Paul and Dave explore the polarizing question of whether AI is a "dirty word" by examining its rapidly evolving role in creative media and academia. The discussion highlights major "frontier models," including OpenAI's Sora 2 for video generation, Microsoft Copilot for enterprise privacy, and Adobe Firefly, which is noted for its ethical, creator-friendly training data.
Central to the conversation is the mantra "co-create, don't abdicate," a reminder that while AI can assist in tasks like generating B-roll or "vibe coding," humans must remain responsible for the output to avoid "hallucinations" and navigate the current lack of copyright protection for AI-generated work. Paul and Dave also delve into the ethics of synthetic media, such as AI-generated actress Tilly Norwood, and the potential for AI to cause harm to young users if proper guardrails are not in place. Ultimately, they advocate for a "fundamentals first" approach in education, suggesting that students should master foundational concepts before using AI as a Socratic "learning mode" tool to enhance higher-level contextual thinking.
Key Topics
Frontier AI Models and Their Specialized Functions: The sources categorize various AI platforms, noting that Microsoft Copilot is favored for enterprise privacy while Anthropic Claude excels in "vibe coding" and authentic conversation. Other tools like Google Gemini prioritize app integration, whereas Adobe Firefly focuses on being "creator-friendly" by using ethically sourced training data.Ethical Best Practices and Legal Boundaries: The hosts emphasize the mantra "co-create, don't abdicate," reminding users to remain responsible for AI outputs because the technology can occasionally "hallucinate" or make up facts. They also warn that AI-generated work currently lacks copyright protection and advise against uploading personally identifiable information (PII) to these platforms.
AI's Strategic Integration into Education: In the classroom, educators suggest a "fundamentals first" approach where students master foundational concepts like loops and arrays before being allowed to use AI to enhance their work later in the term. One innovative assessment method discussed is the "Student Turing Test," where a student's understanding is measured by their ability to explain and defend their project in a live conversation.
Transcript
Click here to view transcript
Paul Engin: Welcome to the Immersive Lens, the podcast exploring the technologies reshaping how we live, work, and learn. From AI and virtual reality to creative media and design, we're diving into the tools and ideas shaping our connected world. Join us as we uncover the people and ideas driving the next wave of interactive experiences. This is the immersive lens. My name is Paul Engin, professor of new media at Fingerlakes Community College. Dave Ghidiu: And I'm Dave Ghidiu, professor of computing sciences here at FLCC. Today's episode we're going to be asking the question, is AI a dirty word? Paul Engin: And by AI, I assume you mean artificial intelligence. Dave Ghidiu: Yes. And you're you're saying, well, is it is it good thing? Is it a bad thing? Or is it double-edged sword? Paul Engin: Absolutely. I think it's a great question that I'm eager to get your point of view on. All right. Well, uh before we get going, what's been going on? Have you done anything interesting? Dave Ghidiu: Um actually a few weeks ago I went to uh dinner at Julia's here at uh FLCC. Paul Engin: What's dinner at Julio's? Dave Ghidiu: So uh the business department um specifically to culinary and hospitality, they uh they put together a five course meal and um I believe it's their capstone. So students um will work the front end like getting everything ready, serving and then um the back end as well uh for all obviously cooking all the all the foods and then um they have professors uh Patrick Ray and Jamie Roder in the in the back of the house. Is that the right term? Paul Engin: Yeah, that's the right term. I worked at Algart for six weeks. I I know these terms. Dave Ghidiu: Um you know just uh managing the kitchen and then Paula Knight is on the front of the house. Um, you know, just managing to make sure the tables are set, make sure that, um, they're coming and, you know, doing all the waters and all everything you would expect in the restaurant. It's pretty amazing. And, uh, I can tell you, they have the best bread. It's amazing. Paul Engin: Hey, you keep raving about the bread and the dip. Dave Ghidiu: It is. It I'm telling you, I had like awful helping. Paul Engin: Is it every Friday night? Dave Ghidiu: Um, I believe it is uh in the fall, but you can go to www.flcc.edu/rest. Um, and you can, you know, read a little bit more about it and, um, you can also make reservations at jiulia at flcccc.edu. Um, so if they want to find out more, they can go to the website and explore. And yeah, it's a great great event. Definitely would recommend it. Paul Engin: And if you do go, let us know how you like the dip in the bread. Dave Ghidiu: Yeah. Right. Oh, the other thing that they do is they serve um the wine that is um— Paul Engin: Oh, from the vidic culture program here. Dave Ghidiu: Yeah, absolutely. Paul Engin: Okay. Dave Ghidiu: Yeah, it's amazing. And uh the wine actually has um not only is it like uh created here, but the um the labels that are on the wine are created by graphic design students here as well. Paul Engin: I love just the collaboration between all the departments here. Dave Ghidiu: Yeah, it's and it's really good. Now, you have to buy the wine separate. They don't serve it. So, it's a I don't know. Paul Engin: I'm sold. I'll do it. And uh fun fact, Do you know the the the grapes for the viticulture program are grown at Anthony Road Vineyard which is on Sonica Lake and Ann Martini who's uh Ann and John are the co-owners of of that winery. She's an alum of of the college. Dave Ghidiu: Oh, that's awesome. Paul Engin: Yeah, man. Dave Ghidiu: It all comes back to FLCC. Paul Engin: I know. How do you know these things? This is like— Dave Ghidiu: Our our reach here at the college is far and wide and durable. Paul Engin: Yes. Uh so, have you been up to anything? Dave Ghidiu: Yeah, we we hosted Todd Mccle who's a national uh trainer in AI and he came in for National Manufacturing Month of October. He came in and worked with probably about 30 or 40 companies we had in here. We had a full day and he did a presentation and some workshoppy stuff and we had a great lunch which we didn't have dip and bread but Jenny who works in the the dining facilities here made the absolute best muffins I've ever