trip. Oh, wow. It's starting to break out, um, like eight, 30 to 9 AM. Jenn Carney Wow. It got really detailed for you. Now, uh, now it's, um, leave Canada. Mine is still thinking. Yeah. Yeah. So don't use cloud. If you want speed, use chat GPT for speed again. Paul Engin We'll get, if we leave between 839, we'll get there at around two. We'll check into our hotel. Oh, actually this says it's gonna be about 140 a night. Jenn Carney I found you a cheaper hotel, huh? Yeah. Yeah. Paul Engin And then it says up for the afternoon. We're gonna head to the strip district have a big lunch and Then in the evening, we're gonna go to Point State Park Speaker 3 Mmm Paul Engin And then Jenn Carney Is there, there's a famous statue there, right? Does it mention it? I can't remember. Is it Einstein or something? I can't really, I know it doesn't say people from Pittsburgh or if the, you are listening, I'm sorry. Paul Engin And then it gives us a Saturday breakdown of what we could do in the morning, a lunch, afternoon, dinner, and then Sunday, grab breakfast, and then get ready to leave. Jenn Carney So, um, I gotta say, Claude's, um, it's a very detailed itinerary. It has picked out a hotel for me, uh, Comfort Inn North Shore. That's going to put me, um, uh, right across the Allegheny river from downtown on the Andy Warhol bridge and walking distance to the museum of the same name. Um, so it has me arriving on Friday at like 3 PM, checking in, walking the Warhol bridge and going to the Andy Warhol museum. Um, and then it, uh, has found a specific, uh, tapas restaurant for dinner. Um, and, uh, Saturday, it has me starting early 7 30 AM breakfast, um, in the strip district, um, followed by, uh, um, a morning at the strip district mattress factory museum, Duquesne incline, and then, um, Sunday. Oh, visiting Randy land. Have you heard of Randy land? No, Randy land. So if you are somebody who likes Atlas Obscura kind of things, like you want to see the quirky and the weird Randy land is the place to go. Um, so it's a particular artist who transformed his home with crazy things. Um, and, uh, and then it has us, uh, visiting the Heinz history center before driving home. Paul Engin Okay, so I wonder if yours is slightly more catered toward your previous prompts. Yeah, I mean, I noticed I like art, right? Jenn Carney It has leaned in towards art. Yeah, absolutely. Paul Engin Absolutely. And so when we're doing this, I think we'll go through our list of 10 things that we can look at. And the first one is give context about yourself. So obviously yours had a little bit more context. So mine did not because I have my memory off. But things like age preference, if you have physical limitations, so I can't walk far or I can't. And then your travel style. So, you know, one of the things that it notes is I hate touristy spots. So something like that. So this way it knows we're not going to hit the regular tourist. Jenn Carney Right. Yeah. Another best practice would be to set the hard constraints up front. So if you tell it things like the budget, the time window, the location, and non-negotiables right from the start, it's going to plan around those boundaries better than trying to add it in later. Paul Engin Yeah, which makes sense, because if you do have a budget, you know, you don't want to go extravagant and it'll help gear it toward that. Be specific about the goal is the other one. So I want to really have a fun day. I want to feel relaxed, a little adventurous. So this way it has some context as to things to look for for that. Jenn Carney Right, include what you don't want. So negative prompting can be really powerful. So if you tell it no chain restaurants, no more than 20 minutes driving between stops, then it's gonna eliminate you from having, it's gonna prevent you from having to eliminate those bad options later. It's gonna avoid them to begin with. Paul Engin And to kind of piggyback off that context about yourself, you also wanted to give it, you know, who else is coming because obviously that's a big factor if you have kids under a certain age or if everyone's over 21, that's a different trip. Jenn Carney Yeah. And as always remember that if you're using the memory, it's going to remember who's in your family and so only give it details you want it to know. Also a great, um, idea is to ask for options, not just a single plan. So, um, if you give it a request for specific contingencies, like give me a plan B in case it rains. Paul Engin Yeah. And I think that's really good. I think it's also good because if you ask it to give you one or two ideas, that second idea might be something more relevant or something that you'd be more interested in. So you have the choice. You're not asking it to give the choice for you. Request is scheduled with times. So this way you have something that's more realistic. So it's not giving you a list of 15 things to do in two hours. Jenn Carney Right, and then ask it to explain its trade-offs. Ask it why it chose this option over that option. And a lot of times that'll surface another different, better alternative, too. Paul Engin Yeah. And then from there you can iterate. So, you know, you don't have to start from scratch every time in the same chat. It'll remember what you're talking about. So you can keep iterating, saying, why don't we swap this lunch with this? Is that so realistic? And it has all that context from before as well. Jenn Carney Yeah. And then stress test it, throw in a curve ball after the plan is set and see how it adapts to it. Personally, when I've used AI to plan itineraries for me, it packs the itineraries. So I usually then have to respond and say, hey, let's cut this down to half the number of activities in the day. Yeah. Yeah. And maybe that's just it needing to know my personal pacing. Yes. And yeah. Paul Engin Just give me buffer time, right? Yeah, right. Yeah, absolutely Because it doesn't understand some of the things like you know I think it's those nuances if you have four kids that are under 10 Alright, well, that's gonna be a lot harder to like move from location a to be sure if you're with like if I'm with my Mom or something that movement is gonna be a lot slower to go through a museum or whatever Activity we're doing versus if it's just my wife and I and we're just like, you know So well, I think that's all the time we have for thank you so much for stepping in for Dave This is fun. Appreciate it. My name is Paul Engin Jenn Carney And I'm Jennifer Carney, filling in for Dave Ghidiu. If you enjoyed today's conversation, be sure to subscribe so you never miss an episode. Paul Engin and share it with a friend or colleague. Until next time, stay curious, stay connected, and thanks for looking through the immersive lens with us. Jenn Carney This episode was engineered by Hugh Laird. Paul Engin recorded at Finger Lakes Community College podcast studios located in beautiful Canandaigua, New York, in the heart of the Finger Lakes region, offering more than 55 degrees, certificates, micro credentials, and workforce training programs. Jenn Carney Thank you to Public Relations and Communications Marketing and the FLX AI Hub. Paul Engin eager to delve into passion, discover exciting and immersive opportunities at www.flcc.edu. Now, usually Dave messes me up here. Jenn Carney I know. Yeah, as part of our mission at FLCC, we're committed to making education accessible, innovative, and aligned with the needs of both students and employers. Paul Engin The views expressed in this podcast are those of the hosts and guests and do not necessarily reflect the official position of Finger Lakes Community College. Jenn Carney Music by Den from Pixabay. Paul Engin This is the immersive lens. Thanks again, Jen. Jenn Carney Thank you, Paul.
Listen