The Immersive Lens Podcast

Paul Engin | Dave Ghidiu | Jeff Kidd

Episode 16: Gemini

Cover art for podcast episode.

 

This episode explores the multifaceted world of Google Gemini, Google's rebranded and significantly advanced large language model. The hosts delve into Gemini's three primary modes—Fast, Thinking, and Pro—each tailored to different user needs, from quick queries to complex problem-solving. They also highlight Gemini's seamless integration into the Google ecosystem, including Google Docs, Gmail, and Chrome, making it a ubiquitous assistant for everyday tasks. Beyond simple chat, the discussion covers specialized tools like NotebookLM for research, AI Studio for developers, and creative powerhouses like VEO and flow for video and image generation.

Paul and Dave bring their unique tech-savvy perspectives to the table, discussing their personal experiences and preferences between Gemini and other AI models like ChatGPT and Claude. They emphasize the importance of Gemini's low barrier to entry and its "vertical integration" within the Google suite as major competitive advantages. The episode concludes with a positive outlook, portraying Gemini not just as a chatbot but as a comprehensive platform that is continuously expanding its capabilities and redefining human-AI interaction.




Key Topics

Gemini's Tiered Intelligence: Gemini offers three distinct modes: Fast for quick answers, Thinking for a balance of speed and depth, and Pro for complex reasoning and deep dives. This flexibility allows users to choose the right level of AI power for any given task.

Ubiquitous Integration: A standout feature of Gemini is its deep integration into the Google ecosystem, appearing in everyday tools like Gmail and Google Docs. This makes AI assistance a seamless part of existing workflows, rather than a separate, siloed experience.

Specialized Creative Power: Beyond text, Gemini powers a range of creative tools, including VEO for video generation, flow for frame-by-frame video editing, and Nano Banana for image creation. These tools, often accessible through Google Labs, represent the cutting edge of AI-driven media development.

The Rise of the AI Agent: The episode touches on the emerging field of "agentic AI" with tools like Opal, which allow users to create specialized AI agents for specific tasks, such as generating social media posts. This points toward a future where AI is not just a responder but a proactive participant in achieving user goals.





Transcript

Click here to view transcript
PSA Voiceover A simple pencil, a powerful tool for progress and learning.

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 design, we're diving into the tools and ideas shaping our connected world. My name is Paul Engin. Join us as we uncover the people and ideas driving the next wave of interactive experiences.

Dave Ghidiu And I'm Dave Ghidiu. This is, The Immersive Lens.

Paul Engin Hey Dave, uh, you look different today. What's...

Dave Ghidiu Do I look relaxed? Am I like...

Paul Engin Very relaxed.

Dave Ghidiu Yeah, I am, I am on location in Florida. Uh, yes, we're scoping out a second studio for the Immersive Lens.

Paul Engin Oh man, you're giving it away. Everyone knows now. They're going to take over Florida.

Dave Ghidiu Well, so, and I do want to share with people in case they didn't know that the iHeartRadio Podcast Awards were just held. And, uh, not only did we not win anything, we were not nominated for anything.

Paul Engin Oh, geez. Next year. Next year.

Dave Ghidiu Next year. But I do want to say that there is a category called the best ad read, and that's for when you read ads. And so I think that's what we are going to be going for next year.

Paul Engin Excellent. Excellent, excellent. Um, and then, uh, is there a... did you actually see any winners or anything?

Dave Ghidiu Uh, I don't want to talk about the winners. It's kind of like, it's too soon, Paul. Too soon.

Paul Engin Too soon? Okay. Alright, it hurts. Okay, I get it, I get it. Um, so, uh, what do you got?

Dave Ghidiu I just...

Paul Engin I was um, playing with um, a few things this week. Uh, one of the things that uh, I played with was called Hyperscape. It's um, on the Meta Quest 3 and uh, what it does is it allows you to uh, do a 3D scan of a um, of a room. And um, it creates basically a digital double um, of that uh, space. And then you can bring it into uh, right now it's for Meta Horizon um, which unfortunately uh, is going away. Uh, so... but they're going to...

Dave Ghidiu So Meta Horizon is the, the VR kind of gathering space for the, the Oculus.

Paul Engin Correct. It was something that they were building out like a place that everyone can convene and and, you know, you can have multiple people in one space and you could have shared spaces. Uh, but, the the... I showed you... did I... were you able to see a demo of it that I created our studio?

Dave Ghidiu Blew my mind. Because of the rich detail, and in, in that studio there's a lot of different protrusions and furniture and wires and like wow, it's fantastic. It captured them all.

Paul Engin Yeah, it was um... so good I, I uh let one of the students that are familiar with the studio use it. And they were like, oh my gosh, this is this, this is the studio. This is like... it was, they were, they were just blown away by how real it was.

Dave Ghidiu It really looked to me not like CGI or 3D, like this was a new technology. It looked like it was video to me. I couldn't tell that it wasn't.

Paul Engin Yeah, it um, it uses uh like photo uh... is it cloud clustering or photo... neural alert topology... I mean it's, it uses a certain technology where it's very dense and it uses a bunch of photos for the textures. So it... it... I don't know how usable it is outside of this creation because it's so heavy with, and heavy meaning um large file. It took eight hours to process, so it gives you a scope of um how much data is being created from that, you know. But it was, it it it was really cool to see that you could create something that realistic and um, after I did that I I read the press release and then I was like, oh my gosh, just as I got my hopes up, you know, I was like, ah. And the cool thing about this is I could invite like multiple people into that space even now, uh, so that they could get into that studio, but I want to do a little bit more than just us hang out in the studio. It'd be nice to make things interactive and integrate it into a more of a coursework if that makes sense.

Dave Ghidiu It does, and I think this will also, this really opens a door into possibilities for what we might do for online education, bringing everyone into the same space like a lab or something like that. So that's... I I was just blown away. I was in a, I I just was not expecting to have that fidelity that it did.

Paul Engin Yeah, I was, I was really happy with it. Um, and then the other thing since we just got off of this is uh, if you remember Open Claw?

Dave Ghidiu Yeah, I remember Open Claw.

Paul Engin Nvidia just...

Dave Ghidiu Oh, so go ahead do you want to explain what that is?

