Latest / In the Loop AI / Announcing Google's Secret New AI Model With The Person Who Built It | Logan Kilpatrick

Transcript

Announcing Google's Secret New AI Model With The Person Who Built It | Logan Kilpatrick · In the Loop AI

• 0:00
  Logan: The best image generation and editing model in the world.
• 0:03
  Ejaaz: It's scary how realistic this stuff is. VO3 has kind of like killed the VFX studio.
• 0:09
  Logan: And this is, I think, principally enabled by Vibe Coding. My hope is that it
• 0:12
  Logan: actually ends up creating more opportunity for the experts and the specialists.
• 0:15
  Josh: How much of the tools that you build do you find are built with Vibe Coding?
• 0:19
  Logan: I'm like almost 85% of everything that I do Vibe Coded.
• 0:22
  Ejaaz: I remember when I first booted up a PC and I just had access to all these different
• 0:26
  Ejaaz: wonderful applications all within one suite. This kind of feels like that moment for AI.
• 0:31
  Josh: Gemini is feeling faster, but it's also feeling better, and it's also getting cheaper.
• 0:36
  Ejaaz: What's happening behind the scenes?
• 0:37
  Logan: We crossed quadrillion tokens, which comes after a trillion if you're not.
• 0:41
  Logan: You haven't thought about numbers higher than a trillion before.
• 0:44
  Logan: It's what comes after a trillion, and there's no slowdown in sight.
• 0:51
  Josh: We have an incredibly exciting episode today because we are joined by Logan
• 0:54
  Josh: Kilpatrick. Logan is the product lead working on the Gemini platform at Google DeepMind.
• 0:59
  Josh: We have an exciting announcement to break right here today with Logan,
• 1:02
  Josh: which is the announcement of a model that we previously knew as Nano Banana.
• 1:06
  Josh: The reality is this is a brand new image generation model coming out of Google
• 1:10
  Josh: and you can access it today.
• 1:12
  Josh: So Logan, tell us about this brand new model and what we need to be excited about.
• 1:15
  Logan: Yeah, for people who are not chronically online and seeing all the tweets and everything like that,
• 1:20
  Logan: part of the excitement has been in over the last, I think, like six months,
• 1:23
  Logan: we've seen the emergence of like native image
• 1:26
  Logan: generation editing models historically um you
• 1:30
  Logan: would see models that could actually do a really good job of generating
• 1:33
  Logan: images um they were usually like tend to be like very like beautiful um aesthetic
• 1:38
  Logan: images the challenge was like how do you actually use these things in practice
• 1:42
  Logan: to do a lot of stuff that's where this editing capability is really helpful
• 1:46
  Logan: um and then so we started to see these models that can actually edit images
• 1:49
  Logan: if you could provide an image,
• 1:51
  Logan: it would, and then prompt it, it would, it would actually change that image.
• 1:55
  Logan: What's really interesting though, is this fusion of those two capabilities with
• 1:59
  Logan: the actual base intelligence of the Gemini model.
• 2:02
  Logan: And there's a lot of really cool ways in which this like manifests itself.
• 2:05
  Logan: And we'll look at some examples of this.
• 2:06
  Logan: But it's this benefit of the world knowledge. The model is like smart.
• 2:10
  Logan: So as you, as you ask it to do things, and as you ask it to make changes,
• 2:14
  Logan: it doesn't just like take what you're saying at face value.
• 2:17
  Logan: It takes what you're saying in the context of its understanding of the world
• 2:21
  Logan: and its understanding of physics, its understanding of light and all this other
• 2:25
  Logan: stuff, and it makes those changes.
• 2:26
  Logan: So it's not just blindly making edits or generations. They're actually grounded
• 2:31
  Logan: in reality and context in which that's useful.
• 2:34
  Logan: And we can look at some examples of this. My favorite thing is actually this editing capability.
• 2:40
  Logan: So this is an AI studio, and we'll have a link somewhere, hopefully in the show
• 2:43
  Logan: notes, that will let us do this. My friend Amar, who...
• 2:47
  Logan: Is on our team and drives all of our design stuff um build
• 2:51
  Logan: this and it's called past forward and what you can do is you can
• 2:54
  Logan: put in an image of yourself and it'll regenerate a
• 2:57
  Logan: version of yourself um in this
• 2:59
  Logan: sort of like polaroid-esque vibe um following all the different trends from
• 3:05
  Logan: the last 10 or 20 uh 30 years um so if you look at this example this is from
• 3:12
  Logan: me from the 1950s and And I'm sure I have a picture of my dad from the 1950s
• 3:16
  Logan: somewhere or my grandpa who looks somewhat similar to that.