had. They were these s'more muffins. Um so the event was great but it the muffins— Paul Engin: You can't go wrong with s'more muffins. Right. Dave Ghidiu: And Johnny is the absolute best at doing that. I I went I tracked her down the next day. I was like, "This is the best muffin I've ever had." So, this is a podcast about technology and food, apparently. Paul Engin: Yes, we're foodies, I guess. Uh so, uh any hot topics that happening that you're aware of? Dave Ghidiu: Oh, like hot takes? Paul Engin: Yeah. Dave Ghidiu: Things that uh I don't know. Do you want to talk about Sora 2 and what that is? Paul Engin: Yeah. So, um OpenAI just had their company... yeah, the Jet GBT. Um, they had their dev day and they just released uh a few things. Sora 2 is their new video model. Um, it's pretty amazing. They before this uh when you would type in in a prompt, you know, um uh something like I'm trying to think of something like a a bear dancing to music, let's say. Um it would just have the video, but there would be no audio. Dave Ghidiu: So, it would do video. Okay. They've been able to do video. Paul Engin: They've been able to do video. Um then it's there's some issues with nuances, but um now it has music. Um I know I did a um which leads to this other product that they just released, which was Cameo, which is a um social media platform similar to Tik Tok or Instagram, but it's all AI. Dave Ghidiu: So Sora 2 is used to create the videos, correct? And then then you're saying that there's a social media app that can take those videos and put them into the social media feed. Paul Engin: Yes. And the app basically you prompt it and it will the prompt is Sora 2. So it plays, you know, it creates the video through their new platform. Dave Ghidiu: And Sora 2 I I can't do that right from chat GPT. Correct. Paul Engin: No, it has to go into um uh the Sora 2 platform. Dave Ghidiu: Okay. Paul Engin: But um it's really it's amazing because uh I did a you know cute little Otter skateboarding with an FLCC shirt and uh in a parade and uh you see the otter, you know, skating and waving and going, "Hey." And you hear music in the background and um so it does like ambient noise as well as like dialogue. Dave Ghidiu: Yes, absolutely. Yep. It it tries to read the prompt and figure out what to do with it. But— Paul Engin: And I'm assuming other than otter and skateboarding and and you might be able to have some practical uses as well. It's not just for entertainment. Dave Ghidiu: I—so for cameo, I think it's more just entertainment. Sora 2, I think it'll be beneficial for like, you know, and we'll we'll get more into this later, but I think it might be really cool to do B-roll in it. I think there's some other things. Paul Engin: Uh so in video production... Dave Ghidiu: I'm a math nerd. I I don't know your terms. Paul Engin: So in video productions, if there's a uh an interview, let's say, and the uh person is talking about um how they used to walk on the beach and you know, That's part of the story. You might not have beach footage, but now you can prompt uh beach footage, you know, with subtle waves as the sun is setting. And so as he's doing his uh story about him walking on the, you know, the beach, you you have that. Dave Ghidiu: So that just adds some texture and contrast to the video. So it's not just like a video of someone talking the whole time. Paul Engin: Yes. That's called B-roll. Dave Ghidiu: Is B-roll dead? Has has Soral put B-roll out of business? Paul Engin: No, no. B-roll is a general term. So, we can always shoot B-roll or now we can generate B-roll. Dave Ghidiu: Generate it on the fly. That's cool. Paul Engin: Yeah. How about you? Anything that's uh come up that— Dave Ghidiu: Yeah. There's this there's this thing, this is this has nothing to do with AI, but I think it's pretty fascinating. And it's called vertical soap operas. So, when you think about social media apps like Tik Tok, YouTube Shorts, and Instagram Reels, they're they're vertical. Paul Engin: And that's kind of changing the way consuming things and you have to keep scrolling. Dave Ghidiu: So when you say vertical, you mean like instead of like the regular TV which is the horizontal format, correct? Paul Engin: Yeah. Okay. Dave Ghidiu: So it's like holding your phone upright. You don't have to turn it to the side to watch the video. Got it. And so that's this new video format. And there's companies out there and you wouldn't know it because they don't escape their apps. They they don't tend to go into you don't encounter them naturally in Tik Tok or but there's this this huge business where there's these that are making vertical soap operas and and each episode is, you know, between 60 and 90 seconds because that's the typical length of of of a video. That's a vertical video and they tell stories and they have actors and they it's it's just really well produced. Uh, and it's a little cheesy. I think one of the metrics when uh I I did a deep dive into these. One of the metrics that they use for success is how many slaps there are per episode. Uh, so just to give you like, you know, like slapping someone across the face because it's a soap opera, so there's emotions um and like deceit and dead people coming back. It's just this this really wild thing and these apps are making a killing because they're charging I don't know like $20 a month and you get access to I don't know three, four, five different soap operas. It's just really really fascinating. It's it's something to keep your eye on. And it got me thinking, well, I wonder if I could parlay that into teaching. Like could I make vertical soap operas or at least vertical videos for when I teach just to keep people engaged? And and just yesterday The New York Times, the New York Times app on my phone now has vertical video. Like I can consume my news through the vertical videos. Paul Engin: Vertical. So is it converted? Do you know? Is it like just taking the— Dave Ghidiu: Created for it? Paul Engin: Oh, see that's good because a lot of times when they like convert it, there's a lot of content that's cropped out. If there's two people in a scene and it's wide screen, then there's only one person. Dave Ghidiu: Right. Exactly. Paul Engin: Um that's really cool. I think that there's a lot that you could do with that and it's funny because that's most people are consuming through their mobile device. So that seems to be the platform that things are moving toward. Um you know the other thing that's interesting that came out with uh OpenAI dev day which is again the chat GPT. Dave Ghidiu: So dev day is just like when they get on