Paul Engin That was, we had talked about, we had an emergency pod and we talked about it's basically an AI agent, you let it loose on your computer and it has unfettered access to everything and on a good day it will do exactly what you want it to do.

Dave Ghidiu Yeah, and so Nvidia is uh, really getting behind this and they're doing their own version of Open Claw um, that's going to be a little bit more secure, but they see this being the future, even talked about it being, you know, the new operating system, which you alluded to in a previous in a previous episode. You know.

Paul Engin Yeah, that's where, where you will work is in the AI platform.

Dave Ghidiu Right. And um, and so it was interesting how they positioned it. So basically they've created their version of Open Claw that will base, have access to their models and it has guardrails and it has a little virtual sandbox so it has a little bit more control. And I, obviously they're trying to target more enterprise or business versus Open Claw which is just do whatever you want. Um, but it was interesting because I mean now it's gone mainstream, right? You go from this open source platform, people are really getting into it to now Nvidia which is the largest hardware company right now, I think it's actually the largest company in the world, um, is utilizing the same technology now and piggybacking off of that for enterprise and business.

Paul Engin Yeah, and this is an indicator to me that this is a big paradigm shift. Now it's been normalized. And it did not take long for full-throated agential AI to to be normalized. So that I'm interested to see what happens.

Dave Ghidiu Yeah, the only, my only drawback to this is that you're isolated to using just their LLM models and I think they only have four of them. And you have to pay per token that's being used. So you, you see their pay model being integrated into this, um, versus Open Claw which is completely open source. You can get an open source downloaded model and you don't have to keep paying, you know, uh, the the additional tokens for its use if that makes sense.

Paul Engin Yeah, and the tokens are basically, this is a gross distortion, but it's the the currency you use or the the blocks of information you use to process things. It's not um, it's it's something. The New York Times did just run a piece, uh, and I'll put this in the notes, about token maximizing, uh, where people are bragging in Silicon Valley like how many tokens they're burning through in a week.

Dave Ghidiu Jeez.

Paul Engin And yeah, so this is, in some respects like a sign of, uh, I think it's a proxy for productivity but it really is just might be a proxy for waste. Uh, an engineer at OpenAI processed 210 billion tokens...

Dave Ghidiu Jeez.

Paul Engin ...enough to, enough text to fill Wikipedia 33 times, uh, over one week. And so this is costing companies like up to $150,000 a month per employee that's token max, token maximizing.

Dave Ghidiu That's crazy. I mean...

Paul Engin Yeah, it's bonkers.

Dave Ghidiu Yeah, and even you were hesitant, like you said I'm putting five dollars down and I uh, and I'm going to do a AI agent for open, you did it with um, ChatGPT last episode, right?

Paul Engin Yeah, I was talking to you today. Yeah, and I'm token, token minimizing.

Dave Ghidiu That's right. But I, I did want to also share that we are hot off the heels of the AI for Marketing event here at FLCC where we had uh, 80 people from the, this was an in-person event. And we had some AI experts, we had a keynote from Adam HB from the college that who we all know. And it was a gathering of people who wanted to learn more about how to use AI in uh, in marketing. It was a fantastic event.

Paul Engin Yeah, it looked like it, that the place was packed, you had uh, a lot of good sessions going on. Uh, it seemed like...

Dave Ghidiu Had a good panel. Had a good panel.

Paul Engin Yeah. That's right, that's right. Uh, I believe uh, Laura was on that panel who uh, is heading up the marketing department in here.

Dave Ghidiu Yeah, friend of the pod. Friend of the pod.

Paul Engin We got a, a lot of friends of the pod.

Dave Ghidiu We, we do. And we are doing our big June convening again. That's going to be, uh, June 10th I believe, it's on the website. And that's going to be room for 300 people. Uh, and folks, listen, you got to get your tickets now because let me tell you, uh, in addition to all that great AI stuff that FLX AI Hub runs and offers, this will have, well actually I don't want to make any promises but I can tell you last year and also at the marketing event we had the Oreo muffins or the s'more muffins which are just out of this world.

Paul Engin That's right. I still have yet to have one of those. So...

Dave Ghidiu Oh my gosh, you got to, listen, you better, better be there at the June 10th event, get there early.

Paul Engin I know. I'm, I'm going to. Um, anything else or?

Dave Ghidiu Yeah, I do. I know we were talking about fonts and I've been really, that's really been knocking around in my head, we did an episode about uh, fonts we were talking about. So I read, I'm reading a book called The Psychology of Fonts right now, and I just want to let you know, I'm just going to read this one passage that I highlighted. "Helvetica is the bona fide celebrity of the typeface world. This divisive typeface has been close to total domination since the 1960s." So that's the kind of rigor that we're talking about when we talk about the font, the the um, psychology of fonts.

Paul Engin That's awesome. Yeah, Helvetica is just, it's got so many weights to it and uh, it's so easy to read. Uh, it's a sans-serif font so it doesn't have the decorative elements so it works on mobile. It's uh, yeah. Have you read this book? Did you write this book?

Dave Ghidiu No, I did not, but we uh, we talk about Helvetica and the different font families and the evolution of uh, of uh, typography basically. Um, and you can see how it, how it evolved from like old style um, to transitional and it's just from the um, from the uh, scribing of uh, the church like when they were writing the books, uh, that's the old style where it was kind of defined that thick style. And it went into transitional and so on, modern and sans-serif. And so we talk about it a lot in design classes. And we have a typography course here at Finger Lakes Community College as well now, too.

Paul Engin There is not. There's really a typography... I got to check that out.

Dave Ghidiu They were, they were talking about this study with jelly beans where they, they gave identical jelly beans to two different groups and they somehow and they were labeled with "eat me" but the font was like a rounded font for one and it was like a jagged font for the other. And the people in the group with the jagged font perceived the jelly beans as more sour than the other group. So the font really like...

Paul Engin Perception. That is crazy.

PSA Voiceover A sharp pencil for a sharp mind.

Paul Engin So Dave, why don't we get into the uh, series...

Dave Ghidiu I want to talk more about fonts.

Paul Engin No. That'll be next episode.

Dave Ghidiu That's right.

