Imagine a scientific paper. Okay, now imagine it has over 3 ,000 authors. Wow. Yeah, more authors than some small towns have people. Just think about that for a moment. Welcome everyone to the Deep Dive. So today we're digging into this really fascinating newsletter. We're going to unpack what it's telling us about, you know,
the absolute cutting edge of AI. Our mission basically is to explore how AI research itself is changing, the incredible new tools it's building, and get this, even how it's creating labs that drive themselves. Self -driving labs. Yeah, should be some real aha moments in here for you. All right, let's dive right into that first kind of startling fact then. The recent Google Gemini 2 .5 paper. Yeah. It listed, astonishingly, 3 ,295 authors. I mean, just let that number sink
in. Yeah, 3 ,000. And that's not a typo. Seriously. To put it in perspective, right, the first Gemini paper that had, like, 1 ,250 authors and GPT -4, you know, the open AI model. That's smaller. Way smaller. Only 417. It seems like... OpenAI and Anthropic are, I don't know, maybe a bit more selective with credits? Could be. But Google's list for Gemini, it jumped, what, 144 % in less than two years? It's wild. And what's really fascinating is that the paper apparently had
this hidden message. Oh, yeah. The first initials of the, I think the first 43 authors, reportedly spelled out something like, AI is a team sport. Nice. Which isn't just, you know, a clever little Easter egg. Yeah. It feels like Google was intentionally highlighting this new reality. That AI development is now just this massive collective effort. Exactly. These papers are starting to look like phone
books. Seriously. If this keeps going, I mean, just projecting it out, by 2040, we could be looking at like 2 .6 million names on one paper. You'd need AI just to read the author list. Right. We'll need AI to summarize its own author list. It's kind of funny, but also a bit... Mind bending when you think about the scale. It really makes you wonder, though, doesn't it? When you have thousands of contributors, likely all over the world, how do you even begin to coordinate something
like that? Yeah. How do you track who did what specific piece? Right. It feels like we're genuinely stepping into a new era for science collaboration. More like particle physics or the human genome project, maybe. Yeah. Big science. Yeah, absolutely. And this huge growth in authors, it's not just more bodies, right? It really signals that AI isn't just a coding problem anymore, not by a long shot. It's become this grand, like interdisciplinary symphony. You've got linguists, ethicists, policy
people, neuroscientists. All sorts. All working together to build something. Well, pretty unprecedented. It's not a sprint by a few geniuses anymore. It's like a huge expeditionary force. Yeah. So stepping back, this huge shift in authorship. What's it really telling us about AI development right now? Well, I think it clearly shows that AI research now demands vast, diverse and interdisciplinary collaboration. It's just that complex. OK, that makes sense. And here's where it gets really
interesting beyond just the giant papers. We're seeing new AI capabilities pop up super fast. Like what? Well, breakthroughs in video generation, for one. There's a new real time model. It lets you direct a full minute of AI video. A full minute. Yeah. And you can actually adjust the prompts while it's generating. No more of those like eight second clips like VO3. That's a big jump. And I saw something about FluxPro combined with Seedence producing hyper realistic video.
Oh, yeah. The results from that are. Honestly, pretty remarkable. Getting harder and harder to tell what's real anymore. But maybe the biggest thing gaining traction right now, this idea of agent mode. It's OpenAI's latest thing. Agent mode. Think of it like ChatGPT, but it's connected to a virtual computer. It can actually do stuff. So let me get this straight. An AI agent is basically, it's a program that can think and act for you. Yeah. By controlling a whole computer. Exactly.
So it can carry out complex tasks all by itself. Autonomously. Precisely. It's like delegating to a really, really smart intern who can actually use your software, browse the web, write code, the whole deal. Wow. And look, Mistral just upgraded its assistant, LeChat. It's got a deep research mode now, native multilingual reasoning, even image editing built in. So it's becoming a serious alternative. Yeah, definitely a compelling open alternative to something like Microsoft Copilot.
And speaking of movement. We've seen some pretty big talent shuffles too, right? Anthropic hiring back key cloud code leaders. Yeah, that was kind of an unreversed card on the usual talent poaching, wasn't it? Pretty interesting move. Definitely. And Perplexity AI, the search company, just got a huge valuation boost, like $18 billion. Yep. Positioning itself as a major rival to Google search using AI. Things are moving incredibly fast in that space. It really is. So these new
agent capabilities. How do they really change how we're going to interact with AI, do you think? Well, fundamentally, it shifts AI from just being a thing you talk to to something that can actually do things for you, a digital colleague almost. Right. So in short, these agentic capabilities mean AI can now take initiative and complete complex actions on your behalf. Exactly. It's a big step. Okay. So beyond those big headlines and these new agents, there's also just this
flood of new AI tools, right? Designed to automate stuff, boost creativity. Oh, yeah. It's like a Cambrian explosion of tools right now. Take Uncursor, for instance. It lets you build and deploy live apps. in literally seconds in seconds yeah imagine what that unlocks for people who aren't coders the speed of innovation there that's huge or fast3d .io you type text or feed it an image and boom eight seconds later you have a 3d model eight seconds That could totally transform
design workflows. Artists could iterate like crazy. Totally. And then there's stuff like force equals, which claims to turn just an idea into like a full execution ready plan. Interesting. And symbol, which can take a long PDF or an article and just turn it into a video tutorial for you. So it's all about making complex things easier, faster, radically supercharging what one person can do. Pretty much. And then you have the quick hits. Seen some viral AI art coming from Perplexity's
Ambassador. And an interesting little note that Anthropic might be quietly tightening the usage limits on Claude for some people. Something to keep an eye on. Yeah, definitely. And Adobe's new tool, turning silly noises into realistic audio effects. That just sounds like pure fun. Right, a creative playground. And OpenAI is apparently working on a native checkout system. So the AI could potentially... complete purchases for you. Whoa. Okay, that's interesting implications there.