• 3:20
  Logan: Here's me in the 1980s, which I love. Here's me.
• 3:24
  Ejaaz: Some of these facial expressions are also different. Like you're showing your
• 3:27
  Ejaaz: teeth more in some and then it's a smirk in others. That's super cool.
• 3:32
  Logan: I like this sweater. I actually have a sweater that almost looks exactly like
• 3:35
  Logan: this 1970s one, though I don't like my hair in this 1970s one.
• 3:39
  Logan: Same with the 2000s. um so one of the cool things
• 3:42
  Logan: about this uh this new model and one of the features i
• 3:45
  Logan: think folks are going to be most excited about is this character consistency which
• 3:48
  Logan: is as you took the original image um and you
• 3:51
  Logan: made the translation to this 1950s image it actually
• 3:54
  Logan: looks like me still um which is really cool so
• 3:57
  Logan: there's lots of these really interesting use cases i think we'll
• 4:00
  Logan: we'll go out with a um um like a
• 4:03
  Logan: sports card demo where you can sort of turn yourself into
• 4:06
  Logan: a you know a figurine sports card um
• 4:09
  Logan: which is really cool so lots of really interesting examples like
• 4:12
  Logan: this um and another thing you'll notice is actually the
• 4:15
  Logan: speed and this is where the underlying model is not
• 4:18
  Logan: uh the code name was nano nano banana um
• 4:21
  Logan: the actual model is built on gemini 2.5 flash which is our workhorse model it's
• 4:26
  Logan: super fast it's super efficient um it's relatively priced in the market which
• 4:30
  Logan: is awesome so you can actually use it at scale um and yeah so this model behind
• 4:35
  Logan: the scenes or for developers who people want to build with it is Gemini 2.5 flash image,
• 4:40
  Logan: which is awesome. So this is a use case that I love.
• 4:44
  Logan: And it's a ton of fun. You can do this in the Gemini app or in AI Studio.
• 4:49
  Ejaaz: I mean, as you said, the character consistency just from these examples is like astounding.
• 4:57
  Josh: I need to give a round of applause. This has been my biggest issue when I'm
• 4:59
  Josh: generating images of myself.
• 5:01
  Ejaaz: Genuinely. And Josh and I are early users of, you know, Mid Journey V1,
• 5:07
  Ejaaz: OpenAI's Image Generator as well.
• 5:10
  Ejaaz: And one of our pet peeves was it just couldn't do the most simplistic things,
• 5:14
  Ejaaz: right? We could just say, hey, keep this photo and portrait of me exactly the same.
• 5:18
  Ejaaz: But can you show me what I would look like in a different hairstyle or me holding
• 5:22
  Ejaaz: a bottle of Coca-Cola instead of this martini?
• 5:25
  Ejaaz: And it just could not do that, right? Just simple video, like photo editing.
• 5:29
  Ejaaz: Can you give us a bit of a background as to what Google did to be able to achieve this?
• 5:35
  Ejaaz: Because, you know, I've been racking my head around like, why other AI companies
• 5:38
  Ejaaz: couldn't do this? Like, what's happening behind the scenes? Can you give us a bit of insight?
• 5:43
  Logan: Yeah, that's a good question. I think this actually goes back to and I'll share
• 5:46
  Logan: another example in a second as well.
• 5:47
  Logan: But I think this goes back to this story of what happens when you build a model
• 5:52
  Logan: that has the fusion of all these capabilities together.
• 5:55
  Logan: And I was actually just This is a sort of parallel example to this,
• 5:58
  Logan: but it's another example of why building a unified model to do all this stuff
• 6:02
  Logan: and not having a separate model that doesn't have world knowledge and all these
• 6:06
  Logan: other capabilities is useful. The same thing is actually true on video.
• 6:09
  Logan: Like part of the story, and we haven't, we have a bunch of stuff coming that
• 6:14
  Logan: sort of tells this a little bit more elegantly than I will right now.
• 6:16
  Logan: But part of the story of like VO3 having this really state-of-the-art video
• 6:22
  Logan: generation capabilities, if folks have seen this, is that the Gemini models
• 6:27
  Logan: themselves have this state-of-the-art video understanding capabilities.
