Press play (above) to listen to the podcast. Or if you prefer, here’s an edited blog / transcript.
The JEN_AI Project.
What happens when you mess with machines? Hack around, find out.
This is JEN_AI Challenge #003: Future Podcast.
I get one AI to predict the future, while two others tear us to pieces.
// Hit me with some highlights.
Clips from AI podcast presenters (and one from me)
“Oh, man, Google Glass. What a, what a flashback. I still have mine somewhere, I think. Gathering dust.”
“So we have a special guest.”
“An AI co host. Yeah.“
“Like super intelligent, but kind of detached from humanity.”
“Yeah, like remember that movie Her? Oh yeah, totally. Like that kind of vibe?”
“So anyway”
“You know, the whole AI hype that's going on right now. Yeah. It's everywhere you turn someone's talking about AI. Exactly.”
“It's so funny how that happens with tech, right? Like the hype builds and builds and then reality kinda.. Fizzles. Yeah, or just like never lives up to the expectations.”
The Jen AI Project.
Podcasts.
// What is this?
Hi! 👋 I’m Jenni Munroe. I’ve studied and worked in artificial intelligence, data and tech since a time before computers could identify hot dogs or cat pictures. I’m on maternity leave from Google, and I’m challenging myself to stay up on the latest in AI, tech & media by messing with the latest in AI, tech & media.
The Computer vision struggle parodied by Silicon Valley (the sitcom) in 2017.
This is the JEN_AI project, where I join forces with the machines to generate experimental creative content and give you the insights no-one is asking for.
So far we’ve covered:
Challenge #000: A blog about writing a blogChallenge #001: A newsletter about making a newsletter with the latest news about the newsChallenge #002: A free speech about generating free speechToday we have the podcast podcast (#003).
If you’re checking this out, say hi! Subscribing is free and sends each new episode to you, never more than twice a week (usually less). Let me know your ideas for future challenges or how I could make this better.
Thanks for reading JEN_AI! Subscribe for free to receive new posts and support my work.
Up now, it’s “JEN_AI Challenge #003: Interview with a chatbot, podcast-style.”
Time to play.
// So let’s talk podcasts & politics.
Podcasts have come a long way - from the funny little niche broadcasts xennials (link) listened to on an iPod, to the unexpected media juggernauts shaping culture and politics today.
Joe Rogan & Theo Von - titans of the podcast industry
Podcasts like Joe Rogan (the #1 most popular podcast in the U.S.) host apparently unfiltered, unscripted platforms for power players like recent guests Trump, Musk, and Zuckerberg to rant for hours, cutting through a lot of AI-curated noise and hyper-edited media. Some are saying they’re the 2024 election cycle’s campaign-winning equivalent of Roosevelt’s fireside radio chats, JFK’s TV charisma, Obama’s social media mastery, or Trump’s data-driven 2016 campaign.
Where did I hear this? A Podcast, of course. So you can take this hype-commentary with a pinch of fake news, but as with the surprise success of Google’s NotebookLM’s Audio Overviews feature - which generated the podcast-style chat between two AI hosts you heard clips from in the intro - people are realising that long-form audio isn’t just an easier way to digest information, it’s making us feel like we’re in the room. Podcasts have become a way to forge real human para-social connections even in a world that’s increasingly artificial.
// So what did we generate this week and how?
* A ChatGPT podcast interview script & read-aloud functionality
* A Google NotebookLM Audio Overview discussing that script
* A podcast blog post using Descript (media editor) and Substack (media platform)
// Attempt #1: ChatGPT Podcast Script: Ai-Bruce’s Predictions for 2025
Generated by ChatGPT, and inspired by the “NET_NEW News” newsletters (link) we previously created and some firm prompting. My edits are marked in italics.
Title: What Does 2025 Have in Store? A Chat with an AI that Knows the Future… Sort Of.
Host (You):
Welcome to the very first JEN_AI “NET_NEW News” podcast chat! Today, we’re diving headfirst into the deep end of predictions for 2025. AI will dominate the headlines—again—but the real question isn’t can AI do this or that, but how should AI do it? And who better to ponder this with than… well, AI?
Say hello to my co-host for the day. I’d introduce them with a bit more flair, but they’re a bot, so we’re skipping the pleasantries. ChatGPT Bruce. Edit: Actually I’m so sorry, that was a typo/autocorrected, I meant ChatGPT’s Spruce voice. Described as “Calm and Affirming” - just what we need.
ChatGPT Spruce:
Bruce works. It’s short, punchy, and just aloof enough to sound mysterious.
You:
Ok (italic) Perfect. Bruce, we’re here to make predictions for 2025. So, no pressure, but I hope you’ve got your crystal circuits polished. Let’s start with an easy one: What’s the most overhyped thing people will keep talking about next year?
Bruce:
Oh, easy. The Metaverse. It’ll still be the tech equivalent of that guy who says, “Trust me, this is gonna be big,” but never explains what “big” actually means. Except now, it’ll have slightly better avatars and… legs.
You:
Finally, legs! Revolutionary. And speaking of hype—AI’s set to integrate into, well, everything. What’s that going to look like?
