🎙️ EP 81: Google Kills Assistant, NASA’s AI Reads the Sun, and Sam Altman Calls the Bubble - podcast episode cover

🎙️ EP 81: Google Kills Assistant, NASA’s AI Reads the Sun, and Sam Altman Calls the Bubble

Aug 22, 2025•19 min
--:--
--:--
Download Metacast podcast app
Listen to this episode in Metacast mobile app
Don't just listen to podcasts. Learn from them with transcripts, summaries, and chapters for every episode. Skim, search, and bookmark insights. Learn more

Episode description

What happens when Google retires Assistant and replaces it with an AI that can reason, cook, troubleshoot, and even tell bedtime stories? And what if NASA built an AI that literally reads the Sun to predict solar storms?

We’ll talk about:

  • Google’s big move to replace Assistant with Gemini for Home — and why Alexa should be sweating
  • NASA + IBM’s Surya model that beats humans at predicting solar storms
  • Why 95% of enterprise AI fails, and what Altman’s “bubble” talk actually signals
  • Our take on the tools, trends, and funding chaos you need to know this week

Keywords: Gemini for Home, Surya AI, Sam Altman, AI bubble, OpenAI, Google AI, HuggingFace, NASA AI, Anthropic, Nvidia H20, AI Fire

Links:

  1. Newsletter: Sign up for our FREE daily newsletter.
  2. Our Community: Get 3-level AI tutorials across industries.
  3. Join AI Fire Academy: 500+ advanced AI workflows ($14,500+ Value)

Our Socials:

  1. Facebook Group: Join 251K+ AI builders
  2. X (Twitter): Follow us for daily AI drops
  3. YouTube: Watch AI walkthroughs & tutorials

Transcript

Imagine a model, an AI, right, learning from nine years of solar data. We're talking millions of images. Just so it can tell us when our son might, you know, decide to throw a tantrum. Wow. And not just predicting, but actually doing better than human experts by a pretty big margin. 16%, yeah. Yeah, 16%. That's a really profound thought, isn't it? The sheer scale of that learning. Whoa. It really is something else. And welcome everyone

to the Deep Dive. Glad to be here. Our mission, like always, is to take this, well, stack of the latest insights, like the big newsletter we just got, and really pull out those key bits of knowledge. We want to make sure you're not just, you know, hearing the news, but genuinely in the know about what's happening on the cutting edge of AI. Yeah. Get past the headlines. Exactly. So today we've got quite a journey planned. First up, we'll really unpack Google's big move with

Gemini for Home. That's replacing Google Assistant. Big change. Huge. Then we're taking a faster look, kind of rapid fire, at other critical AI developments happening across tech, finance, even education. Lots going on there. Always. And then finally, we'll circle back to that amazing solar AI we just mentioned, Syria, and dig into what it really means for us down here on Earth. Okay. Sounds good. So let's jump straight into that first big story. Google Gemini for Home.

Right. I mean, for almost 10 years now, Google Assistant has been, well, a fixture in a lot of smart homes, right? Answering questions, turning lights on and off. Yeah, the usual stuff. The usual. But now... It's officially being phased out. End of an era, almost. It kind of is. And in its place, Google is introducing something called Gemini for Home. They're pitching it as this next generation AI assistant. And honestly, listening to the details, it feels like way more

than just an update. It feels like a fundamental shift. How so? Well, Gemini for Home is designed to be much smarter, way more conversational. And it's really built for those complex, multi -step things that are just, you know, part of running a busy household. of it less like that old smart speaker gimmick and more like a household AI co -pilot. A co -pilot. I like that analogy. It feels fitting. Yeah. It's like having a truly intelligent assistant there, almost anticipating

what you need before you ask. Okay. So co -pilot it is. What are some of the... The standout features them. What makes Gemini for Home so different, maybe more capable than what we've had before? OK, well, the media search is a huge leap forward. You know how you're trying to find a show? Maybe you describe it kind of vaguely. Oh, yeah. And the old assistant just gives you that. Did you mean this dead end? All the time. Right. Gemini for Home is supposed to understand context and

intent much better. So it finds exactly what you mean across all your different streaming services. No more guessing games. I know that feeling all too well. Honestly, my family spends half our movie night just trying to figure out where a show even lives or which service it's actually on. It's frustrating. Exactly. It's that little friction point, right? It really takes the smart out of smart homes sometimes. And the smart home control itself is supposed