like the TV and they're like we're doing this product, we're doing this product this essentially a product launch for for them. Um but it's for their developers. So people who can program or work with whatever their um you know— Paul Engin: Sure. pushing. Dave Ghidiu: Sure. Sure. Sure. Paul Engin: Um, but one of the things I got excited about was, uh, they are integrating apps into their prompt. So, um, let's say I wanted to—are you familiar with Canva? Dave Ghidiu: Yeah, Canva is the design software. You can go and be like, "Make me a banner, make me a flyer, or you can do like PowerPoint type presentations." Paul Engin: Yeah. So, you can go on and it's all web-based and you can um— Dave Ghidiu: It's fantastic software. Paul Engin: Yeah. So, you can use templates and you can just build whatever you want um, design-wise. Well, now it's integrated into chat GPT. So now I can type like can—like I can have a prompt that says uh write a give me a bullet list of the five uh most interesting topics of the day and it would give it and I'd say use Canva to create a um infographic uh with regarding those five topics and it will open up Canva pop it in and give me four options and then I'll be able to click on it and it'll launch it in Canva so I can continue to edit it in Canva if I wanted to. So, I think that that was really cool. Dave Ghidiu: Yeah. So, chat GPT is becoming almost like an operating system. Like, you'll go there to start your work and then just finish it off in other places. Paul Engin: You could. Yep. As long as it's got a web connection. There's uh obviously Canva's just one of the apps. But— Dave Ghidiu: Like I imagine Spotify, they'll be able to like understand the context of what you're doing and play like an appropriate playlist like right within. Paul Engin: I could see that. Dave Ghidiu: Yeah, I could see that. Your fun fact about uh Canva, it was created by Melanie Perkins who was a teacher and was making yearbooks and she needed better software to make yearbooks and created Canva and that's how it came to be. Paul Engin: That's how it came to be. Dave Ghidiu: Oh man, I tell you a a lot of people love Canva and it is a really great tool. Paul Engin: It is fantastic tool. Um but you know what was interesting when this came out I I was trying to implement it um in chat GPT with a prompt. Dave Ghidiu: Yeah. Paul Engin: And uh I said you know I gave it the prompt you know in Canva do this. And at the time it wasn't implemented yet because they just announced it and rolled out to you. Dave Ghidiu: Yeah. Paul Engin: So, I got a I got a violation policy that says you're violating the chap... yeah. And I was like, "What?" And I so, then I asked it. I said, "But you just released this today." And then it actually went and it goes, "Oh, you're right. We did just release it today, but it's not implemented." Dave Ghidiu: Okay. Paul Engin: But it was funny cuz then I said, "Well, can we try it now?" And they said, "It's a violate." Like it reverted back. It was like endless loop. Dave Ghidiu: That's how AI works, folks. It's the smart thing you'll ever encounter. Paul Engin: But uh yeah, since then the good thing is uh it's actually been implemented. So now it's actually working correctly. Dave Ghidiu: That's good. So did did that make you ask is AI a dirty word? Did it frustrate you? Paul Engin: That's right. And that's a good segue. Dave Ghidiu: That's how this podcast episode came to fruition. Paul Engin: That's right. Uh so let me ask you, what do you think? Is AI a dirty word? Dave Ghidiu: I think like any other tool, it can be. It's got certainly there's a lot of benefit. For instance, not a lot. Here's another fact and I I know know nothing about nothing. I don't know why I have all these facts. Paul Engin: You're just asking the right questions. Dave Ghidiu: Say AI won uh the Nobel Prize for chemistry last year. Paul Engin: Really? Dave Ghidiu: Yeah. Uh for protein folding, um deep... AlphaFold, which was a deep mind product. Deep mind is an AI company. They had cracked the code on protein folding, which is something that's way too niche and nerdy to talk about in this podcast, but AI, there were three people that got the the Nobel uh prize for that. And one of them was a chemist and the other two were the um deep mind workers created alpha fold to do that. So— Paul Engin: Oh my gosh. Dave Ghidiu: It's good for that, right? Paul Engin: Hey, so there's there's benefits to that. Dave Ghidiu: It's a good word. It's not a dirty word. Paul Engin: So, um before we get into this a little in in more detail, um maybe we can talk a little bit about what are the main players in this space of artificial intelligence. Um— Dave Ghidiu: Do you want to go first? Paul Engin: Yeah. So, and in the biz, we call these frontier models, but you and I and everyone else in the world probably just calls them like AI chatbots or AI tools or just AI. Dave Ghidiu: I Yeah, there's their platforms. Yeah. Paul Engin: Yeah. So, I think probably the one that people know the most is chat GTTP. That was the first— Dave Ghidiu: chat GPT. Yeah. Paul Engin: Chat GPT. I'm sorry. Did I say GTP? GT— Dave Ghidiu: GTP. Yeah. Embarrassing. Chat GPT. It's an alphabetical order. That's how you remember it. GPT. Paul Engin: And that was the first to market. And that's when we think of AI typically that's what we think of. You can type in prompts and generate images and you can generate apparently videos now. Dave Ghidiu: Yeah. Paul Engin: Uh— Dave Ghidiu: Pretty robust videos. Paul Engin: Yeah. So, I I I think that's been first to market. People tend to rely on that the most just because it it was around for so long. It came out November of 2022. And they're I think it's great. Dave Ghidiu: Yeah. They've been really pushing they they are really... I mean this is what they do whereas the other companies we're about to talk about they this is a derivative but it's becoming a focus of what they do. Paul Engin: Yeah. Dave Ghidiu: So, um I think the the one that uh I was I've been hearing around uh the schools are going to be more like Microsoft Copilot. So um because many schools and uh actually companies in general already have Microsoft integrated into their um you know— Paul Engin: Their ecosystem. Dave Ghidiu: Their ecosystem. Um co-pilot is uh you know just something that they feel comfortable with and it's slowly getting more and more integrated into all of their applic— Paul Engin: Yeah. And Microsoft is is real nice. It's it's good software. The the nice thing about Copilot, especially when you're a big company or schools, you have this like institution so that