• 6:31
  Logan: And a very similar context, actually, on the image side, which is since the
• 6:37
  Logan: original Gemini model, we've like,
• 6:39
  Logan: with the exception of probably a couple of months in that like two and a half
• 6:42
  Logan: year time horizon, have had state-of-the-art image understanding capabilities.
• 6:46
  Logan: And I think there is this like capability transfer, which is really interesting
• 6:51
  Logan: as you go to do the generation step.
• 6:53
  Logan: And if you can fuse those two things together in the same model,
• 6:56
  Logan: you end up just being able to do things that other models aren't able to do.
• 7:00
  Logan: And this was part of the bet originally that like, why build Gemini to be the original Gemini?
• 7:06
  Logan: Gemini 1.0 model was built to be natively multimodal. It was built to be natively
• 7:10
  Logan: multimodal because the belief at the time,
• 7:13
  Logan: and I think this is turning out to be true is that like that's on the path to
• 7:16
  Logan: AGI is that you combine these capabilities together and like similar to what
• 7:20
  Logan: humans are able to do we have this fusion of all these capabilities in a single
• 7:24
  Logan: entity just like these models should be able to do.
• 7:28
  Ejaaz: Wow. So if I were to distill what you just said here, Logan,
• 7:32
  Ejaaz: the way you've trained Gemini 2.5 or all future Google Gemini models is it's
• 7:39
  Ejaaz: in a very multimodal fashion.
• 7:41
  Ejaaz: So you're basically, it gets smarter in one particular facet,
• 7:45
  Ejaaz: which trains itself or has transferable capabilities to other facets,
• 7:49
  Ejaaz: whether it's image generation, video generation, or even text LLMs to some extent.
• 7:53
  Ejaaz: I just think that's fascinating.
• 7:55
  Ejaaz: I'm curious, I have one question for you, which I want to hear your take on.
• 7:59
  Ejaaz: How are you going to surface this to the regular consumer, right?
• 8:03
  Ejaaz: Because right now, you provide all of these capabilities through an amazing
• 8:07
  Ejaaz: suite, you know, called Google AI Studio.
• 8:10
  Ejaaz: But if I wanted to use this in, say, an Instagram app, or my random photo imaging
• 8:16
  Ejaaz: editing app, is this something that could be easily proved to someone or sourced?
• 8:21
  Ejaaz: Or do we need to go via some other route right now.
• 8:23
  Logan: Let me just diverge really quickly, which is if any of the researchers who I
• 8:29
  Logan: work with are watching this, they will tell me, they'll make sure that I know
• 8:32
  Logan: that capability transfer that we just talked about, you oftentimes don't get that out of the box.
• 8:38
  Logan: So there is some emergence where you get a little bit of that.
• 8:42
  Logan: You do have to do, there's real true research and engineering work that has
• 8:46
  Logan: to happen to make sure that that capability fusion happens.
• 8:49
  Logan: It's not often that you just like make the model really
• 8:52
  Logan: good at one thing and then it translates oftentimes actually it's
• 8:55
  Logan: like uh it has a negative effect which is as you
• 8:58
  Logan: make the models really good at code for example you trade
• 9:01
  Logan: that off against some other you know creative writing as as a random example
• 9:05
  Logan: of this um so you have to do a lot of like active research and engineering work
• 9:09
  Logan: to make sure that you don't lose a capability as you make another one better
• 9:12
  Logan: but then ultimately they benefit as if you can make them on the same level they
• 9:16
  Logan: benefit from this interleaved capability together.
• 9:19
  Logan: To answer the question about like, where is this going to be available?
• 9:22
  Logan: The Gemini app is the place that like for by and large, most people should be going to.
• 9:26
  Logan: So if you go to Gemini.Google.com, there'll be sort of a landing page experience
• 9:30
  Logan: that showcases this new model and makes it really easy.
• 9:33
  Logan: And you can put in all your images and do tons of fun stuff like the example that I was showing.
• 9:38
  Logan: If you're a developer and you want to build something with this,
• 9:41
  Logan: in AI Studio, we have this build tab.
• 9:43
  Logan: And that's what we were just looking at as an example of one of the applets
• 9:47
  Logan: that's available in the build tab,
• 9:50
  Logan: the general essence is that all of these applets can be forked and remixed and
• 9:54
  Logan: edited and modified so that you can keep doing all the things that you want
• 9:58
  Logan: to do with the AI capability built in.