to be far deeper, too. It can apparently reason through complex commands more naturally. Reason through them? Yeah, like connecting multiple devices without you having to spell out every single step. Imagine just saying, like, get the house ready for movie night. Okay. And it maybe dims the lights, lowers the blinds, turns on the TV to the right input, adjusts the thermostat. You know, seamlessly. It moves beyond just simple commands to like actual integration based on

the goal. So it's really anticipating what you actually mean, not just reacting to the literal keywords you say. That feels like a pretty significant leap in understanding intent. It truly is. And for families, they're saying it's a potential game changer for coordination. You can manage complex routines just using plain natural language. Makes sense. And here's where it gets really interesting, I think. say, hey, Google, let's chat. And then you drop the hey, Google wake

word and just talk. Oh, OK. So like a continuous conversation. Exactly. It pulls information from search, your streaming platforms, all your connected devices all at once. It's a genuine conversational mode. It should feel much less like you're talking at a machine. So it's going beyond just simple trivia, basically handling anything from. complex questions to managing your whole home setup just through a flowing conversation. That sounds incredibly intuitive if it works well. Yeah, if it works

well. When can people expect to see this actually rolling out into their homes? Well, it's starting pretty soon. Gemini is set to replace Google Assistant on Nest speakers and smart displays. Early access is actually beginning in October. Okay, this fall. Yeah. And there will be both free and paid tiers, which is an interesting detail to watch. Subscription AI in the home. Could be. But the real significance here, I think,

is what this means for the broader market. This shift really puts competitors like Alexa and Siri in a, well, a tough spot. If Gemini becomes the household sort of default AI brain, the one that really understands and connects everything, the entire category could shift overnight. It forces everyone else to step up their game significantly. Right. Okay, so let's crystallize that. How fundamentally does this change our interaction with home technology?

Yeah, in short. It makes AI a truly conversational, intelligent home co -pilot. Okay, so now let's pivot a bit from the living room to the wider and, frankly, incredibly dynamic world of artificial intelligence. Moving so fast. It really is. It's such a fast -moving space, and we're seeing these rapid developments just, well, across the board. It feels like every single day brings a new headline, sometimes several. It's hard to keep up. What are some of the most striking highlights from

the sources we looked at today? beyond the smart home stuff okay well let's start with coding agents this is pretty interesting a founder john rush recently tested get this 61 different viral ai coding agents 61 yeah He created this really comprehensive list, put demos and notes together, and it apparently got over 1 .1 million views in less than a day. That's just wild. The sheer volume of interest in these tools is incredible.

It is. And just for anyone unfamiliar, AI coding agents are basically specialized AI programs. They can write code, help you write code, even debug human code sometimes. Like digital assistants for developers? Exactly. Like digital apprentices almost. That is a massive response. his list. You know, I have to admit, I still wrestle with prompt drift myself sometimes. Oh, yeah. Yeah. You know, where the AI starts to wander off from the initial instructions or loses context halfway

through. Trying to get really consistent output from these tools can be tricky. I know what you mean. So the idea of having a comprehensive, maybe vetted list of coding agents sounds incredibly helpful for a lot of people. myself included. Absolutely. And moving into higher education for a sec, Anthropic just announced two pretty major moves. First, they're forming a higher education advisory board with leaders from top universities. Second, they've also launched three

new completely free AI fluency courses. Oh, free. That's great. Yeah. Available for anyone to access. So that seems like a big push for making foundational AI concepts more accessible and better understood. Free courses are always fantastic for broadening access to about critical knowledge like this. What else is making waves out there? Well, Google also shared some really excellent use cases for their Gemma 3270M model. Ah, the compact one. Exactly. Remember, that's a remarkably small

and hyper -efficient AI. It's all about packing powerful AI into smaller packages. Which is important. Why? Well, it's crucial for running AI directly on devices, phones, laptops, maybe even cars, instead of always relying on the cloud. It expands accessibility and reduces the need for that massive cloud infrastructure for a lot of tasks. This whole on -device AI idea is a major step towards making AI more widespread and personalized. Right.