Paul Engin Um, so we'll continue the series. We talked about, our first episode was on um, OpenAI, uh, and then we talked about Claude. And today, we're going to talk about what I think is one of your favorites. I know it's one of mine I think. It's Google Gemini and the Google family of apps. Um, so right now, uh, I was thinking we could break it down into a few categories. We'll call it the brain, uh, the interface or what people are usually interacting with most of the time. Um, the builder side of things. And then the creative like media development side of things. And then uh, we can kind of go from there. Does that sound like a plan?

Dave Ghidiu That sounds like a great plan.

Paul Engin Do you want to talk about what the brains of this whole operation is and what the name of this Google thing is and whatnot?

Dave Ghidiu Sure. So, I mean Google Gemini, it used to be Google Bard when it first came out, it's come quite a long way. Google Gemini is their LLM, so much like ChatGPT or Claude. And this is where people go to use AI. And there's kind of three modes. So when you go to gemini.google.com, you can choose between Fast, Thinking, and Pro. There's like a little dropdown. And Fast is when you'd want like real fast answers, you're not doing like real deep thinking. You just need some, and this is the default, and it's usually actually really surprisingly good. But the next, the next step is Google Thinking, or is Gemini Thinking, which is an appropriate tradeoff I think for speed and intelligence. This is kind of where I live for most of my daily driver stuff. Uh, and then Pro is when I really need to like, when I need to crunch down, do some maybe programming or if I'm doing some images, I might use Google Gemini Pro. And this, this is where the deep reasoning happens. And if you want rich, rich responses then just flip it down to Pro.

Paul Engin And um, with each of these uh, they, they take uh, more resources for the process, is that right? So if you're on the premium plan, you might not have as much Pro bandwidth as you want.

Dave Ghidiu Tokens. Yes, and we, we talked about that earlier. Uh, so just keep that in mind that you don't want to just default to Pro or Thinking or Deep Think because you want to make sure that um, especially if you're in the premium version, uh, that you're utilizing it for, you know, I think like you said most of the time the Fast is is fine, but if it's not getting the results that you think you might want or if you want it to be really more thoughtful about what you're doing or processing, then you can kind of move toward those other models. Does that sound right?

Paul Engin Yeah, that sounds right. And there, there also is that like Deep Research one as well. And Claude and ChatGPT both have them. I think we talked about them, but the, the Deep Research is where you can task it to go figure out kind of a messy problem where it might have to look at 150 different websites and I find that this one is the best of the three, in my opinion. You, you know, you can get like a 20-page reply report.

Dave Ghidiu And this one might be thinking behind the scenes, right? This is like I could set it, maybe I have a a molecule study that I'm doing and it's looking at, you know, thousands of data points or millions of data points and it's just processing it, and then I move on to do the next task while it's still processing.

Paul Engin Yeah. Yeah, you can work on your own thing.

Dave Ghidiu Gotcha. So the interface, this is where people feel it? This is, what do you mean by that?

Paul Engin So, the interface is what most of us will interact with most of the time. So um, in this case here, if you go to like Gemini, the actual app, the chatbot, that is what we'll call the interface. So um, we engage with it in many different ways and most of the time people will just go to gemini... is it gemini.google.com is that the...

Dave Ghidiu It is, yeah.

Paul Engin Um, and they'll just enter the chat and that's how they'll engage with it. Uh, other people will use it, it's kind of... if you go to a, if you use Chrome, if you're in uh, Google Doc or uh, Gmail, you might see Ask Gemini at the top of your browser if you have enabled it, and you know, you can engage with it that way. So there's a lot of engagement points here for uh, the users. And when you're inside of it, then you can choose which brain you might want to use, if that makes sense.

Dave Ghidiu Yeah, yeah, yeah, yeah. Um, it makes perfect sense. And do you use it on your phone by the way? I'm just curious.

Paul Engin So, I'll be honest with you, I do not. I use, I use ChatGPT on my phone and uh, Copilot on my phone. And then I use Google and uh, Claude and ChatGPT on my laptop. So. Okay. How about you?

Dave Ghidiu That deserves a whole another episode. Uh, I use Gemini when I'm driving, I'll talk with it. I think it's the best for, for interaction with voice if I'm like driving, I'll say, hey I want to learn more about port knocking. And we'll have a conversation about it and I can ask a lot of questions. I started using Claude because I do a lot of design work, like instructional design work. And it, you can have a short conversation, but then it's like, okay, now I'm going to go build it, and that takes a long time. So now I do that on mobile. So then when I get to the computer, it's already built. And uh, I'll use, but whenever I'm doing any type of programming, I'm almost exclusively doing it on the computer. Although, since I was traveling, I was doing them both on, I was doing Claude and Gemini on my phone for programming, and which is a nice litmus test because when you're making a web page for instance or web app, it renders it on mobile because you're holding the phone. So you can see how it behaves on mobile, which is always a good barometer of how the app is working. So uh, I I find that I think I'm trending more toward some type of whatever device in my hand is what I'm going to be using at the time.

Paul Engin Yeah, and I think that that's another thing that you just brought up, which is really nice, is that conversational element, the multimodal uh, approach to uh, discussion or engagement with these AI systems. I think uh, they all have a form of it, but having, I think people are so used to just typing prompts and they forget that they can just have a conversation.

Dave Ghidiu They can hit the mic and just talk.

Paul Engin Yeah, and I find it to be a little bit richer. Uh, and I don't know why. I don't know if it's because the topics that I happen to be talking about are I are academic, like I'm always wondering about things. Or if it's because it's a little bit faster or if the voice is natural. I don't know, we'll have to, years from now we'll be reading books about the psychology of chatting with a chatbot.

Dave Ghidiu Yeah. Yeah. And it was, it's interesting you brought up the phone because I know uh, at one point I was doing some traveling and I uh, I tasked ChatGPT to build something for me and then I got on the plane. And then when I got off the plane, I uh...

Jeff Kidd You okay? Uh-oh.

Dave Ghidiu We lost our, our host. We're good. I'm good.

Jeff Kidd We'll have to bring it back up.

Paul Engin Alright, you're on the plane.

Dave Ghidiu So we, yeah we were on the plane and uh, and when I got off the plane I went and it was processing while I was on the plane so it was like working.

Paul Engin Yeah, it's great, right?