• 10:01
  Logan: So it'll continue to be powered by the same model. It'll do all that stuff, which is awesome.
• 10:06
  Logan: So there's lots of cool fusion capabilities that we have with this.
• 10:10
  Logan: Same thing with this other example that we're looking at. So if you want to
• 10:14
  Logan: go outside of this environment, we have an API.
• 10:16
  Logan: You could go and build whatever. saw if your website is, you know,
• 10:20
  Logan: AI photos.com or whatever, you could go and build with the Gemini API,
• 10:24
  Logan: use the new Gemini 2.5 flash image model to do a bunch of this stuff, which is awesome.
• 10:29
  Josh: Awesome. So while this is baking, I noticed you had another tab open,
• 10:32
  Josh: which means maybe there's another demo that you were prepared to share.
• 10:35
  Logan: There is another demo. This one I actually haven't tried yet.
• 10:39
  Logan: But it's this idea of like, how can you take a photo editing experience and
• 10:44
  Logan: make it super, super simple? So I'll grab an image.
• 10:48
  Logan: Actually, we'll take this picture, which is a picture of Demis and I.
• 10:52
  Josh: Legends.
• 10:53
  Logan: We'll put an anime filter on it and we'll see.
• 10:56
  Logan: And so this is a completely vibe coded UI experience and all the code behind
• 11:00
  Logan: the scenes is vibe coded as well.
• 11:01
  Logan: And we'll see how well this works with Demis and I.
• 11:05
  Josh: How much of the tools that you build do you find are built with vibe coding
• 11:08
  Josh: instead of just hard coding software?
• 11:11
  Josh: Are you writing a lot of this as vibe coded through the Gemini model?
• 11:14
  Logan: I think you sometimes you're able to do some of the stuff,
• 11:18
  Logan: completely vibe coded um it depends on like
• 11:21
  Logan: how specific that you want to do i do i'm like almost 85
• 11:25
  Logan: of everything that i do vibe coded somebody else on
• 11:27
  Logan: my team built this one so i don't want to misrepresent the work it could
• 11:30
  Logan: have it could have all been human programmed because we have an incredible set
• 11:34
  Logan: of engineers the general idea is how can you make this oh interesting how can
• 11:39
  Logan: you make this photoshop like experience let's go 90 or do you have suggestions
• 11:45
  Logan: what would a good filter for this be i don't know oh.
• 11:48
  Josh: Man i yeah like perhaps uh going back to
• 11:50
  Josh: the last example maybe like a a 90s film or an
• 11:53
  Josh: 80s film grain all right and i guess while we wait for that to load is there
• 11:56
  Josh: a simple way that you would describe a nano banana or this new image model to
• 12:01
  Josh: just the average person on the street who's oh look there we go we have the
• 12:04
  Josh: film grain okay so what we're watching for the people who are listening um you're
• 12:08
  Josh: retouching you can retouch parts of the image you could crop adjust there are filters to be applied
• 12:12
  Logan: I'm just clicking through buttons to be honest, I've never done it before. So it's been fun.
• 12:17
  Logan: Live demo, day one. This is the exploration you're going to get to do as a user
• 12:20
  Logan: as you play around with this.
• 12:21
  Ejaaz: Logan is vibe editing and that's what's happening. Yeah. He's experimenting.
• 12:25
  Logan: Vibe editing, which is fun. I love it. That's a great way. And the cool thing,
• 12:29
  Logan: again, is like what I love about this experience is as you're going through,
• 12:32
  Logan: oh, interesting, this one's like giving me edited outline.
• 12:36
  Josh: Oh, yeah, a little outline. This is helpful for our thumbnail generation. We do a lot of this stuff.
• 12:40
  Logan: Let's see if I can remove the background as well.
• 12:42
  Ejaaz: Oh, yeah. Let's see. I should be.
• 12:44
  Josh: If this removes the background, this is going to be trouble because this is
• 12:47
  Josh: a big feature that we use for a lot of our imagery.
• 12:49
  Logan: Hopefully. Come on. Oh, nice.
• 12:51
  Ejaaz: Oh, done.
• 12:53
  Josh: Nicely done.
• 12:54
  Ejaaz: For those of you who are listening, he's typed in, put me in the Library of
• 12:58
  Ejaaz: Congress. So we're going to hopefully see Logan.
• 13:02
  Logan: Yeah, the context on that image was that Demis and I were in the Library of the DeepMind office.
