Game Changers in AI This Week

Ollie Carter

Alright, welcome back to The AI Intelligence Podcast — with thanks for Advancer, The AI Agency. I'm Ollie Carter, joining you from sunny Brisbane, as always, and I'm here with my mate and co-host Llew Jury. Big week, Massive swings in the world of AI tech.

Llew Jury

Yeah, g'day everyone! It’s Llew here. And look—the pace is just ridiculous lately. Kinda feels like every time you blink, there’s another model drop or some lab outpacing the last.

Ollie Carter

Totally. Top of the pile this week: Google’s Nano Banana. Yes, you heard right—Nano Banana. Which sounds more like my breakfast, but no, it’s their latest image editing model. This thing has shaken up the scene; what was it, over ten million downloads already and, like, two hundred million image edits—just since launch?

Llew Jury

Ollie, it’s wild. And Google’s aiming right at OpenAI’s Sora and Adobe’s Firefly. They’ve basically made high-end image and even 3D object manipulation something anyone with a login can do. I’m not sure I’ve even wrapped my head around that adoption curve. I mean, those are start-up-to-scale numbers in literal days.

Ollie Carter

I’m just picturing my uni days, hacking away at this janky 3D image tool I found on a shareware CD. I reckon that’s the moment I got obsessed with the idea that software could make creativity scale, you know? I think that’s part of what got me into tech gigs later—just chasing down this kind of leap. Nano Banana is light-years from what I played with back then.

Llew Jury

Ha! A shareware CD, there’s a throwback. I was still using floppy disks from the late 90's. But you’re right—the bar gets raised every six months now. And it’s not just the flashy stuff—MIT’s FlowER system is a real game changer for labs. It’s predicting chemical reactions way more accurately, which is a big step for drug discovery and even new materials. Proper deep science applications, not just memes and cat pics.

Ollie Carter

It’s a massive leap. If FlowER does what they’re hoping, it speeds up new drugs coming to market but also makes chemistry less about guesswork. I’ve already seen a few AI start-ups circling to spin this out for industry, too. Feels like we say it every episode, but the toolbox for researchers just keeps expanding.

Llew Jury

Absolutely. And—I don’t wanna skip over this—Google’s also putting out VaultGemma 1B, that monster open-weight language model. What caught my eye, though: they’re training it with differential privacy from the ground up. That’s tech speak for, ‘We’re actually thinking about privacy, not just saying it for the press release.’

Ollie Carter

It finally feels like the big players care about privacy in a practical way. Not perfect, obviously, but end-to-end privacy with these models is—in theory—a real win for actual human users.

Llew Jury

And it all links back to where we’ve been heading—more capable, more open, hopefully safer AI. Righto, let’s roll into the next chapter, because the race for smarter language models is heating up even faster.

Llew Jury

Alright Ollie, time to talk LLMs. And this week it’s all about the models pushing into reasoning, not just spitting out generic text. OpenAI has dropped its o1 model—but, and I love this, the whole idea is to get the model to take more time on tough questions. Kinda like a chess grandmaster, not just playing fast, but thinking deeply.

Ollie Carter

Yeah, it’s like, “Don’t rush me, I’m thinking!” which honestly, I could use in real life some days. I remember the last episode we did on infinite context models—now it’s about how long they'll dwell on a tricky bit before answering. I’m curious: does a more thoughtful model mean we’re headed to something genuinely trustworthy, or just, y’know, slightly slower with the same hallucinations?

Llew Jury

Good question. And you’ve got DeepSeek R1 showing up as well, going full open-source. This stuff used to be big lab-only, but now, smaller players can push the boundaries a bit. Honestly—and this is where I might differ from some—I reckon open-source will actually steer most of the practical gains, at least short-term. Lower entry, heaps of community eyes on the code, you don’t get locked into some proprietary walled garden.