And Microsoft Copilot can now apparently see whatever's on your screen in real time. Real time screen awareness. Yeah, that adds a whole new layer of context for the AI. It's amazing, really. All these tools, all these rapid developments, they're incredible. But honestly. Yeah, I still wrestle with just the sheer volume of it all. Yeah. Yeah. It's a lot to keep up with. Oh, tell me about it. You're definitely not alone there. It feels like drinking from a fire hose most
days. Right. But what's fascinating, I think, is that underneath all this rapid fire stuff, there's actually a pretty clear pattern emerging, isn't there? Yeah, I think so. For all their differences, what's the common thread you see weaving through these diverse new AI tools? Well, they're not just automating tasks, you know. They're fundamentally lowering the barrier to creating complex things or getting complex things
done. Lowering the barrier. Yeah. Okay. So the common thread is that they significantly simplify complex processes and dramatically accelerate creative output. Yeah. I think that sums it up nicely. They empower the user in a big way. Okay. So if we kind of connect that idea of empowerment and automation to the really big picture, we're starting to see AI not just assist with discovery, but actually drive it. Exactly. Which brings us to this work at North Carolina State University.
Researchers there have built a fully autonomous AI -powered chemistry lab. Fully autonomous. Like no humans involved day to day. Pretty much. This lab runs continuously 24 -7. It collects something like 10 times more data than traditional lab methods. And it's finding promising materials for things like clean energy, advanced electronics. Finding them in days, not the years it usually takes. It's a total game changer for material science. So it's not just running one experiment
and stopping. Nope. It captures a new data point literally every half second. It analyzes that data, learns, and then adjusts the next step of the experiment mid -run. On the fly. On the fly. And it never stops testing. It's this constant, intelligent loop of discovery. Design, test, learn, repeat, nonstop. And you mentioned it's efficient, too. Yeah. Big time. Uses fewer chemicals, cuts down lab waste drastically. So it's actually enabling sustainable science acceleration. That's
a huge win -win. Whoa. Okay. Just pause there. Imagine scaling that up. A self -driving lab like this, discovering materials for, I don't know, a billion new applications we haven't even dreamed of yet. Right. It's like Netflix for chemical reactions. Just streaming experiments constantly. No pauses. Things that normally take researchers weeks or months, done in hours or days. My mind kind of explodes thinking about what this could mean for medicine, for battery
technology, for, well, everything. What makes this really stand out, though, is, well, we have systems like DeepMind's AlphaFold, right? Right. Predicting protein structures or NVIDIA's drug discovery pipelines. Sure, powerful tools. But those are largely computation, right? They don't typically operate physical experiments in real time in a lab. Exactly. This NC State lab, it fully automates the entire loop design, test, learn, repeat, all within the physical world.
It's a closed loop system actually doing the chemistry. It's the integration that's key. Totally. So now that this kind of autonomous lab is clearly possible. It kind of begs the question, right? For you, for us, for everyone listening, where's the open AI of chemistry or the clod for clean energy materials? What's the next step in really leveraging this kind of autonomy for scientific breakthroughs? It's a huge question. Yeah. But it's undeniable how this kind of self -driving
lab revolutionizes the pace of discovery. Yeah. How would you summarize that impact? I'd say it enables... continuous rapid experimentation, basically compressing years of traditional research work into mere days. Speed and scale. Years into days. Okay. Sponsor. All right. So let's try to wrap our heads around all this. What does it all mean for us, for you listening? We've seen AI research itself scale dramatically, needing thousands of collaborators now. Yeah. The scale
is next. We've seen this explosion of new AI tools and these aging capabilities really empowering users to do much more, much faster. Democratizing complexity almost. And we've explored the emergence of these fully autonomous. AI systems like that chemistry lab that can accelerate scientific discovery at just an unprecedented rate. Yeah. When you put it all together, the sheer scale of the research effort, the rise of agents that can act for us and now labs that literally think
and experiment for themselves. Right. It's becoming really clear that AI isn't just, you know, assisting us anymore. It's evolving into an active, intelligent partner. A partner in discovery and creation, even in basic science. So here's maybe a final thought to leave you with. If AI research papers now need thousands of authors and labs can design, test, and learn entirely on their own, what happens to the role of human curiosity? Human creativity. Where do we fit in a world increasingly driven
by this kind of autonomous intelligence? That's the big question, isn't it? A really profound shift to think about. It really is. Well, we hope this deep dive gave you some valuable insights today, maybe sparked some of your own curiosity about where all this is heading. Yeah, and if something particularly struck you or made you think, definitely let us know. We're curious, too. Absolutely. Thanks for joining us for this deep dive into the latest in AI. Until next time, Aotiro Music.