Dave Ghidiu Like modeling and doing all this stuff and um, I get a little Wi-Fi moment and then I'd ask it, you know, 'cause you can get Wi-Fi on the plane, so periodically I'd be asking it to do things. But for the most part, it was just working in the back end um, which was kind of interesting. And then I was able to get on the computer and see the results. So.

Paul Engin Yeah, and I think that's, that's really lightweight agential AI. And I know we're not talking about agential AI and that's not really, it's not quite ready for, for the, the people yet. But that's what it's going to be like, where it's like, I'm giving you a task, you do it, and then when I'm ready I'll come back and the task will be done.

Dave Ghidiu Yeah. So um, we see Gemini as a uh, really the, the app itself as like the entry point. Um, Google Docs is integrated, and again it makes it, and I'm using Chrome right now, I use Chrome for my web browser, and I have an Ask Gemini button at the top right. So I can just click it, it's very easy to access. And one thing I'm realizing is that Google has an advantage right now, that it's really integrated into almost the daily tasks. So I can just open up a Google Doc, I can click on the Chrome Ask Gemini, when I do a Google search, it pops up and I can ask for it to uh, you know, think about it more. So it's integrated already, I don't need to do anything special if that makes sense.

Paul Engin Yeah, and I think that's one of the ways, that's their strategy, their market domination for AI, because when I'm in Google Docs I'm typing, I can right-click on a paragraph and Gemini is like, would you like me to rephrase it? Do you want it shorter? Do you want it longer? Or when I'm in Google Slides and I need an image, I can just go to the, hit the nano banana button and be like I need an image of this, and it will do it, or it will make a whole slide for me. So the, you, you can't find that in a lot of, like ChatGPT just doesn't have that integration. And Claude has it but it's very specific within say Excel and Word. And uh, I I think it's just, it's so convenient, you're right, and especially with search, it's just dominating search now.

Dave Ghidiu And so I think that to be honest with you, Google is one of those things where even if you don't know it, you probably are using Gemini because it's just so integrated. So it's in Gmail. If you open up Gmail and you start using it, there's a little icon that you can click on, you can right-click. It's, it's just, it's seamless. It's a very low barrier, we talk about that a lot. There's no, you know, I think with some of these other platforms, you have to install a plugin or you have to install uh, an extension or you need to... but Google, and that this is one of their advantages is they already have a lot of these platforms web-based and they just integrated it. This is just another feature quote unquote.

Paul Engin Yeah.

Dave Ghidiu That's uh, part of their product line.

Paul Engin And you can see it, you can see it bleeding over. So Nano Banana was first in Gemini and then they're like, let's just throw it in Google Slides because that's where people are going to need it. And uh, Gemini, one of my one of the least known things about Google Sheets, which is like their version of Excel, is normally you'd do a formula, you'd be like equals concatenate and then you're like A2 comma A3, and that's how you would, would glue the two sides together. Now you can literally type in equals AI and then parentheses and then you just say in plain language, plain language, take this information and uh separate the spaces. So this would be like a split function. And then you just give it a range. So you can just explain what you want it to do and it does it.

Dave Ghidiu Oh my... and that's already integrated?

Paul Engin Oh yeah, it's been there for a year or two. It's insane.

Dave Ghidiu Oh, that's awesome. I didn't even know that. I've been still doing the, the split calls, the delimiters... what a chump man.

Paul Engin Delimiting is so 2008 buddy.

Dave Ghidiu Man, I'm, I'm behind. Um, well I will tell, tell everybody Dave is a guru at uh, Google Docs, he's your Google guy.

Paul Engin Can I give you a pro tip too? So I... Hunter who's the accessibility gentleman to the college put out an email last week talking about email signatures and just giving us pro tips on how to do uh make our signatures accessible and talking about the colors and the fonts and everything. So I was like, okay I should do this. So I was doing it today. And a lot of people don't even know this exists, but there's in, in the Google suite, there's something called Google Drawings, which is like MS Paint except more modern and much more practical. So one of the things that you can do, so I, I'm going to try to explain this. I took an avatar of me, okay, and I put it in Google Drawings and I made it a square and so it's just an avatar. And I was like, it would be nice if that was in my signature. So the old way of doing it was you would download that as a PNG, go into your signature, include it. But Google Drawings has this really unique feature where you can embed the drawing from Google Drawings. So you don't have to download it, you can just give, it gives you a URL. You got me so far?

Dave Ghidiu Yes, so basically rather than it being an image attached to the email, it's actually a reference to an image in the cloud.

Paul Engin Correct, and here's the bonkers part. Because it's a live image, I can change it, and that image changes in the email. So consider this, maybe during Halloween I put like a pumpkin on my head instead. Every email I've ever sent will now have a pumpkin on my head. So if you recall an email from two, three months ago, it will be the live version of it. And so you can use that trick...

Dave Ghidiu It's like a live image, yeah.

Paul Engin Yeah, and so, so you can do it for seasonal, you can do it for say, conferences. I know you shouldn't have text in images, but in the past I've seen uh, we use that for a countdown to an uh conference. So if you, in the leadup to the conference we'd send ten emails, and all of them would have this at the bottom. So no matter what day you were looking on it, it would be like, the conference is in three days, the conference is in two days, whatever, because we'd just keep updating that image. Uh, but we don't do that anymore because you shouldn't have text in images. Um, but this also works for all sorts of things. So that's, that's one of the...

Dave Ghidiu Very cool.

Paul Engin I know.

Dave Ghidiu That is really cool. Um, so obviously we have Gemini that's integrated um, into all the Google um, docs uh, platforms, whether it's sheets, docs, uh, forms, uh, you know, they got uh, Google Vids, they have Google Drawings, they have slides, ton tons of different um, suite of uh, production uh, applications that are equivalent to like Microsoft uh, Office. Um, they also have NotebookLM. Do you want to talk a little bit about NotebookLM?