Ollie Carter

I dunno, Llew—the big tech houses like Google and OpenAI, they’re not gonna give up the moat that easy. Gemini 2.0’s Flash Thinking feature, that’s proprietary all the way. And every time Google or OpenAI launches a shiny new thing, everyone races to match it. Remember last year when GPT-5 landed and the open models had to scramble just to keep up?

Llew Jury

Yeah, and here we are, what, six months on, and there’s, like, two dozen reasoning-optimised LLMs out. I’m not saying open wins everything, but look—DeepSeek R1 gets everyday devs tinkering, prototyping stuff the big labs would never even think to try. That level of access? Gold. Of course, it worries the big labs. But as we saw with Meta releasing Llama and OpenAI’s own GPT-oss, even they can’t totally resist the open movement now. You reckon open-source or proprietary sets the bar, Ollie?

Ollie Carter

For the near-term? I think it’s a hybrid. You’ll get open stuff leading the way for experimentation—like, good enough for 80% of problems—but the absolute bleeding edge, the stuff getting all the VC cash, that’ll be behind paywalls and API keys. But—yeah, like you said—open means more eyes, more mistakes caught. Sorta keeps everyone honest, right?

Llew Jury

Couldn’t agree more. It’s moving way too fast for a single company or model to win outright. Anyway, all these models aren’t just disrupting labs—they’re shaking up how business gets done. Let’s unpack some of that next, yeah?

Ollie Carter

Sweet, so—big-time shakeups this week. Intel, for starters, is seeing a mass leadership exodus. Michelle Johnston Holthaus—in charge of a massive product group—she’s bowing out after three decades. That’s not nothing, especially as Intel’s doubling down on their own AI and foundry bets to stay relevant. I might be reading too much into it, but it screams “changing of the guard,” don’t you think?

Llew Jury

Absolutely. And losing Ronak Singhal—that’s the Xeon server brains trust leaving. You rarely see top hardware architects pull the pin unless there’s something bigger shifting underneath. I know Intel says it’s all about new priorities, but in this climate, could be the classic move to get some serious AI specialist talent in to a business for a refresh or planned renewal.

Ollie Carter

Either way, the old tech giants aren’t immune to all the pressure from the upstarts. But it’s not just the Intels and Nvidias, is it? Look at Yum Brands—the KFC, Taco Bell, Pizza Hut lot—they just promoted Jim Dausch as Chief Digital and Technology Officer. And, key point, he’s steering a full-stack AI platform for them. AI’s everywhere; even your Friday night takeaway deal has a neural net behind it now!

Llew Jury

Yeah get a Zinger box and order it through an AI chatbot! But seriously, we’re seeing this cross-sector: banks, food, entertainment. Even Malaysia rolled out Ryt Bank, its first fully AI-powered bank. That’s pure infrastructure, handled start-to-finish by algorithms. Scary or genius, your pick.

Ollie Carter

I'm still thinking about that, honestly. Even the HR world is shifting; Recruit Holdings—major player—just cut thirteen hundred jobs, AI-driven restructure. Real people, real impacts, but also just the beginning of this trend I reckon.

Llew Jury

And Ollie, I gotta mention this—a mate went to Vivid Sydney last week, and he said he saw a drone light show run entirely by AI. Unreal! It’s easy to forget, but that blend of tech and art—AI isn’t just bank algorithms and chemistry models. It’s the stuff you remember years down the track, and the future is right here!

Ollie Carter

I love that. That’s why we keep coming back—AI’s everywhere, for better or worse, changing whole industries and letting people create new stuff we never imagined. Makes next week’s news anyone’s guess, huh?

Llew Jury

That’s it from us for this episode of The AI Intelligence Podcast. If you’re keen to keep up with the next big shakeup—or even just want to know what that Zinger box chatbot is up to—make sure you subscribe and join us next week. Ollie, always a pleasure mate.

Ollie Carter

Legend, Llew. And before we go, a shout out and thanks to the great team and our sponsor Advancer. Advancer are the AI experts providing real world AI and Automation strategy, AI products and customised AI solutions for your business. Check them out at Advancer.com.au! Thanks for tuning in everyone—catch you next time!