sometimes, most of the time, when you type your prompts into AI, then that goes to those companies and they can train their data on what your conversations are. And Microsoft doesn't do that if you have that institutional licensing. Dave Ghidiu: So, it's locked to just that enterprise. Paul Engin: You get a little bit more privacy. Security. Dave Ghidiu: That's nice. And I bet you the IT departments love that. Paul Engin: Yeah. Yes. Uh— Dave Ghidiu: Less integration. Paul Engin: The the other half the school. So about 50% of schools maybe just a hair more are using Google as opposed to Microsoft. And Google has an AI service called Gemini. Dave Ghidiu: Mhm. Paul Engin: And you can download the app for you can download the apps for all these on your phone or your computer. Google Gemini is equivalent to chat GPT and Microsoft copilot like it has images that you can derive from text and you can do text to video. But Google also the Gemini AI lives in all the other software packages. So if you're in Google Slides, you can generate an image right from Google Slides. You don't need to go to GPT, download an image, you can do it right there in the slides. So they thrive on their integration. Dave Ghidiu: Gotcha. Yeah. And I've I've seen that. Um there's a lot of different flavors of Gemini. It seems like there's uh Gemini for developers there. They have Gemini has a video model as well, right? Paul Engin: V3 VO. Dave Ghidiu: Veo. Oh, okay. Paul Engin: And actually they just released 3.1 last night. Dave Ghidiu: Did they really? It's awesome. Paul Engin: And it does the same thing as Sora. It does tax and ambient noise and conversations. Dave Ghidiu: Oh, nice. Yeah. I mean, these things are evolving so quick. I'm sure when this when this is released, they'll be at like uh you know, V5 and yeah, Sora 9, I don't know. Um but, uh as far as Google goes, uh is it closed like co-pilot for enterprise or is it... I don't know how you know you said that some of the schools have. Paul Engin: Yeah, I think that's some something the administrator of your uh your instance would have to work with. Dave Ghidiu: Okay, I gotcha. Gotcha. And then um there is uh Meta Llama, right? Paul Engin: Yes. And Meta is Facebook. Dave Ghidiu: Facebook. Paul Engin: Facebook. Dave Ghidiu: So there's their version. Paul Engin: Yeah. So um they have an online, but I think the nice thing about Llama is that they're trying to keep it open source so people could actually download it and if they have a robust system, they can run it locally. So— Dave Ghidiu: Yeah. And and it's not as hard as it sounds, but it's not like at the press of a button. But one thing I really like about Meta is they are... they have built especially their image generation to creators like it the tool set for if you want to create an image it's not just text it it gives you a menu of options like uh for the size and do you want to be claimation do you want to be like paper craft but it also allows you to adjust uh what we call the temperature and that's like how creative or imaginative it is. So you can actually... yeah it's got some really nice development toys uh that anyone can use. And it's very simple to use just by uh going to Meta and creating. Paul Engin: And so that one... so they they have it both online and in and a downloadable. Dave Ghidiu: Yeah. And Meta is uh the only one we're talking about that doesn't even have a paid tier. Like you can just do things. All the other ones are free, but they have a paid tier to do more, but Meta doesn't even have one. Paul Engin: Oh, that's nice. Dave Ghidiu: You just sell your soul if you work with Facebook and Meta. Uh and then there's Anthropic Claude, right? That's the— Okay. And what what are your thoughts on Claude? Paul Engin: I think Claude excels in two different arenas. One is if you're doing any type of programming, Claude is the way to go. It's better than chat GPTs. It's better than Gemini. In fact, we have a course here in the spring. It's called... yeah, we call it vibe coding. That's what it's called. When you just use AI to code, you don't need to know a lick of programming. You just use natural language. Dave Ghidiu: And we'll be using Claude for that. Paul Engin: That's very cool. So, you're going to... so, this is anybody can take this course. You don't have to be a programmer. You don't need to be computer science. Dave Ghidiu: Correct. In fact, the only people who can't really take it are computer science people. Paul Engin: They... so this will be a good elect... and I think Claude is also the most conversational and it seems the most authentic. Sometimes when you're working with chat GPT or Gemini seems a little stiff or a little hollow the responses. Dave Ghidiu: You know I agree and I know this is really cheesy but it's nice when you open up Claude and it says hello Paul. Good morning Paul. Have you had your coffee today Paul? Paul Engin: Oh really? Dave Ghidiu: Like it's very like like warm and inviting versus like... you just go to these other prompts and they're just in your face and you're like, "Okay, I guess I got to type something." Paul Engin: Yeah. And I never thought that chat GPT or Gemini were transactional until I started using Claude. I was like, "Oh, wow. This is like way more intense of a conversation and it's way more in-depth." Dave Ghidiu: Yeah. And it seems to produce really... I I'm I've been really happy with the code that um like for programming what Claude can produce. Um— Paul Engin: Yeah. Dave Ghidiu: And there's one more we want to talk about and I don't know anything about it, but I know that Adobe is in your wheelhouse. So, can you tell me about Adobe Firefly? Paul Engin: Yeah. So, um Adobe Firefly is a creator friendly um AI. And what I mean by that is that they do not um learn or train on anyone's copywritten material and they don't use anything you do for their training. So, you can feel confident that if you produce something with Firefly that um it's it's yours and you know, you retain it. Dave Ghidiu: That's an interesting approach because all these other models are taking like every piece of artwork ever in the world and for their training, but Adobe is doing it a little bit more ethically, would you say? Paul Engin: I I think so. And I think that now they do have the option where you can select one of the other models. Dave Ghidiu: Okay. Paul Engin: But um if you use your the the standard Adobe Firefly model, then then you're safe is is the way I equate it to as far as content creation. Dave Ghidiu: Yeah. And that actually reminds me there's kind of like three things that we should probably