That efficiency piece is key for wider adoption and probably enables totally new capabilities, too. It truly is. Then there's the geopolitical angle, which is always simmering. NVIDIA might be looking at losing billions. China has reportedly told its tech giants like Alibaba and ByteDance to ditch Nvidia's H20 chips. Those are the watered down ones, right? The ones designed to comply with U .S. export rules. Exactly. The less powerful versions Nvidia made specifically for the Chinese

market due to U .S. restrictions. But it seems China now views even those as, well, maybe an insulting concession they're not willing to make anymore, according to the reports. Wow. It just shows how sensitive the whole global... tech landscape is and how political statements can have these huge ripple effects on business. Indeed. The strategic implications of something like that are enormous. It impacts supply chains, national tech strategies, everything. They really

are. And, you know, big tech just keeps making these massive deals. Meta just signed a $10 billion six -year agreement to run its AI operations on Google Cloud. $10 billion on a competitor's cloud. Interestingly, Meta's stock dipped a bit after the news while Google saw a little bump. This isn't just about a big check changing hands. It signals a really critical strategic choice,

right? Even tech giants with vast infrastructure of their own are recognizing the specialized, frankly, enormous scale needed for advanced AI. They're choosing to partner, even with rivals. So the market might be seeing it as Google gaining an edge in this AI infrastructure arms race. That seems to be the read, yeah. Almost like meta is becoming reliant on a competitor for something core to its future strategy. That is a fascinating dynamic. Wow. Precisely. And the

fundraising. Oh, man. The fundraising frenzy just continues. Anthropic is reportedly raising $10 billion now. $10 billion? Didn't they just raise a huge round? They did. They're apparently doubling their initial goals for this round. This could push its valuation to, well, an astounding $170 billion. Whoa. Yeah. With major investments already coming in from Iconic Capital and other big VCs, the amount of capital pouring into this AI space is just incredible. It's a staggering

figure. Truly reflects the markets. maybe hunger is the right word, for AI right now. We've also seen a few other quick but pretty notable things pop up. Like what? Well, Google's AI mode in search is now global. Yeah. Rolled out to 180 countries. That really makes AI search almost ubiquitous for information retrieval worldwide.

Right. Part of daily life now. Yeah. And OpenAI executives are apparently claiming that their next model, maybe GPT -5 Pro, might even be able to prove new interesting mathematical theorems. Prove new math. That's what they're hinting at, which, if true, is. That's not just mind -boggling. It potentially moves AI beyond just mimicking patterns into true original discovery. That's a profound leap. Yeah, it could redefine the role of human intellect and feels like pure math.

No. A huge potential shift. And on a completely different note, but still AI, China actually launched its Wukong AI on its space station during some recent spacewalk. AI in orbit during a spacewalk. Yep, bringing AI into space in a very direct way. Meanwhile, back on Earth, Meta, after that big hiring spree for AI talent, you know, poaching people, is now reportedly pausing its AI hiring, which could signal maybe a broader market recalibration or maybe they just got everyone they needed for

now. Hard to say. And finally, Sam Ullman himself, while, you know, successfully selling open AI shares at a valuation around, what, $500 billion? Half a trillion, yeah. Has also apparently warned about an AI bubble. Which for listeners tracking the market raises that critical question, doesn't it? Are we seeing genuine sustainable growth here or is it getting frothy? Yeah. Driven by speculation, maybe like past tech booms. It's

definitely a lot to chew on. So pulling it all together, what's the unifying theme you see across all these really diverse AI developments? I think it has to be AI's relentless, almost pervasive integration into nearly every facet of life. It's just everywhere. Sponsor. Okay, so earlier, right at the start, we touched on this really astounding AI development from NASA and IBM. Surya. Surya, exactly. Now, let's really dive into that. This solar foundation model that literally

learned to read the sun. Yeah, this is a truly fascinating project. It's marrying, you know, cutting edge AI with really crucial space science. Historically, sure, we've had tools to monitor the sun. Right, satellites, telescopes. Exactly. But apparently none have been as fast or as accurate as Surya seems to be at recognizing those really early signs of danger coming from our star. And that's the core problem it's tackling, right? Because space weather isn't just an abstract

thing. It can have serious impacts here on Earth. Oh, absolutely. Disrupting communication. Messing with power grids, even knocking out satellite navigation. Yeah, GPS. GPS, exactly. So this solution, Surya, they're calling it a solar foundation model. Maybe we should quickly explain what a foundation model is in this context. Yeah. Good idea. So a foundation model is essentially a very large AI model. It's trained on a really broad range of data, often unlabeled data. Okay.