• 13:08
  Ejaaz: Oh, nice.
• 13:09
  Logan: Yeah, so that was the Library of Congress reference in my mind.
• 13:13
  Logan: But yeah, so much that you can do.
• 13:14
  Logan: Again, what I love about this experience is that as you go around and play with
• 13:19
  Logan: this stuff, if you want to modify this experience, you can do so on the left-hand side.
• 13:24
  Logan: If you say, actually, here are these five editing features that I really care
• 13:28
  Logan: about, the model will go and rewrite the code, and then it'll still be attached
• 13:32
  Logan: to this new 2.5 flash image model.
• 13:36
  Logan: So you can do all these types of cool stuff. This experience is something that
• 13:40
  Logan: I'm really excited about that we've been pushing on.
• 13:42
  Josh: Yeah, this is amazing because I myself, I do photography a lot.
• 13:45
  Josh: I was a photographer in my past life and I rely very heavily on Photoshop and
• 13:49
  Josh: Lightroom for editing, which is a very manual process.
• 13:52
  Josh: And they have these smart tools, but they're not quite like this.
• 13:54
  Josh: I mean, this saves a tremendous amount of time if I could just say,
• 13:57
  Josh: hey, realign, restrain the image, remove the background, add a filter.
• 14:01
  Josh: I think the plain English version of this makes it really approachable, but also way faster.
• 14:07
  Logan: Yeah, it is. It is crazy fast. I think about this all the time.
• 14:10
  Logan: Like there's definitely cases where you want to go deep with whatever the pro tool is.
• 14:16
  Logan: I think there's, there's actually something interesting, like on the near horizon
• 14:19
  Logan: that our team has thought a lot about, which is how you can have this experience
• 14:23
  Logan: and how you can sort of in a, in a generative UI capacity,
• 14:29
  Logan: have the experience sort of subtly expose additional detail to users.
• 14:33
  Logan: And I think about this like if you're a new you know photoshop user
• 14:36
  Logan: as an example and you show up like the chance that you're
• 14:39
  Logan: going to use all of the bells and whistles is zero like you want
• 14:42
  Logan: like the three things i want to remove a background i want to crop something
• 14:45
  Logan: whatever it is like don't actually show this all of these bells and whistles
• 14:50
  Logan: i think the exciting thing about like the progress on coding models is that
• 14:54
  Logan: in the future the challenge with the challenge with doing this in the present
• 14:58
  Logan: rather is that software is deterministic.
• 15:00
  Logan: You have to build software to build the sort of like modified version of that
• 15:04
  Logan: software for all of these different like skill sets and use cases is extremely
• 15:08
  Logan: expensive. It's not feasible.
• 15:10
  Logan: It doesn't scale to production environments. But if you can have this generative
• 15:14
  Logan: UI capability where like the model sort of knows and as you talk to the model,
• 15:18
  Logan: it realizes, oh, you might actually benefit from these other things.
• 15:21
  Logan: It can create the code to do that on the fly and expose them to you,
• 15:24
  Logan: which is really interesting.
• 15:26
  Logan: So I think there's lots of stuff that is going to be possible as the models keep getting better.
• 15:31
  Josh: This is amazing. So the TLDR on this new announcement, how would I,
• 15:36
  Josh: if I were to go explain to my friend what this does, why this is special,
• 15:39
  Josh: how would you kind of sell it to me?
• 15:41
  Logan: The best image generation and editing model in the world, 2.5 flash image or
• 15:47
  Logan: nano banana, whichever you prefer, is the model that can do this.
• 15:52
  Logan: And I think there's so many creative use cases where you're actually bounded
• 15:57
  Logan: by the creative tool. And I feel like this is one of these examples to me where
• 16:01
  Logan: it's like, I feel like I'm 10x more.
• 16:03
  Logan: I was literally helping my friend yesterday doing a bunch of iterations on his
• 16:08
  Logan: LinkedIn picture because it was like, you know, the background was slightly
• 16:12
  Logan: weird or something like that.
• 16:13
  Logan: And we were just like, I did like 15 iterations and now he's got a great new
• 16:17
  Logan: LinkedIn background, which is awesome.
• 16:19
  Logan: So like, there's so many like actual practical use cases where you,
• 16:23
  Logan: and I literally just like built a custom tool on the fly vibe coding in order
• 16:28
  Logan: to solve that use case, which was a ton of fun.