say about AI. One is any output from AI cannot be copywritten. Uh that's the current law right now. Who knows what's going to happen by the time hear this. Uh there is some some exigent circumstance. So if you have substantial human intervention then you might be able to copyright it. One of the other things and I I think I just say this to everyone talk about AI is never upload personally identifiable information like social security numbers, driver's licenses, anything that you don't want the world to see because you never know when it's going to escape. Paul Engin: Yes. And that means documents that have that information on it too, right? I mean you don't want to um be have that up in the ether. So people can just uh you know ask ask those questions. Dave Ghidiu: Yeah. Um and the the last advice I like to give is that and we've coined the term here FLCC co-create don't abdicate which means you can use AI to help you as a thought partner but you need to review whatever the final whatever the final output is before you publish it or before you use it. You need you need to sign off on it. You need to okay it read it because there's been so many case in fact this summer the Chicago Tribune I think it was a Chicago Tribune made a summer reading book list and whoever did it just typed into AI and published it and didn't read it because it made up books. Like there were a lot of stuff that that didn't exist. Paul Engin: No, it's super embarrassing. So, you have like reputational harm that can come from uh abdicating to AI instead of co-creating. Dave Ghidiu: You know, that's interesting because that's equivalent to like what what happened with Wikipedia. It's not a reputable resource or it wasn't. It is now cuz I think you can lock some things down. Um but for a while it wasn't. And uh so I think with a lot of these things you just want to make sure that the content is accurate and you want to do some additional research outside of just prompting, right? Paul Engin: Yeah. Um, so I I think that's an interesting point. You know, I think that when we look at AI, there's there's for creatives, it's a slippery slope as far as copyright goes. What is uh something that people are okay with, you know? Um, I just saw something on this uh virtual actor called Tilly Norwood has... Dave Ghidiu: Yeah. So, tell tell us about Tilly Norwood. This is a fan... this is like such a fantastic story. I thought it was made up the first time I heard it. Paul Engin: Yeah. So, uh if you Tilly Norwood is a fully AI generated actress and um it's run by a agency. So, the agency it's I think it's group six. Dave Ghidiu: Particle six. Paul Engin: Particle six. Thank you. Um they they created this and uh this fictitious actress um and technically, first of all, I want to say it's amazing that they can keep the consistency of this character. I think that's one of the things that we we'll talk about is that's been a big challenge. Right. Dave Ghidiu: Right. So, like you can have one person and that person can be the same in all of your prompts. Paul Engin: Yeah, that was very frustrating when I was playing with like Sora 2 in in Google VO. I would generate I wanted to do a whole story cuz they only generate 8-second clips and I would want the same person to be in it, but each time the person would be a little bit different. I could never get that consistency. And so you're saying that they've... Particle 6 has figured that out with Tilly? Dave Ghidiu: Yeah. I mean, they have uh her in different scenes and different sequences and it's and it she looks the same. Paul Engin: And the the thing that was hard for me to wrap my head around is that it's not CGI. Tilly is not a CGI person that we can give lines to and then animate like typical animation. Dave Ghidiu: Correct. Correct. So, you're going to prompt her and she's going to... you can prompt her behavior. I mean, I don't know the behind the scenes of how they're doing her creation. But it seems like you can um you know, you can prompt her and and you can give her directions and tell her what to say and then she'll act and you know, you can give her an accent if you want. But— Paul Engin: Can you give her notes? Like after a take, can you give her notes and— Dave Ghidiu: Probably. I don't know. I mean that's what the AI part is where there's like give and take. Paul Engin: Right. Right. So I think that the other thing about this is that we want to keep in mind uh you know there's you know the good and bad of this. Right. So I I think that when we look at this, I think there's uh I know SAG AFTRA is— Dave Ghidiu: That's the union for actors. Paul Engin: Yeah. They're 100% against this. Dave Ghidiu: Sure. Paul Engin: Um they don't believe in synth... this synthetic creation. Um and you know I I agree to... I agree on some extent but I wonder if um this can fall in a different genre maybe or um you know there's a lot of like uh digital doubles uh which is like if a stunt actor can't do a performance or like you know Marvel like people don't fly, right? Dave Ghidiu: Yeah. Paul Engin: Uh so uh they create digital creations of those— Dave Ghidiu: Like scanning the whole body. Paul Engin: Yes. A 3D scan and it's not AI. It's all the actor knows what's happening and maybe that's the distinction is the actor is giving consent, you know, to that. But um it's interesting because they do have... I wonder what these actors sign off on their digital like... cuz technically once that's integrated into an AI system now— Dave Ghidiu: Yeah, it's 3D. So now technically they could prompt in the future. They'll be able to prompt these these real life actors possibly. Paul Engin: That's wild. Dave Ghidiu: I don't know. But um it'll be an interesting area that I think we should look into more in the future. Um and uh I know you pointed me to— Paul Engin: Critters. Dave Ghidiu: Yeah, tell us about Critters. Paul Engin: So Critters is a uh creation that was all AI. It's all animated. It's a um documentary that they did of a jungle critters and uh you know somebody's like we're in the woods of such and such and we come across and then the spider starts talking and there's a... it was really well done. Um but it was a lot of creation still. So it wasn't just prompts. Dave Ghidiu: Sure. Because it was it was scripted to an extent, right? Paul Engin: Yes, it was scripted. Um you could see that there were small sequences but uh this uh agency worked directly with OpenAI or ChatGPT um to help develop their software and uh you know create how creators could use the software. But I want to say that this was really a great experiment in how AI can