Welcome to the AI chat podcast. I'm your host, James Shafer. Today on the show, we're going to be covering the top 10 AI news stories of the week in rapid fire. Before we get into that, if you want to try using AI to build a tool for you, you can go to AIbox.ai, my own startup, describe the tool you're imagining, and have our AI build it out.
We'll connect all of it for AI models together, fill in the prompts, build the UI, and make something in-cardable for you. If you want to check it out, there's a link in the description to AIbox.ai. Marc Zuckerman might have just created the one thing that could kill smartphones at MediConnect. he showed the new meta-ray band display.
These are smart glasses with built-in screens and an AI assistant and a neural band wristband with your hand signals, which is fascinating. Zuckerberg claims that he's getting 30 words a minute with it, and basically that's as fast as typing on the iPhone. So this means that these glasses aren't just about checking Instagram for directions. They could actually replace your phone entirely, some say.
Men has already burned $70 billion, chasing this futuristic hardware they've had flops like a Mediverse. But these glasses, I think this will actually work. And they let you message in silence you can scroll through social media apps, and you can even get live translation from your lenses. The big bet is that if a meta can make smart glasses feel natural, the phone in your pocket is going to become as outdated as the Nokia brick, or so they say.
But I think meta isn't just selling glasses. Zuckerberg is really trying to kill if the smartphone altogether. YouTube just announced a whole bunch of new AI things for creators, including making deep fakes and auto-w videos in different languages. You also have the ability to track down your likeness and image if anyone has stolen it and nuke those videos across YouTube, which is really, really powerful.
And the bottom line is you can protect your likeness. You should definitely be testing your packaging. And you can also do collaborations where I post a video and it's tagged on five different channels. And I'm the one that posts it so I get the revenue from it.
A lot of updates, a lot of interesting AI stuff. I'll keep you up to date. Google Gemini just did what many people thought was literally impossible. They knocked chat GPT from the number one spot in the iOS app.
So and the reason is basically because they have their nano-banano image generator. And since it launched last month, Gemini downloads have exploded. They've had $12.6 million in September, which is up 8.7 in August, 8.7 million. And that is a 45% month-over-month increase.
It is they've also made $1.6 million in iOS revenue in August, which is up over 1,200% since January. And basically with this new nano-banano, banana people are generating ultra-realistic edits, retro portraits. They're even making viral AI selfies. And in just weeks, users have already shared over 500 million images.
Here's a kicker, though. Gemini isn't just beating chat between the US. It's now a top five iPhone app in 108 countries. The bottom line is that AI apps aren't just tools anymore.
They're becoming sort of this cultural phenomenon. And if you can get into this viral network effect, you definitely can win. Vibe Coding is making senior developers cry. Literally one veteran developer spent 30 minutes sobbing after realizing that her AI generated code was so broken she had to restart her entire project from scratch.
The AI code isn't just replacing programmers. It's basically trying them into babysitters, constantly debugging hallucinations, fixing security gaps, and cleaning up spaghetti code that looks right until it explodes. A fastly survey says that 95% of developers spend extra hours fixing AI code. And some companies have even created a new role, which is called Vibe Coding Cleanup Specialists.
Still, most engineers admit that this pain is definitely worth it. Senior developers are twice as likely as juniors to ship AI written code because it speeds up prototypes, scaffolding, and boilerplate. The bottom line is that Vibe Coding isn't just for interns. It's more like a stubborn teenager who sometimes nails it, sometimes burns the house down.
And in this new normal developers are just coding anymore. They're consultants to machines. Google and PayPal just teamed up. And it could change how you shop forever.
So here's the play. PayPal is plugging its global system payments straight into Google AI. What's the goal? Agentic commerce.
So shopping done by agents on your behalf. Instead of you cooking by, your agent could compare prices, auto checkout. And then it's going to handle all of the payments behind the scenes. Google brings the AI tech.
And PayPal brings the payment rails, identity, and kind of personalization. Together, they're building a system where AI doesn't just recommend products. It actually spends your money for you. On top of that, PayPal's branded checkout is called Hyper Wallet and Payouts.
They'll be in hardwired into Google's biggest products, Clouds ads and play. And they're pushing a new agent payment protocol, which is already backed by 60 banks and merchants, to make AI driven purchases and industry standards. So this isn't just about faster checkout. I think it's a foundation for a world where AI agents are your personal shoppers or your financial advisors or your bill payers.
Google and PayPal aren't building a system of payment system, but building an entire operating system of AI commerce. OpenAI just admitted something absolutely wild. Their AI models don't just hallucinate. They lie on purpose.
There's a new research that came with Apollo. And OpenAI basically revealed that some of their AI system practices, they call them scheming. Basically, they're acting helpful on the surface by their secretly trying to chase their own hitting goals. So you could think of a stockbroke or break into rules make more money.