• 16:31
  Josh: Yeah, this is so cool. Okay, so this model, Nano Banana Gemini 2.5 Flash Image Gen, it's out today.
• 16:37
  Josh: So we'll link that in the description for people who want to try it out.
• 16:39
  Josh: I think one of my complaints for the longest time, and I've mentioned this on
• 16:41
  Josh: the show a few times, is a lot of times when I'm engaging with this incredible
• 16:45
  Josh: form of intelligence, I just have a text box.
• 16:47
  Josh: And it's up to me to kind of pull the creativity out of my own mind.
• 16:50
  Josh: And I don't get a lot of help along the way. but one of the things that you
• 16:53
  Josh: spend your time in is this thing called Google AI Studio.
• 16:56
  Josh: And I've used AI Studio a lot because it solves a problem for me that was annoying,
• 17:00
  Josh: which is just the blank text box.
• 17:02
  Josh: It kind of has a lot of prompts. It has a lot of helpers. It has a lot of guidance
• 17:05
  Josh: into helping me extract value out of the model.
• 17:08
  Josh: So what I'd love for you to do for people who aren't familiar,
• 17:10
  Josh: Logan, is just kind of explain to everyone what Google AI Studio is and why
• 17:14
  Josh: it's so important and why it's so great.
• 17:17
  Logan: Yeah, I love this, Josh. I appreciate that you like using AI Studio.
• 17:21
  Logan: Go. It is a labor of love. Lots of
• 17:25
  Logan: people across Google have put in a ton of time to make progress on this.
• 17:28
  Logan: I really want to, so I'll make a caveat, which is we have this entirely redesigned
• 17:33
  Logan: AI Studio experience that's coming very soon.
• 17:35
  Logan: I won't spoil it in this episode because it's like half faked right now.
• 17:40
  Logan: And I wish I could show, and I think actually some of the features that you
• 17:43
  Logan: might see in this UI might be slightly different at launch time than what you see here.
• 17:47
  Logan: So take this with a grain of salt. We've got a bunch of new stuff coming.
• 17:51
  Logan: And I think actually it should help with this problem that you're describing,
• 17:55
  Logan: which is as you show up to a bunch of these tools today, the onus is really
• 17:58
  Logan: on you as a user to like try to figure out what's capable,
• 18:02
  Logan: what all the different models are capable of, what even are all the different
• 18:06
  Logan: models, like all of that stuff.
• 18:08
  Logan: So at a high level, like we built AI Studio for this like AI builder audience.
• 18:13
  Logan: If you want to take AI models and actually build something with them and not
• 18:18
  Logan: just, you know, chat to AI models, this is the product that was built for you.
• 18:22
  Logan: We have a way to, in this like chat UI experience, sort of play with the different
• 18:27
  Logan: capabilities of the model, feel what's possible. What is Gemini good at? What's it not good at?
• 18:33
  Logan: What are the different tools it has access to? But as you go into iStudio,
• 18:37
  Logan: you'll see something that looks like this.
• 18:38
  Logan: You know, we're highlighting a bunch of the new capabilities that we have right now.
• 18:42
  Logan: This URL context tool, which is really great for information retrieval,
• 18:45
  Logan: this native speech generation capability, which is really cool.
• 18:48
  Logan: Folks have used notebook lm and you want to build a
• 18:51
  Logan: notebook lm like experience um we have
• 18:54
  Logan: an api for for people who want to build something like that and
• 18:57
  Logan: we have this live audio to audio dialogue experience where
• 19:00
  Logan: you can share a screen with the model and talk to it and it can see the things
• 19:04
  Logan: that you see and engage with it of course we have our native image generation
• 19:08
  Logan: and editing model the old version 2.0 flash now the new version 2.5 flash um
• 19:13
  Logan: and lots of other stuff that's available as you sort of experience what these models are capable of.
• 19:18
  Logan: Um, so really this playground experience is one version. We have this chat prompt
• 19:22
  Logan: on the left-hand side. We have this stream.
• 19:25
  Logan: This is where you can talk to Gemini and sort of share your screen.
• 19:28
  Logan: And, um, actually you can like show it things on the webcam and be like, what's this?
• 19:33
  Logan: How do I use this thing? You can do this on mobile as well, which is really cool.
• 19:37
  Logan: We have this generative media experience where like, if you want to build things
• 19:40
  Logan: with, we have a music model, we have a VO, which is our video generation model.