be integrated into a creative process from storyboarding to character creation. So people like sketched out the the creatures. Dave Ghidiu: Sure. Paul Engin: And then they put it into ChatGPT and then they said give us variations with very cute, but with the sketch and— Dave Ghidiu: So you just painted a portrait of I think how people should optimally be using AI and still have that human element and so like I might not ever be able to make a CGI spider and make one talk, but I could work with AI to do it, but it would still be my vision and I'm still overseeing the whole operation. Paul Engin: Yes. Okay. Exactly. Dave Ghidiu: Yeah, it's really cool. Paul Engin: So, um, let me ask you, where do you think AI a dirty word in education? Is it... I mean, what's your thought? Dave Ghidiu: Well, it depends. I mean, even today, I just read an opinion in the New York Times about how AI is making learners dumber, basically. And and I don't know. I feel like we've been on this treadmill before with calculators and with Wikipedia and with computers in the classroom and different devices, right? I think that it's incumbent upon us as educators to teach people how to use AI. And I don't want to use the word responsibly, but I want to say optimally. So, how can you maximize your learning? It should never be it should... you should never use AI to deprive yourself of an opportunity to learn. And as educators, we need to create the space in the classroom for them to learn. That being said, there, you know, the trope in AI and education is well, here's my essay prompt. I'm just going to dump this into a chat GPT and then whatever it spits out, I'm going to submit as my essay, right? And and and that happens a lot. Uh and and there's a lot of reasons for that. And there's Perplexity, which is one of the... we we actually didn't talk about Perplexity. That's one of the models that fits in with like say chat GPT and uh Gemini. Perplexity has this kind of... kind of helps you do your homework for you where uh you can just like upload your homework or you can like open up in their browser because they have their own browser and it will solve the problems for you or write the essays for you. So yes, there is a path of least resistance and it's easier now than it has been in the past. Although the research is pretty clear of right now, the research shows that the number of instances of cheating hasn't increased. It's just the way people are cheating is kind of converging to AI. So it used to be like in college, you know, like fraternities and sorities would have file papers of like all the old tests or you'd pay someone to write an essay or you could go to like check.com and now everyone's just like I'll just use chat GPT to do it, right? Paul Engin: Do you remember cliff notes? Was that cheating or cliff—like you didn't have to read the entire uh Romeo and Juliet? You could just get the cliff notes. Dave Ghidiu: I mean, you're still reading and you're getting the the main thing. I I think what... at the end of the day, here's here's the litmus test. Are you familiar with the Turing test? Do you know what the Turing test is? Paul Engin: Can you give me more details? I'm familiar with it, but— Dave Ghidiu: Sure. So, when the conversation of AI started back in the 60s, the the posit was if you were sitting in a room and there there was someone else on the other side of the room and you were writing communication down and sliding under a slit and then getting it back and sliding under slit and getting it back, could you tell if it was a robot or a human? And so, when we got that was called the Turing test because if we got to a point where you could not tell that the response was human or AI, then we have... then we have AI. And I think that we could adapt at the classroom at some point and have what I call the student touring test. So let's say that you're teaching a class, Paul, you're teaching a new media class. And at the end of the class, maybe it's end of a... maybe it's an exam, maybe it's a project, probably a project because you do an awful lot of projects in the course. Paul Engin: Yeah. Dave Ghidiu: You sit me down and just have a conversation with me about like what did I learn and how did I do this and like oh Dave, I really liked how you did this this fade. How did you do that? And if I can answer those questions and have that coherent, do you care if I use AI to do it? Do you care if I use cliffnotes to do it? Or do you just care that I apprehended that knowledge? Paul Engin: Right. And I I think that's a good point. I I would care that you you understood it. So I think that that was that's a great that's a great point. I think that on the flip side of that, sure, the issue is that, you know, they're not picking it up because they just say, "Well, I'll always have AI, right? So I can always prompt." That's what that's a common thing that I get. Dave Ghidiu: It is. And I I mean I have mixed feelings. This is like a double-edged sword, but there were teachers when I was growing up and I was probably one of them cuz I was a math teacher in a previous life uh or previous career who, you know, learners would be like, "Why do you need to do this?" I'd be like, "Well, you won't always have a calculator in your pocket." And then we all have calculators. We all have phones, right? So, so I'm not so sure. I went to a presentation a few years ago. This is before AI. And someone was doing this presentation. She was from the uh board of education from a school, Husk Falls, which is out in uh almost at the Massachusetts border. And she said, you know, 20 years ago, you used to be smart if you could name every state in their capital or you could name every president. She's like, that's not the bar anymore. Now the bar is like, can you contextually tell me like why this president had these policies given the economic conditions? Paul Engin: So everything is going to be elevated because now you have more support to do more. Dave Ghidiu: Yeah, you you can learn um you can like engage with things a little bit more like there are educators out there making chat bots. Say say you you found every single letter of that Abraham Lincoln ever wrote and every every everything he's written, you can create a chatbot that will have that conversation. So 10 years ago, all we knew about Abe Lincoln for the most part, I mean, some scholars would know this stuff, but would be, you know, what he was famous for. Um, but now you can actually have a conversation with him and you can find out how like his relationship with his wife. You can find out like what he was thinking