And that broad training makes it highly adaptable for many different, more specific tasks later on. Think of it like a... Like a highly educated generalist AI that can then specialize quickly in various fields. Got it. So Surya was trained as this generalist, but specifically focused on the sun to become a specialist. Yeah. How exactly did they train it to be so effective? What was the data? They fed it an absolutely

immense amount of data. Millions of images gathered over nine years from NASA's Solar Dynamics Observatory, or SDO. Nine years of constant observation. Yeah, it's an incredible volume. And through that, Surya basically learned to track everything, from emerging sunspots, you know, the little dark patches, to tracking solar wind speeds. And crucially, it processes multiple wavelengths of light simultaneously. Ah, so not just visible light. Right. Different wavelengths reveal different

things happening on the sun. This allows it to spot tiny surface changes, things that humans often miss, or that current automated systems just aren't sensitive enough to pick up reliably. So it's not just seeing what we see, but it's quite literally seeing more detail across more spectrums and processing it faster than we ever could manually. Precisely. And what can it do once it's learned all this? What are its main capabilities? Well, its primary capability and

the really vital one is prediction. It can predict when a solar eruption like a flare or a coronal mass ejection might blast out towards Earth. Okay. And the key thing is it's already outperforming the existing predictive models by a remarkable 16%. 16 % better prediction. Yeah. That's a significant leap in accuracy for something that's so crucial to protecting our infrastructure and frankly,

our safety down here. Outperforming. by 16 % is huge when you're talking about potentially damaging space weather events that could impact global systems. But what I find truly exciting about this beyond the science is the accessibility aspect. Oh, absolutely. This is maybe the coolest part for the wider community. NASA has open sourced the entire Syria model. They put it up on Hugging Face. Hugging Face, yeah, the AI platform. Exactly.

So this means any researcher, any student, even a private company working on space weather, they can access and use Syria for their own projects. Wow. It truly democratizes this incredible scientific tool. It should really help accelerate further innovation in the field. That's such a powerful

move. Making sure its impact can be felt far and wide, not just with... in NASA or IBM, you know, it really makes you realize with climate change already putting so much strain on Earth's systems, knowing when our nearest star plans to, well, throw a tantrum, as we said, that might just be the most important forecast of all. It directly impacts our planet's resilience, our

ability to prepare. So thinking about the direct impact then, how does this specific scientific AI innovation really affect us, you know, day to day on Earth? Well, fundamentally, it crucially improves forecasts for space weather, which helps protect our planet's vital infrastructure communications power. Okay, so let's try to synthesize our whole deep dive today. We've covered some truly transformative ground, really highlighting how pervasive and powerful AI's evolution is becoming. Yeah, it

feels like three main themes emerged. Right. First, we saw AI making this deep integration into the home. Gemini for Home isn't just, you know, another update. It feels like a fundamental shift. The co -pilot idea. Exactly. Moving AI from just simple command and response to being a true... household co -pilot, maybe intelligently anticipating needs. And that challenges all the astounding players, Alexa, Siri, and hints at this new era of genuinely conversational intelligence

right in our living spaces. Okay, that was theme one. Then theme two was that whirlwind tour of AI's explosive growth and all the strategic plays happening. Yeah, from those specialized AI coding agents getting millions of views. Crazy numbers. To multi -billion dollar cloud deals between giants. And even those geopolitical shifts driven by chip technology. AI is just everywhere now. It's evolving at this incredible pace, constantly weaving into new industries, new aspects of our

lives. And that's prompting both massive innovation and some serious strategic recalculations across the board. And then finally, the third theme was seeing AI tackle these really grand challenges, like with the Syria project. Right. That demonstrates AI's immense power to help solve complex scientific problems. It's offering unprecedented accuracy in vital areas like space weather forecasting

in this case. Yeah. It's really about leveraging AI for... some of the biggest questions facing humanity, pushing the boundaries of what's possible in scientific discovery and, well, planetary protection. It really has been a fascinating journey through the latest developments. And honestly, there's so much more to explore within each of these topics. Always more to learn. Always.

If you, the listener, are curious to continue your own deep dive after this, I'd highly recommend checking out Google's use cases for their compact Gemma 3 model. The efficient one. Yeah. Understanding how powerful AI can actually run on smaller devices is, I think, a really crucial insight into where things might be heading. Or, you know, for the truly adventurous, go explore the open sourced Syria model on Hugging Face yourself. See the sun's secrets. Exactly. There's a lot to learn

there. And these powerful tools are increasingly right at our fingertips. So thank you all for joining us on this deep dive today. As AI integrates more deeply into our homes, into our global systems, even into our very understanding of the universe, what new questions do we maybe need to start asking about its role in shaping our future? It's definitely something worth pondering.

Transcript source: Provided by creator in RSS feed: download file
For the best experience, listen in Metacast app for iOS or Android