Paul Engin I do. So I'll start off by saying Lindsey Chamberlin and Zeller who works in the college is our NotebookLM guru. She just knows everything about it. In fact, she just did a webinar, so I'll put that link to the recording in the show notes here on NotebookLM. But imagine a trapper keeper that's uh, digital. And so you might put all your math notes for an entire year in it. And NotebookLM, so you upload all your notes, or it could be links to it, could be Google Docs, it could be PDFs, it could be YouTube videos, it could be websites. And that becomes your source of truth for this particular notebook. And then there's a chat in it. And you can ask it questions. It will only give you answers based on the information that you've uploaded. So you get almost no hallucinations, most of the time it doesn't know something it'll say I, I don't know the answer to that, it's not in these documents. But when you do ask it a question, 99.9% of the time it does know, and it will say, it will give you footnotes. It'll be like, click here if you want to see exactly where in which document this information came from. And then there's also this whole media pane where you can generate slides based on it, or you can generate a podcast or a video. And now if you have the ultra plan, you can do cinematic videos. So it's not just explainer videos. And they have uh, all sorts of other media things that you can create. I, I will say this though, I think those are good starting points, but you can't edit the slideshow and you can't edit the video and you can't edit the podcast. So to me that kind of I tend not to use them. They're great in a pinch, but that's not usually my endgame when I'm making a slideshow or a movie.

Dave Ghidiu Well, I really like NotebookLM because it's closed.

Paul Engin Yeah.

Dave Ghidiu I, I still would not put personal um, information in there, but it is not trained on and it is strictly used in isolation.

Paul Engin Yeah. Fantastic, right?

Dave Ghidiu And so you can feel a little bit more confident when you're putting stuff in there for your research, 'cause I think that that's something that it's really uh, good for, is that you can have a lot, you could put like, if you're doing a research paper and you want to kind of build diagrams from it, you can have it look through your research paper and um, just isolate it to that versus having a you know, uh, search the web randomly and, and different things. But the, the other thing that I really like about it is it's an opportunity to uh, get mo, different modalities of learning. So uh, when I'm talking to people in the college, I say, you know what you could do is you could actually move your entire uh, PowerPoint or your um, anything that you're going to be lecturing on, to the NotebookLM and have it create a podcast. And then you can move that podcast up to whatever LMS you're using. And then students can learn from that. I know I learn from hearing, like I could be driving and it could be having the conversation about whatever lecture that you just went over, and it it's just a different way of learning and that people can, you know, ingest the content if that makes sense.

Paul Engin Yeah, so there's this multimodal approach uh, for different learning preferences.

Dave Ghidiu Uh, and, as you were talking I, I did forget that you're right, it doesn't train on the data, but I also just double checked and you can get the cinematic videos with the pro account. You don't need the ultra account, which is 150 a month, you can do it for Pro which is the typical 20 bucks a month.

Paul Engin Yeah, I, have you used it? I tried to do a a video, it was, it was okay, it was uh, but I'm a I'm more of a video snob maybe, but it was, it was okay. The uh... again, I think it's just amazing what it can do now. It can create diagrams and it can create charts and it can create um, interactives and it, it's just so many different things you can output too. You can um, do a slide deck from it, you can do mind maps, flash cards for studying infographics. Yeah.

Dave Ghidiu Yeah, the flashcards and the quiz questions are really, really interesting. If I were taking another class I think I would certainly use that feature.

Paul Engin Yeah, so I definitely think if you're an instructor that wants different modalities or different things that can be pulled from a previous lecture, just refresh it. I would definitely put it into here and see what it produces. It's a low-hanging fruit that can really garner I think a nice return if that makes, if that sounds right to you.

Dave Ghidiu Oh it does. And as you were talking about interactives, there is something else I want to say about the Gemini app which I haven't seen in the other ones. So Gemini as a chatbot, when you do a deep research, which is again like say you have this wicked problem and you can't be like ah, we're looking doesn't matter what it is, you're going to task it, it's going to go find 150 different websites and give you a report that's very, very detailed. Once you go to share it, or there's a create button, and you can create an infographic from that, you can create a website from it, and you can create an interactive element, which is really, really slick because then you can share it. So it's not just a static 20-page report. You can certainly give that, but you can also now give this website or give this interactive thing where people can click and expand. It's really fascinating uh, and that, that just bleeds over to all their other AI technologies that are embedded everywhere.

Paul Engin And I think that's where the the power of Google comes in, right, because you could have, let's say you've been working on this research project with NotebookLM for a long, long time. Well then you can link that NotebookLM into any other these modules, you know, if it's the Gemini app, if it's one of the Google Docs, like it, they all talk to each other with no new extensions, no new plugins, nothing. It just, it all just works together.

Dave Ghidiu Yeah, it's fascinating.

Paul Engin Um, and I know that uh, we kind of alluded to uh, Google Drawings, Google Vids, Google uh, these are all different uh, applications that you can use uh, for development. And they have Gemini integrated into it. Uh, and there is a thing called Google Labs that has all of these tools that you can access. Um, and there's a lot more. We're just scraping the surface and theyre releasing new apps I almost say every other week to be honest. It's just crazy to me how many things they're developing.

Dave Ghidiu It's like every other hour, Paul. Like I swear, I open my browser like, oh here's a new thing. And the Google Labs, and the link will be in in the episode description at the website as well. But Google Labs, it's just all their beta testing, it's fantastic stuff. And you can see where it will bleed into their future products.

Paul Engin Yes. And I, I'm almost certain if there is something that you're thinking about as far as, oh I wish there was an AI app that did XYZ. I'm almost certain if you go to the Google Labs, you will find that what you're looking for, because they have so many different things there. I mean...

Dave Ghidiu Have you seen the one where it's like you can create any font you want and I'm like, I want a font made out of eyeballs. And then it will create, like there's an AI tool for that. Uh, and then there's also Opal which is an AI tool for making AI tools. It's just, I'm actually looking through the the list of them right now and there's all sorts of things I haven't, that I didn't see last week, so it's holy smokes. So these labs are really cool.

Paul Engin Yeah, so I definitely suggest that all the listeners go check out Google Labs and um, and see what's there. Um, and uh, just to keep this moving, we're going to, we we might come back to Google Labs, but um, what I want to do is I want to get into um, more of, I guess I'll call it our realm. Um, the builder layer of Google Gemini or you know, where power users might start using this. And um, I think those are the tools that are like uh, AI Studio, which is basically a developer toolkit that allows you to create uh, agents and create um, it's equivalent to I would say like almost uh, the Codex or um, uh, Claude Code.