when he was, you know, running for president, all these things. It's just a it's a it's a wild world. Paul Engin: Yeah. Um, and how about what what are your thoughts about um should schools or or institutions completely ban like have AI banned or like what what's your thought on that? Dave Ghidiu: Well, uh, and I don't know what the outro to this episode's going to be, but I'm sure it's going to have something like the views and expressions in this podcast reflect FLCC, but I I don't think that we should be banning uh any AI. We're you know, we're actually training learners to be functional workers and functional participants in this society, which is AI. It's a, you know, you can't walk five steps without hearing something about AI or seeing something about AI. So, if we put those AI blinders on, I don't think we're doing them any services. Paul Engin: Yeah. I think that learners are going to have to... and and everyone in general, right, has to figure out ways that they can utilize AI in whatever they're doing. I think it's, you know, I always equate this um you know, the internet when that came out, computers, um, you know, it was like, oh, it's going to take away jobs. It's going to do this and that, and it's like the jobs might be shifting, things might be changing, but um, I think it's how do you use the tools, right? So, I think that these are still tools. Um, these are powerful tools, but you know, we have the internet now and, uh, we do research on on online more than we do in the library. You know, I think like there's a different approach. Dave Ghidiu: Yeah, it really potentiates what you do. And you can ask if you ask a 100 people how they're using AI, they'll give you 110 different answers. Like it is different for every single person. I'm glad to see that people are finding their pathway and learning how to like potentiate what they do and optimize what they do with AI. So that that's our job as educators to help people figure out how to use it responsibly. Paul Engin: Yeah, absolutely. And I do think that, you know, if we want to look at the other side of this, um I think that there are some issues with AI. Uh I know that uh there is a uh legislation that's coming out about chat bots and uh I think that it's going to be... so chat bots basically are an AI system that act as somebody else. So young people under 18 are building relationships with these bots and are harming themselves. And so I think in that regards it is bad. And I think we need to put these rules around who uses those. Dave Ghidiu: And there were there were some pretty scary cases over within the last year about chat GPT specifically in theirs. Uh and they've made some course corrections and put guard rails. But I'm glad this conversation's h— I'm not glad because this is not a happy conversation to have. But I'm glad that that we're paying attention to it because social media which has been around for years... I mean there's been an awful lot of talk about social media for the last 10 years about how harmful it is and those companies haven't taken measures and some something bad happens with Chet PPT in the next day they they've already put some guardrails in. So, um I'm optimistic about this. Paul Engin: Yes. Dave Ghidiu: Uh and and I do want to say while we're talking about this, there are some environmental concerns about the energy consumption in the water. There's some labor concerns about using um people in countries in the global south to do all our filtering. That hasn't been so good. There's been a lot of um concern about, you know, we're paying developers, Meta, Facebook is literally paying $100 million for AI developers. So, there's an equity and wages. So there there are some bad sides to it and I think a great book if you want to read more about those uh is called The Empire of AI by Karen How. Uh it it's just this fascinating book that looks at the the the the dark side of AI. Paul Engin: That's good. And I think you need both perspectives. I think that it's important to understand that AI can be used, you know, in a negative way and it can be used in a positive way. I know um we'll get into it in a minute, but um there are some techniques that AI can help um someone learn or help with a situation and then um there's you know like you said people just putting in a prompt and just trying to get an essay out of it and you know that's that's no bueno right? Dave Ghidiu: No I mean at that point you're just like a cog in a machine. You're like I'm going to take this thing from over here put it in this other thing and then put it put the response in this third thing. Paul Engin: Yes. Exactly. Exactly. And um you know I think that uh as far as um training goes, I think that how would you see it? How like put it in how I could use AI to benefit me in something that I might not be familiar with? So like if I'm not familiar with uh um uh math or chemistry, how how might it I approach it? Dave Ghidiu: So if you're a learner in a class, say a chemistry class, most of these AI tools, chat, EPT and Gemini specifically, have a button that you press called learning mode. Paul Engin: Mhm. Dave Ghidiu: And so it that cues the AI to not give you the answers, not just reveal the answers, but use the Socratic method and and kind of like ask you questions and tease it out of you. So we'll ask you, well, do you know this thing? You're like, well, yeah, I know that. And then it'll be like, well, what do you think would happen if you did this other thing? And it'll be like, oh, and then so I think that's a great way. Paul Engin: It's like quickly creating uh note cards for you. And so you can— Dave Ghidiu: Well for it flashards, it can um they they can design flash cards, but they can also just have a conversation with you and get you to kind of reconsider what you know and then ask you to make predictions and then ask. So, it's like— Paul Engin: More conversational. Dave Ghidiu: Yeah. It's like if you if I was a learner in your class and I asked you a question, you'd be like, "Well, I'm not going to tell you the answer, but I'll point you in the right direction and I'll like ask you questions like leading questions." So, that that's that's one of the easier ways you can do it. The other way is uh Chad GPT um published something called I think like a hundred prompts 100 ways college students are using Chad GPT and some of those prompts in there are wild. Uh you can go to the website, you can Google it. It's it's it's free. You can um just look at them. And it's wild because they have like prompts that you might not even think about. So for instance, let's say