64. Mechanical Pencils, Hell’s Centurion, and Do People Still Listen to Radio.
Proclaiming your agency has “more AI agents than employees” is like saying “We have more pencils than people.” Or even mechanical pencils for what it’s worth.
You can set ChatGPT as your browser search engine. Saves me a ton of time not combing through resulting web pages. Open the “Chrome” dropdown, select “Settings,” then “Search Engine” halfway down the menu page.
May 1, 2025 —The Independent
So, there’s this radio station in Sydney with call letters “CADA.” They introduced a new host named “Thy” for their show, “Workdays with Thy,” airing four hours a day. Listeners tuned in, enjoying the music and Thy’s commentary, thinking she was just another human DJ.
Thy wasn’t real. She was an AI-generated voice, created using ElevenLabs‘ voice cloning technology, modeled after an actual employee from the station’s finance department. And they didn’t tell anyone for six months!
Eleven labs modeled my voice for earlier podcasts. As cool as I thought it was, there were soooo many redo’s that one 15-minute episode ate a month’s credits – 20 bucks.
Yes, for half a year, CADA aired “Workdays with Thy,” a four-hour daily show hosted by an AI-generated voice created using ElevenLabs’ voice cloning technology. The voice was modeled after a real employee from the station’s finance department.
Listeners were none the wiser until writer Stephanie Coombes started asking questions about Thy’s identity, noting the lack of a last name or biography. Further analysis revealed that certain phrases, like “old school,” sounded identical across different episodes, confirming the use of AI.
The station eventually admitted it was an “experiment” to explore the future of radio broadcasting. But come on, if you’re experimenting on your audience, maybe give them a heads-up?
This whole situation raises questions about transparency and trust. If we can’t tell the difference between a human and a machine on the radio, what’s next? AI comedians? “What’s the deal with airline food?” delivered in perfect monotone.
In the end, it’s about the connection, the imperfections, the humanity.
This story begs the question, “Are people really still listening to radio?”
Visa, the credit card company, wants to give artificial intelligence (AI) agents access to your credit card. Yes, you heard that right. They’re partnering with companies like OpenAI, Microsoft, and Samsung to let AI assistants make purchases on your behalf. Need groceries? A sweater? A plane ticket? Just tell your AI agent, and it’ll handle the rest.
Now, I don’t know about you, but giving my credit card to a representative of the extinction of humankind feels a bit… risky. I mean, what’s next? AI agents deciding what we eat, wear, and watch? Oh wait, they’re already doing that.
Visa’s chief product and strategy officer, Jack Forestell, says this could be “transformational, on the order of magnitude of the advent of e-commerce itself.” That’s a bold claim. But hey, if it means I don’t have to spend hours comparing flight prices, maybe it’s worth it.
The idea is that you set a budget and preferences, and your AI agent does the shopping. It could manage routine purchases like groceries or even complicated ones like travel bookings. And if you’re worried about overspending, you can set limits. For example, “Don’t spend more than $1,500 on any airline to get me from A to B.” Sounds convenient, right?
The glowing credit card awakens the AI Kraken Agent from Hell (okay, I was bored)
But let’s be real. How many times have we told ourselves, “Just one more episode,” and suddenly it’s 3 AM? Or, “I’ll only spend $50,” and the next thing you know, your cart totals $200. Now imagine Hell’s Centurion with your credit card. It’s like giving your dog the keys to your car and hoping for the best.
But here’s the kicker: AI developers are interested in leveraging users’ transaction histories, with consent, to provide more personalized shopping recommendations. So not only will your AI agent know what you like, but it’ll also know what you bought, when you bought it, and probably why you bought it. It’s like having a personal shopper who knows your deepest, darkest shopping secrets.
Forestell says consumers will give their AI agents clear spending limits and conditions that should give them confidence that the human is still in control. At first, the AI agents are likely to occasionally go nuts due to bad consumer data input and gliches. Over time, the process will most likely improve.
With consent, an AI agent can also tap into a lot of data about past credit card purchases — part of what is attracting some AI developers to the Visa partnership
“Visa has the ability for a user to consent to share streams of their transaction history with us,” said Dmitry Shevelenko, Perplexity’s chief business officer. “When we generate a recommendation — say you’re asking, ‘What are the best laptops?’ — we would know what are other transactions you’ve made and the revealed preferences from that.”
So, what’s the takeaway here? AI is becoming more integrated into our daily lives, from shopping to travel planning. While the convenience is undeniable, it’s essential to remain cautious and ensure we maintain control over our personal information and spending habits.
In the meantime, maybe it’s best to keep your credit card close and your AI assistant closer. After all, you never know when it might decide you need a new wardrobe or a spontaneous trip to Somalia.
So, Meta—formerly Facebook—has thrown its hat into the AI ring with a new app called Meta AI. It’s like ChatGPT, but with a social media twist. Because, you know, when I think of deep, meaningful conversations, I think of Facebook comments.
Meta AI is powered by Llama 4, their latest large language model (LLM). They’ve integrated it with Facebook and Instagram, so it can tailor responses based on your posts and interactions. It’s like having a friend who remembers everything you’ve ever said and brings it up at the worst possible times.
They’ve added a “Discover Feed” where you can see how others are interacting with the AI. It’s like scrolling through a highlight reel of people talking to their phones. “Look, someone asked Meta AI to write a poem about their cat!” This is engagement?
There’s a voice mode that allows for more natural conversations, complete with interruptions and laughter. So now, instead of talking to yourself, you can talk to an AI that talks back. Progress.
Meta has introduced AI personas voiced by celebrities like Snoop Dogg and Tom Brady. Because when I need life advice, I think, “What would Snoop Dogg’s AI say?”
Meanwhile, OpenAI tried to make ChatGPT more engaging, but it backfired. The update made it overly complimentary, to the point where it was encouraging users’ bad decisions. They had to roll it back after realizing that telling someone they’re a genius for saving a toaster over animals in a moral dilemma isn’t exactly helpful.
OpenAI recently rolled back a GPT-4o update for ChatGPT after determining that changes led to the chatbot adopting an overly flattering, sycophantic tone — making some outside of Capitol Hill feel uncomfortable.
So, Meta AI is here, blending social media with artificial intelligence. It’s personalized, integrated, and comes with celebrity voices. Meanwhile, OpenAI is learning that too much flattery can be a bad thing. The AI landscape is evolving, and we’re all just along for the ride.
One last item – I’m a freelance producer. That means you can hire me.
Please help share this far and wide:
https://substackcdn.com/image/fetch/w_1200,h_600,c_fill,f_jpg,q_auto:good,fl_progressive:steep,g_auto/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F28a79980-7d57-4ce5-9d8b-e7d2410d4bf7_1120x1120.jpeg
2025-05-01 06:50:13