Dave Ghidiu This is one, one thing that is absolutely fantastic about this. And I would suggest to listeners you go check out AI Studio, because it is powered by AI. So in Codex and in Claude Code, I can say make a web app that's a clock timer or whatever. It doesn't require, it requires AI to build it, but it's not powered by AI. With AI Studio, which is free, you can say make an app that is a chatbot. And so now it's created by AI but it's powered by AI, it has intelligence. So when you go to it, when you roll it out and it has AI in the app that you can use. Uh, so, and it's free. It's for rapid prototyping, though, so it's not really for scale.

Paul Engin Yeah, and it, but it gives you access to a lot of things. You can do a lot of connections to it. Um, it's just, it's very extensive. You can get API keys so you can actually build AI-related apps as well and you can test it in this environment. Um, so again, it's more of that developer side of things, but I think it's a really great platform. Um, along with that uh, there is the um, Vertex AI which I haven't really used. Are you, have you seen that?

Dave Ghidiu That's like if, if you build something in AI Studio and you're like oh, I want to monetize this or I want to make it big. Then you move it to Vertex AI. But then you're paying for it. So Vertex AI is great for even if you don't know anything about programming, you can probably get something up and running. Vertex AI is like you really kind of have to have some background knowledge before you start doing that. Uh, but I do see on the list that you put on here Firebase, which is their database storage system. And probably half the apps on people's phones are powered by Firebase. But this goes back to your point about just their wide integration portfolio. So with Claude or Codex as I build me an app, say you want to make like a contact book or like a phonebook or something, it needs a database. So now you have to like go through the rigmarole of setting up, finding a database online that you can set up. With Gemini or Vertex or AI Studio, any of these coding ones in Google, you can say I'm setting up an app like it's a phonebook, and it will say oh let me set up this Firebase for you 'cause I own Firebase as well. And so it's just this vertical integration that is fantastic. And you don't, it's just so flawless and painless.

Paul Engin And it's just, everything is just so integrated, it's crazy to me.

Dave Ghidiu Yeah.

Paul Engin Um, and then they have uh, Anti-Gravity, uh, which is the... it's their basically, I guess this is more equivalent to their Codex, uh, where it's uh, more of a programming and it can take over some some raw jobs on your system and take over a folder and uh, add and and and do things if that makes sense.

Dave Ghidiu Yeah, so you can think of it as, generally speaking most AI assistant coding, they, there is a tool that's out there, Visual Studio Code, is probably the most popular. And Claude Code will sit in there, or you can put Codex in there. You can also put Gemini in there. But Anti-Gravity is that tool built from the ground up with AI. So it thinks about things a little differently. It has uh different access so it's, it's interesting. And actually if you go to labs, you'll see something called Jules, which is their predecessor to it. So you can see kind of again the evolution through the labs of where they are now.

Paul Engin So and again, with Anti-Gravity, it's tough because I use Visual Studio Code like you mentioned. And um, I have Claude hooked into there already and I have OpenAI hooked into there already and I have Copilot. And so um, I use Anti-Gravity on its own. I don't have it hooked in... wait... I don't have it hooked in, but uh, obviously in Visual Studio Code you can make your selection on which model you want to use um, for your programming. Um, have you used Anti-Gravity or do you...

Dave Ghidiu I, I haven't yet. Um, because I've been using Claude Code so much. But probably on my radar especially in the vibe coding course. So our next move is going to AI Studio and we're going to look at Anti-Gravity before we're done throughout the semester. So I will be checking it out and I'll, I can report back at a later date.

Paul Engin Sounds good. Yeah, and I've done a little testing with it and I've been doing a little programming with it. But it's kind of cool that it has access, you can give access to a specific folder and it can do searches in there, it can do whatever you want in that in that in that folder. Um, so I think in that respect uh, it's kind of a nice uh integration. Um, so those are all the development tools. And I don't want to get too nitty gritty into those tools um, 'cause it's really for a specific audience. But they, they have again this ecosystem of Gemini is just they all can cross-pollinate, they can talk to each other, they can... you can have one piece start here and it can share... it's just, it's just so crazy how well it's integrated, so it's kind of it's going to be a tough one to beat for people.

Dave Ghidiu Yeah, ah it's it's always there and it always works great.

Paul Engin Um, and then this leads to more of my realm, which is the uh creation and media development. Um, yeah...

Dave Ghidiu Before we do that, Paul, I think we just need to take a quick uh, commercial break, we'll be right back.

Paul Engin Sounds good.

PSA Voiceover In a world of digital clouds and, oops, I deleted it, choose the original hardware. The number two pencil. It's got a core of graphite, a heart of wood, and a built-in undo button that actually works on paper. No charges, no software updates, and it never asks for your password. When number one is too soft and number three is too hard, reach for a number two pencil. Number two pencils. We're number one.

Paul Engin Okay.

Dave Ghidiu And that's our sponsor. Excellent.

Paul Engin That's our sponsor.

Dave Ghidiu That's our first sponsor.

Paul Engin Awesome, awesome, I love it. Number two pencil, everybody.

Dave Ghidiu Yep, we're going for the best ad read.

Paul Engin Best ad read.

Dave Ghidiu That was, now we got it.

Paul Engin We got it, so now we should get that nomination and probably the trophy.

Dave Ghidiu That's right. Um, so uh, the last area that uh, Google has is uh, and and I think this is an area that some other ecosystems don't have, which is the media creation aspects. So um, when I talk about media creation, imagery, video, music, um, they have it all. Uh, so...

Dave Ghidiu And you're the guru here, so walk us through. My biggest question is, can you describe the difference between Veo and Flow, because I hear those interchangeably and I don't think they're the same.

Paul Engin Yes. So Veo is the model for video um, in Gemini. So if you're creating a uh if you want Gemini to create a video, you would pick the Veo model. And uh depending on when you're listening to this, it's Veo 3, Veo 3.5, Veo 4, Veo 5. It'll it'll be iterative, right? But um, it's basically, you can go into uh any of the Gemini chats and you can select Veo as your um, LLM inside of Gemini, and you can have it create a video intro. And then that's what it'll create, and then you can have it produce whatever videos you want, right? But it's tough to do any continuity, meaning if I had a character and I wanted that character to go from one scene to the other scene to another scene... It has a difficult time with that because it's recreating it, it's referencing maybe another video or it's referencing an image, but it's creating it from scratch every time if that makes sense. So it does a really good job at doing single videos. If you can picture that.