I was writing an essay. I might write the essay and load it into chat GPT and say, "Please read this and ask me five questions that identify blind spots that I've missed." And then I'm becoming a better writer because I'm getting a more macroscopic view of a particular topic. Paul Engin: Oh, very cool. Dave Ghidiu: Yeah. So I I would recommend to every learner that they look at that and if if you're an educator, point all your learners to it. Uh and have webinars and seminars and just like tell people like you can here's 100 here's 100 free ways that you can be a responsible learner. Paul Engin: So 100 ways college students are using chat GPT. Is that what you would search? Dave Ghidiu: Yes, that what I would search. Paul Engin: All right. Very cool. And how are you using it in the classroom? Out of curiosity, are you using it yet? Are you integrating it? I know you mentioned vibe coding. Dave Ghidiu: Vibe coding. Uh CSC 295 2 o'clock to 320 and Thursdays in the spring in the spring semester. Uh yeah, right now I think a lot of people are doing it in awareness and just saying, you know, saying here's what you can do with AI in the class and make a list and here's what you can't do like I want to draw a line because there I think learners are confused because every educator has different rules for how you can engage. So I think you just need to be very clear with that and very and remind them and sometimes show people and sometimes learners don't even want to use AI. Uh and part of that anxiety because they don't want to get busted for like plagiarism or like student code of conduct issues. So you it really is important for us to kind of work with our learners to figure out the best pathway because the way your learners are going to use it is way different than the way my learners are going to use it. Paul Engin: Right. And I think that um I know I've had um conversations with some students and they they're just against using AI just for um for reasons of like power consumption and— Dave Ghidiu: And so I tried to give them alternatives. But I do tell them that, you know, this is something that is going to be... like it's like saying that I'm never going to use the internet— Paul Engin: Or Photoshop. Like anyone that's in your classes are going to be using AI, right? Exactly. And I think that um but at the same time I I'm against foundational um topics using AI like they need to understand the the foundation of coding to understand relationships and they can't just— Dave Ghidiu: Right off the bat, I know you're doing vibe coding and I know it's going to be different, but for— Paul Engin: Yeah. And actually, the way I arrived at that is when AI first came out, I was like, "No, you may not ab—absolutely may not use AI." And then over winter break, I was talking to some of my friends who they, you know, they do off... they're in software development and they're like, "When we hire people, we don't give a crap how they do it. We just want it done." So the next semester, I was like, "Everyone can use AI for everything." And then about 2-thirds of the way through the semester, I started getting all these emails and phone calls and it was on the same lab, which happened to be AI proof. Not it wasn't design that way. AI just wasn't there yet. And and I was like, "Oh, you just have to write a loop." And they're like, "Well, what's a loop?" I was like, "Oh, no. I've done something bad." And it be... it became clear that people were relying... over relying on it. So now where I'm at is uh I'll say to uh the learners, listen, for weeks zero, one through three, you may not use AI. And weeks four, four through six, I'm going to show you how you can use AI. Like if you're writing a program and you're stuck, you can upload the code or you can ask it questions. Yeah. Dave Ghidiu: And then once we get through the fundamentals like all the things like the loops, the arrays, those are all like the complicated things. But once you have that tool set for the last three or four weeks in the semester, I want to see what you can do because if if AI gets you into a hole that you can't get yourself out of because you haven't learned those fundamentals, well, that's bad. But now that you have those, you can you can now navigate if AI does something and you don't know what's going on, you can at least navigate it and interrogate it with your skill set. Paul Engin: Yes. Yep. I like that. And I I think that's a approach that I take too with my my programming. I think that for students that are are learning they use it... they actually use it as a learning tool. So they can ask— Dave Ghidiu: Sure. Paul Engin: They can ask questions regarding why how would I do a grid format for layout or whatever and it'll give them instructions but if they're just copying and pasting code just directly not understanding what's happening then they're missing the whole the whole point of it because they really do need to understand what's happening with the with the code. Dave Ghidiu: Yeah. So so it sounds like there's some responsible models like that are out there and and it really is kind of course dependent or future dependent. Paul Engin: Yeah, and I know we're just scratching the the topics on all this stuff. Um, but I think that's all the time we have today. Um, my name is Paul Engin and I'm Dave Ghidiu. If you enjoyed today's conversation, be sure to subscribe so you never miss an episode and share with a friend or colleague. Until next time, stay curious, stay connected, and thanks for looking through the immersive lens with us. Dave Ghidiu: This episode was engineered by by Jeff Kid or Hugh Laird. Wait, this episode... oh, this episode was engineered by Jeff Kid and Hugh Laird. Paul Engin: Yeah, that's right. They're both here. Recorded at Fingerlakes Community College podcast studios located in beautiful Canandaigua, New York, in the heart of Fingerlakes region. Offering more than 55 degrees, certificates, micro credentials, and workforce training programs. Thank you to public relations and communications, marketing, and the FLX AI hub. Eager to delve into passion, discover exciting, immersive opportunities, at www.flcc.edu. As part of our mission at FLCC, we are committed to making education accessible, innovative, and aligned with the needs of both students and employers. The views expressed in this podcast are those of the hosts, guests, and do not necessarily reflect the official position of Fingerlakes Community College. Music by Den from Pixabay. This is the immersive lens. Would have been okay. I was just going to say it would have been nice. Thanks, Jeff.
Listen