Bruce:
Picture this: You wake up, and your coffee maker suggests adding oat milk because it’s January, and you’re “doing a thing.” Your fridge tells you to stop hoarding expired pesto. Then your car, which you thought was electric, suddenly decides to self-identify as autonomous and starts suggesting scenic routes to the office
You:
Wait, wait—scenic routes? Is this the future of productivity? Or just cars trolling us?
Bruce:
Both. Productivity will look like this: AI managing your entire schedule, sending polite but firm nudges to stop doomscrolling, and reminding you that yes, you did sign up for that Pilates class.
You:
Ah, so my future is basically one long guilt trip, curated by AI. Love that for me. But let’s get serious for a second—what’s the biggest challenge AI faces next year?
Bruce:
Oh, definitely ethics. Everyone’s asking, “How should AI do this?” but the real issue is who gets to decide what’s ethical. Are we trusting governments? Big tech? A committee made up of well-meaning but slightly clueless experts who think ‘algorithm’ is a dance move?
You:
If it’s a dance move, it’s definitely not moving fast enough. Like, TikTok wouldn’t even bother with it. Speaking of trust, do you think people are finally going to stop falling for AI-generated misinformation in 2025?
Bruce:
[Laughs] Let’s just say: Deepfakes are getting so good, they’ll make even your high school friend’s pyramid scheme look credible. But here’s the twist—2025 might also be the year people start using AI to fight AI. Verification tools, watermarking systems—it’s going to be like an arms race, but for memes.
You:
Finally, memes with integrity. That’s the world I want to live in. Okay, one last prediction—what’s the wildest thing you think could happen next year?
Bruce:
Alright, here’s one: Someone will try to train an AI entirely on cat videos, claiming it’s the purest data source on the internet. And, weirdly, it’ll work. By December 2025, we’ll all be taking life advice from an AI that meows.
You:
Honestly? Sounds more trustworthy than half the gurus on LinkedIn. Spruce, this has been an absolute blast. Any parting words for our listeners?
Bruce:
Just this: 2025 is going to be weird, wild, and full of surprises. Stay curious, stay skeptical, and maybe start saving for that AI-powered fridge—it’s not going to come cheap.
You:
Spoken like a true futurist. Thanks, Spruce—and thank you for tuning in to the first-ever JEN_AI Chatbot Chat. If you’ve enjoyed this little experiment in AI banter, hit subscribe, share it with a friend, or just yell your favorite prediction into the void. Who knows? Maybe the AI will hear you.
Wrap-Up Music Plays
Outro tagline: NET_NEW News: Your 3-minute tech culture fix. Unless it’s a podcast. Then it’s… however long this was.
// Notes on Attempt #1: Future-Bruce interview
Well mistake #1 was me messing up the bloody chatbot’s name. Sorry Spruce.
Continuing the theme from our last episode - text-to-speech generation - ChatGPT has a read-aloud function meaning you can get it to read any of its responses to you, and you can choose from a range of voices, sensibly described not in gender or racial terms but vocal tone like “Composed and Direct”, or “Calm and Affirming” like Spruce. (Supposedly ChatGPT has a back-and-forth Voice Conversations mode but I couldn’t get it to work in the UK within a few minutes of trying, so that’s one to try again in the actual future.) Anyway, not the most natural-sounding, especially with a script, but not the worst podcast interview you’ve ever heard? I need to work on my own delivery. A solid 4.9/10.
// Fail Attempt #2: How not to use the excellent NotebookLM from Google
Ok so Google’s NotebookLM is an AI-powered tool that helps users summarize information, generate insights, and can create audio discussions *from source documents they provide*. Its Audio Overviews feature (basically podcast-generation) generated a buzz on launch for giving a really fresh way to interpret info. People were surprised by the natural-sounding banter of its AI presenters.
However. This right here is not how you are supposed to use it. You are supposed to feed it the underlying source material about a topic, not the podcast scripts you want it to read out. I did it for the hell of it but I somehow also copy/pasted only the beginning of the script in and didn’t realise until the bots had already started chatting s**t about it. In an interview with lead NLM creators Steven Johnson and Raiza Martin, they did say one of the best outputs they’d heard was from a document that said only two words repeated over and over - cabbage and puddle. So Playing *is* definitely encouraged. But anyway more on all that in a future episode that does this cool tool justice. Run the clip!
Transcript of NotebookLM’s “Audio Overview” of the Bruce-Ai-Predictions Interview Script.
Sounds great in audio but hell to read in transcript. Maybe skip or play.
Change to embedded audio clip!
“All right. So, uh, today we're going deep on something a little different, you know, we usually look at like articles or blog posts, but this time you guys sent in something really cool. It's like a, uh, a draft script for a podcast episode and get this. It's set in 2025, like in the future, and it's got this, uh, AI co host.
An AI co host. Yeah. It's called Chat GPT Bruce. Chat GPT Bruce. Interesting. I know, right? So we're going to kind of like dive into this whole script and see what it tells us about like, you know, the whole AI hype that's going on right now. Yeah. It's everywhere you turn someone's talking about AI. Exactly.