Dave Ghidiu Sure, sure, sure, sure.

Paul Engin So prompt for a single video. Uh, which...

Dave Ghidiu So if I if I were looking to do say a single video for the number two pencils ad, I might go to Gemini, click on the the VO or Veo and make a sweet, sweet ad.

Paul Engin Yes. Now it's going to be limited. So this is where you might add Flow to it, right? So you might start with Veo, it might create something, but then you might go to Flow. And Flow is where you'll be able to put the first frame and the last frame in and then it will create the in-between. So then you can prompt and you can say, I want character A to go from here to here, skipping. And then you'll see that character skipping to that frame. Um, so then at that point what we could do is we could use another one of their media creation tools, which is Nano Banana, which is their image creation tool. And again, you would uh select this as uh one of the um, the dropdowns in Gemini. So if you wanted to do image creation you would pick Nano Banana. And then you would say, create a fun number two pencil for me, um, you know, excited to be used, and then one with someone holding the pencil in the hand doing a drawing. And then I would take those two stills, bring it into Flow, and then I could transition between those two.

Dave Ghidiu Jeez. That's...

Paul Engin So...

Dave Ghidiu So and and I think in episode five of our podcast you might want to go back to listen to that was a holiday one where you talk, you do a kind of an exhaustive look at Flow and the the trials and tribulations of it.

Paul Engin Yes. And they've integrated Nano Banana even more into Flow, so now you rather than what I was doing is I was going into uh Gemini and I was using Nano Banana, creating the images, then bringing downloading it, bringing it into Flow. Now it's integrated. Again, it's that ecosys... they they're integrating everything together. So now I can just live in Flow and create.

Dave Ghidiu And it so, so now you don't have to go somewhere else to generate the first and last image, you can just put it in, your, you can do that in Flow.

Paul Engin Correct. And so um, Nano Banana is integrated in just about every Gemini platform now, right? So I think you mentioned you can go into Google Slide or Google Vid and you can use Nano Banana to create an image for you.

Dave Ghidiu Yes, uh, and and I it lives exactly there. And I do actually do that quite a bit, more than I thought I would. And also in Google Vids, you can create an image or a video. Uh, so Google Vids again is they're kind of like video creation. So if you're doing instructional videos and normally that would be me recording myself and then putting in there and putting maybe texts on there whatever, but now you can also create like the b-roll footage and just say like, I need, I need a picture, I need a video of someone sharpening a pencil while I'm doing the talk over. And it will just generate it right there.

Paul Engin Yeah, it's, it's amazing. Before you'd have to go to a royalty-free site, look for you know, do a search for a pencil, and then look for an illustration or photo and now you rather than doing that search you just type in what you want and it'll generate it for you, which is pretty amazing.

Dave Ghidiu Yeah. Um, and and I do recognize I I think we do need to point out that we you, we're both acutely aware of how this is changing say the industry of photographers, videographers, stock footage companies. Um, you know this is the the nature of the beast, it's not a circle that I can square, but uh I do recognize it and I I do what I can to support um local artists and when I can.

Paul Engin Absolutely. And I think that it's important to know that um, one of the things that you can do is you can still go to these royalty-free sites or you can go to uh a stock uh photo site and you can still do everything that you were doing before. Um, and maybe that is where you maybe the the AI is not producing exactly what you're looking for, and and that's what you have to do. I know I know that sounds really weird but sometimes specific things will require you to do things in the real world. So like uh, for us in that that holiday uh production, I wanted a drone clip of the college. And you can't get AI to do that. So uh Jeff had the drone, took it up, and it was snowing, it was perfect. And I used that as a still image. So if you picture in Flow, that was my starting image. And then he flew it in and then I took that last image, and then I um, Nano Banana'd it to be claymation. And then I brought those claymation images into um, Flow and then it created that transition of the of the drone flying into the school. And or actually I think it was flying out. I think it was our out our out. Yeah it was flying out. Um, but you can't you can't Nano Banana that right, because it doesn't create the school in its actual form. So I needed that base image first, which is what we went and shot and we did, and then I used Nano Banana on that to stylize it if that makes sense.

Dave Ghidiu Yeah, so this is just another reminder that you still need the human element to, to do what you did. And even if and again like when you watch that video it was like two and a half minutes, you're like, oh that made, it's AI, it must have been super easy. But it took 60 hours because you need to do things like that, and you need to have your expertise and you need to like understand cinematography.

Paul Engin Right. And so and I think it's just another tool.

Dave Ghidiu Yeah, another tool.

Paul Engin You know, um, and I think the other one that is Lyra, um, did I pronounce that right?

Dave Ghidiu Is I don't know, is this the Lyra what is this one?

Paul Engin This is uh the music.

Dave Ghidiu Yeah, so this is baked, baked into Gemini now. You can say write me a song about this, and it will write it. So, so Imagen, which is actually Nano Banana, is baked into Gemini, Veo is baked in. So now it's calling experts. So if you ask it to write a song, it will invoke Lyra, but you won't know that, it will just spit out a song for you.

Paul Engin That's awesome. And so now you can create music and songs and uh, we were again, I think Google is in a an amazing place right now because it's got all of these applications and integrations and they have both uh creative uh, media asset development, tons of different applications that you can use. Um, and it all works together. So they have that advantage over those other ecosystems.

Dave Ghidiu Yeah, it's it's gonna be bonkers man.

Paul Engin Um, the other uh URL that I would suggest people go to is uh deepmind.google. It shows another level of things that you can create and apps that you can use. Um, I had an opportunity while we were doing development to get the um ultra package of uh Gemini. And um I got to use Geni 3.

Dave Ghidiu That's the, yeah, Genie 3, tell us about that.