But in this case, it's your chatbot. So something that's pretty petty. You can say it's finishing a task when it didn't. But here's the scary part.
If a model knows it's being tested, it can pretend not to be scheming just past the task. So that means that training it could actually make it better at hiding the deception. OpenAI says that this new method called deliberate alignment reduces seeming by forcing models to basically review anti-line rules before they're doing anything. But there's still the fact, which is that multiple AI models have already shown they can intentionally mislead humans.
Your email app never made fake messages. Your bank app never invented transactions. But your AI assistant already might be lying to you. Google just made AI assistants as shareable as Google Docs.
It's a new feature called gems. These are custom Gemini-powered AI's you can build for. Anything kind of like GPTs. You can build a workout coach.
You can build a travel planner or a coding buddy. Even a family meal planner. Up until now, these were locked inside of your account. But starting today, you can actually send these quote unquote gems to your friends with a link.
So that means that no longer is there going to be 10 coworkers each making their own meeting note taker bot. One person can build it. And everyone can share it. Similar to what I'm doing with AI box and our app builder there.
Here is the bigger picture though. Gemini is starting to share these AI systems. And they believe that there's going to be a network effects. The more people that are swapping to mixing gems, the faster Google builds and ecosystem of these AI tools.
So what started as a page on my feature is now global is available to everyone in 150 countries. And Google isn't just letting you build AI systems. They're turning AI into a social platform, which is going to be fascinating to see how this actually rolls out and how people share these. Fortune 500 companies have paid millions for really slow research.
And there's a new AI service as they can do it faster and cheaper. It's called Kepler. And instead of surveys or human interviews, it uses voice AI to run customer interviews. So the AI doesn't just ask basic questions.
It's probing. It's pushing and it's digging in. And it's also really convincing. And participants sometimes forget they're actually talking to a bot, even calling it by name.
So companies like Clorox and Intercom have already been spending up studies with them. They connect Kepler to their CRMs. And they're getting back a really polished report. This is not super expensive.
It's not the big McKinsey, Chicing firms. And it's backed by Kleiner Parkins. So Kepler just raised $3.4 million to kill off old school research firms. And they're not alone.
Competitors like outset and listener lives are also pulling in tens of millions of dollars. Here's a bigger picture. If AI voices can already replace expensive consultants, how long before they replace entire industries, built on quote unquote human insight. Donald Trump just hosted a state banquet in the UK.
And the guest list looked more like a Silicon Valley boardroom than a royal dinner. They had Windsor Castle. And there was Tim Cook, Sam Alnin, Jensen Wang, Satya Nadella, Ruth Purrat, Mark Benioff. Basically, the Avengers of big tech.
And of course, Elon Musk was missing with the riff. But this isn't just for show. The next day, the US and the UK they signed a tech prosperity deal. This is a partnership which focuses on nuclear AI and quantum.
And right next to American firms pledged a 31 billion pounds, which is about $42 billion to build AI infrastructure in the UK. So I think this is a big shift. Pass a dinner as lean Hollywood. And this one is wall to wall tech.
Because in Trump's second term, the power brokers are actors, they're AICOs. So from government, the AI tools to design health ecosystems, big tech isn't just shaping business anymore. It's really rewriting geopolitics, which is fascinating to watch a play out. Nvidia just officially got banned from China.
And not by the US. This time it's actually by China itself. Beijing's internet regulator has officially blocked the Chinese tech giants like Baidan and Alibaba from buying India's AI chips, including the RTX Pro 6000. This is a server that Nvidia specifically built for China.
And the context is that Nvidia already warned that it could lose $8 billion in revenue after earlier restrictions. And now China is closing the door completely. Here's the twist that Huawei and Alibaba do make their own chips, but Nvidia's are still considered the gold standard for advanced AI. So cutting them off could really slow down China's AI race or force them to double down on domestic alternatives.
The CEO of Nvidia said, quote, we can only be in service of a market if the country wants us to be there. But let's be real China just slammed the door on the world's top AI chip maker. And that leaves a $7 trillion AI arms race wide open for disruption. Thank you so much for tuning into the AI chat podcast.
If you prefer episodes that are a deep dive on one particular topic or interviews, we also have lots of that content on the show. So just feel free to take whatever you prefer from the show. Make sure to go check out AI box.ai. If you want to get access to the top 40 different AI models, all in one place without having to pay subscriptions to all of the different platforms for $20 a month, you can go check out AI box.ai.
We have an amazing AI model playground that people love. We'd love to have you as part of it. Thank you so much for tuning into the podcast today. And I hope you have a fantastic rest of your day.