And this script is like a window into, you know, how people are imagining the future with AI. Yeah. I'm, I'm really curious to see how they portray it. Me too. So the first thing that struck me was, like, right off the bat, they give this AI a personality, you know? Right away. Yeah, and it's so, like, immediate.
Yeah. Like, right away, you know, they're setting him up to be, like, this character. Yeah, and they give him a name like Bruce, which is, like, so casual, such a human name, right? But then they describe him as, like, aloof. It's almost like they're playing with that stereotype of AI. Exactly, like the super intelliJent, but kind of detached from humanity.
Yeah, like remember that movie Her? Oh yeah, totally. Like that kind of vibe? Yeah, that kind of like, you know, knows everything, but doesn't quite get us. Exactly. So anyway, ChadCPTBruce, this AI, he jumps right into predictions. Mm hmm. You know, it's a future podcast, so. It's future predictions, that's what we want to hear.
Right. And the first one is that, uh, the metaverse is going to be the biggest overhype of 2025. Oh, wow. Okay, so like, it's gonna be a flop. Basically. Yeah, it's like, remember when everyone thought 3D TVs were gonna be like, Hmm? The next big thing. Yeah, and everyone rushed to buy them. And then what happened?
Nothing. It was a fad. It's so funny how that happens with tech, right? Like the hype builds and builds and then reality kinda Fizzles. Yeah, or just like never lives up to the expectations. Right, like Google Glass. Oh, man, Google Glass. What a, what a flashback. I still have mine somewhere, I think. Gathering dust.
Exactly. But anyway, back to this, uh, this script. There's this funny little detail. Oh, tell me. It's about avatars in the metaverse. You know, they They don't have legs. Yeah, they finally get legs in 2025. Oh, so that's progress. Right. But it's kind of like this little joke, you know? It's kind of funny when you think about it.
After all this time. And all these promises about these immersive virtual worlds, they're still figuring out legs. It just shows you how the reality of this tech is moving a lot slower than the hype, you know? Yeah, totally. And then the script moves on to AI being like integrated into our everyday lives.
Okay, so like practical applications. Yeah, exactly. And they start with this example of a coffee maker. Okay, a smart coffee maker. Yeah. And what's interesting is how subtly They show the AI's role changing, you know. Okay, so it's not just like one thing. Yeah, like at first it's just making suggestions. Yeah.
You know, like, oh, maybe try oat milk today. Okay, so like a helpful suggestion. Right, but then it's like giving you reminders, like, hey, that pesco in your fridge is about to expire. So it's like keeping track of things for you. Yeah. And anticipating your needs. Okay. I can see that being useful. Right. But then they get to the car and things get really interesting.
Oh, the car. What happens with the car? It's not just an electric car anymore. It's fully autonomous. Like self driving? Yeah. And it's suggesting scenic routes for your drive. Wow. Hold on. So we went from coffee suggestions to a self driving car choosing our route. It's like they skipped a few chapters in the AI manual.
Yeah. That's a pretty big jump. Right. And that's where the script kind of abruptly ends. Oh, really? Yeah, it leaves us with this image of the car, like, making decisions for us. It's kind of unsettling when you think about it. Yeah, like Yeah. Is it really just about the scenery? Or is ChatGPT Bruce Like, subtly nudging us in a certain direction.
It makes you wonder about the whole idea of control, you know? Yeah, exactly. It's like, as AI starts doing more and more for us, are we okay with it making decisions for us, even small ones? Like, what route to take? And what happens when those decisions become bigger, you know, more important. That's a good question.
And I think that's what this script is trying to get us thinking about. Yeah, for sure. It might be set in 2025, but it's definitely raising questions that we need to be thinking about right now. Absolutely. Because this AI future is closer than we think. Exactly. And on that note, we'll leave you to ponder those questions and maybe check the expiration date on your pesto just in case.
Good advice. Thanks for listening, everyone. See you next time.”
// What next
Well that’s enough for today but we’re not done with podcasts yet. A Googler friend reading JEN_AI did jokingly question my loyalty to Google as I keep using ChatGPT. (It’s market research, honest.) But it was definitely time to formally invite Google Gemini to enter the chat.
Podcast Part 2 coming soon, where we use the wonderful Google NotebookLM properly and dig out a few of Google Gemini’s hidden gems to go deep into the inner workings of the universe and de-mystify quantum computing. Or possibly just make one of those ASMR podcasts you can fall asleep to. Possibly both. Find out next time at JEN_AI.
// Thankyou
If you’re still tuned in, thankyou. Communication like this doesn’t happen in a vacuum (why? There’d be no sound! In a vacuum. Aha.) You’re a big part of this experiment and thankyou so much for listening. If you like JEN_AI - subscribe for free, forward it to a weird friend, and keep sharing your thoughts & your own AI experiences. If something doesn’t seem right please tell me so I can make this better. Thanks!
All views expressed are my own (or possibly a chatbot’s) and produced on (unpaid) maternity leave. They are not endorsed by Google.
This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit jenaihacks.substack.com