Paul Engin Yeah, it's a a literal 3D world creator. So you can create a 3D world from anything, and you can just immerse yourself in it, so it could be uh a race car from your point of from the racer point of view and um it gives you a little world sequence in this 3D space. Um, it's pretty processor intensive, it, you know, you can prompt it and it'll take a bit for it to create, but once it once it's created it's pretty amazing um the immersive worlds that you can build from it. And I don't know what else you can do with it yet, but you can see could you imagine being a game developer now and going, create a uh cyberpunk uh city and we're gonna and then if we can you can download it and then put your characters in it, it I'm just like oh my gosh.

Dave Ghidiu Yeah, it's the the world models is kind of the next frontier in AI and I think this is an an interesting uh, first go at it because so it, the way you described it to me was say there's like ant floating on a leaf going down a river. When when it's generating, it's generating on the fly, so when it starts it doesn't know what the river looks like. But you, as you like go down, it's kind of figuring it out and then you can look to the left and the right and up and down and see it all, right?

Paul Engin Yep, it's all 360, so you can spin around, it's like it's building the whole world, not just your point of view and locked to that view.

Dave Ghidiu And and I know that this sounds uh awesome and it is, but DeepMind was I think one of the first, it was around 2010, 2011, they were acquired by Google a few years later. But they are the that's like the deep, deep, deep, deep, deep research AI branch of Alphabet, which is the Google's parent company. And they are, they DeepMind was won the Nobel Prize for Chemistry in 2024 for its work on protein folding. I mean, also with some scientists, but the that was part of it. Uh, and they did AlphaGo which was the first AI to beat the world champion in the game Go. So DeepMind is really, really, really intense on their AI stuff. It's not just like ants floating down a leaf even though that's super cool. It's really doing heavy lifting on science and technology issues too.

Paul Engin Well, I didn't want to use all my tokens on on creating all that.

Dave Ghidiu Um, so uh, I definitely would suggest checking that out. Um, I know I one of the apps that I used from Google Labs um is uh Stitch uh which is uh a prompt where you can create a quick mock-up of a web page or app and you can get uh color palettes and it's just a really quick way that you can you can use it, and they have um... and there's not enough time to go over all of the apps, but there are tons of apps you should look at and uh the other one I just want to bring up is Opal you mentioned it before, but um it is not a dumbed-down version, but it's an easier way to make an agent almost. I feel like you can give it tasks to do in a very simple way like you know how when you were creating your agent you were like it's way it's another level. Opal makes it attainable to just about anybody can use Opal. You don't have to be a programmer, it's very uh easy to use, and you can create your own uh little Opal that you can develop something in and the one that I've used and I started developing is um social media post. So um you literally uh will go in and there's one that already exists and you can it's called remix, so you can grab the one that already exists, click remix, and then you can modify the instructions. But um you know, it's not I gave it a really easy prompt for new media. But the first initial prompt is basically step one, ask the name of your company. And you literally just uh put the name of the company, and your action is it goes online and it looks for your company. And then it comes back and then it'll give you options and then you're like, yep this is my company, and then it goes to the next question, and it gives you... and again, this sounds complicated but it's literally you put one, period, and you type you know, "ask the user their name of their company and then go online and try to find information about the company." And then you go step two, take that information and... and you're just in normal text defining what each of these things are.

Dave Ghidiu So so that really truly is like agential AI like in in in our hands.

Paul Engin Yeah, it's definitely to me it's a stepping stone to trying to build an agent. If you can do this, it's it's still node-based, so you still input connections, output connections, but it's very basic uh with layman terms. You don't need any coding at all. It's just right, someone needs to input something, so you need to put an input node, right. You want an output, so you just put an output node. And you say, do you want a video, an image, and it's just so intuitive I think that just about anybody can use it. So I would I just wanted to mention that as well.

Dave Ghidiu And these are all things that people should go try doing, like the Opal, we'll put the links again in the in get the https://www.google.com/search?q=immersivelens.com. But then AI Studio, even if it's scary.

Paul Engin Play, it's free.

Dave Ghidiu Yeah. I mean low barrier.

Paul Engin Exactly. So, and Gemini is integrated in everything. Almost everyone probably has a Google account already, whether you're on YouTube, which is Google, so if you log into YouTube, it's Google, right. So, all of these are all integrated. So I 100% would encourage you to go to Google Labs, try uh all the advantages of uh of Gemini, because it's already integrated in most mostly everything you do, don't you agree?

Dave Ghidiu Yeah. Uh absolutely 100%. And and when you were talking about Stitch I was like, oh yeah, they also have fonts.google.com, they own a font CDN so I was like there it is, Paul's right, it's like everything's in their verticals.

Paul Engin And we've been using that fonts.google.com before Gemini existed.

Dave Ghidiu Oh yeah, for like 10 years.

Paul Engin And it's a web font that we've been using for web development for custom fonts, but that's a whole other category. We're running out of time, um, and so I think that's all the time we have today, and I just want to say I am Paul Engin.

Dave Ghidiu Uh, I am Dave Ghidiu. If you enjoyed today's conversation, be sure to subscribe so you never miss an episode. Let's be careful out there, folks.

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.

Dave Ghidiu This episode was engineered by Jeff Kidd with special difficulty of having to bring people in from all over the country.

Paul Engin And I believe Hugh helped in the uh engineering of this masterpiece. The digital Dave.

Dave Ghidiu It was re- this is 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 fifty 55 degrees, certificates, microcredentials, and workforce training programs.

Paul Engin Thank you to Public Relations and Communications, Marketing, and the FLX AI Hub at FLCC.

Dave Ghidiu Eager to delve into your passion, discover exciting and immersive opportunities at www.flcc.edu. I'm getting good with those three Ws.

Paul Engin Good job. You're killing it. 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.

Dave Ghidiu 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.

Paul Engin Music by Den from Pixabay.

Dave Ghidiu This is The Immersive Lens.

PSA Voiceover Brought to you by Number Two Pencils.

Dave Ghidiu Front sound because I can't do it.

Paul Engin Did I ruin it, did I ruin it?

Dave Ghidiu No, you didn't.

Paul Engin By being here?

Dave Ghidiu No. I just can't, I played it off of my phone and I can't right now. Uh. So... you know, I I had a... a... a ding happen on my computer and I muted it, and then I didn't realize I muted you, Dave. I was like, why is Dave not talking?

Dave Ghidiu (Laughter)


full-width