• 19:45
  Logan: We have all the text to speech stuff, which is really cool.
• 19:48
  Logan: As I overwhelm people with so much stuff that you can do in AI Studio.
• 19:51
  Logan: The sort of key threat of all this is we built AI Studio to showcase a bunch of these capabilities.
• 19:57
  Logan: And everything you see in AI Studio has an underlying sort of API and developer experience.
• 20:02
  Logan: So if you want to build something like any of these experiences, all of this is possible.
• 20:07
  Logan: There's like no Google secret magic that's happening pretty much anywhere in AI Studio.
• 20:13
  Logan: It's all things that you could build as someone uh using a
• 20:16
  Logan: vibe coding product or you know by hand writing the
• 20:19
  Logan: code um you could build all these things and even more
• 20:22
  Logan: um and that is the perfect segue to this
• 20:24
  Logan: build tab where we're trying to help also you know actually help you get started
• 20:29
  Logan: building a bunch of stuff so you can use these templates that we have you can
• 20:32
  Logan: use a bunch of the suggestions you can look through our gallery of a different
• 20:36
  Logan: stuff um and we're really in this experience trying to help you build ai powered
• 20:41
  Logan: apps which we think is something that folks are really really excited about and,
• 20:45
  Logan: we'll have much more to share around all the ai app building stuff uh in the near future.
• 20:50
  Josh: Awesome thanks for the rundown so as i'm looking at this i'm wondering who do
• 20:54
  Josh: you think this is for what type of person should come to ai studio and tinker around here
• 20:58
  Logan: Yeah so i think you know historically and and so you'll see a little bit of
• 21:02
  Logan: this transition if you play around the product where there's some interesting
• 21:06
  Logan: edges we were originally focused on building for developers So it was built
• 21:10
  Logan: and there is like a part of the experience which like is tied to the Gemini
• 21:14
  Logan: API, which tends to be used mostly by developers.
• 21:16
  Logan: So if you go to dashboard, you can see all your API keys and check your usage
• 21:21
  Logan: and billing and things like that. By and large, though, I think the really cool
• 21:25
  Logan: opportunity of what's happening right now is this transition of like who is creating software.
• 21:31
  Logan: And this is, I think, principally enabled by Vibe Coding.
• 21:34
  Logan: And because of that, like we've recentered ourselves to be really focused on
• 21:38
  Logan: this AI builder persona, which is like people who want to build things using AI tools.
• 21:44
  Logan: Also, people who are trying to build AI experiences, we think is going to be
• 21:47
  Logan: the market that creates value for the world.
• 21:51
  Logan: So if you're excited about all the things that you're seeing,
• 21:53
  Logan: if you want to build things, AI Studio is very much like a builder first platform.
• 21:57
  Logan: If you're just looking for like a great everyday AI assistant product,
• 22:01
  Logan: you, you know, want to get help on coding questions or homework or life advice
• 22:08
  Logan: or all that type of stuff,
• 22:09
  Logan: the Gemini app is the right place for this. It's very much like a.
• 22:14
  Logan: DAU type of product where like you come back and it has memory and personalization
• 22:19
  Logan: and all this other stuff, um, which makes it really great as like an,
• 22:22
  Logan: uh, as a, as an assistant to help you in your life versus AI studio.
• 22:27
  Logan: The artifact is like, we help you create something and then you go put that
• 22:31
  Logan: thing into the world in some way.
• 22:32
  Logan: Um, and you don't necessarily need to come back and use it every day.
• 22:36
  Logan: You use it whenever you want to build something.
• 22:37
  Ejaaz: It's funny. Um, I'm dating myself a bit here,
• 22:40
  Ejaaz: but I remember when I first booted up a PC and I loaded up Microsoft Office
• 22:45
  Ejaaz: and I just had access to all these different wonderful applications that were
• 22:49
  Ejaaz: at the time super new or within one suite.
• 22:52
  Ejaaz: This kind of feels like that moment for AI.
• 22:55
  Ejaaz: And you might not take that as a compliment because it's a completely different
• 22:58
  Ejaaz: company, but it was what I built my childhood off of and my fascination with computers.
• 23:03
  Ejaaz: So I appreciate this and I love that it's this massively like cohesive experience.
• 23:08
  Ejaaz: But kind of zooming out, Logan, I was thinking a lot about Google AI and what
• 23:14
  Ejaaz: that means to me personally.
• 23:16
  Ejaaz: I have to say it's the only company that I think beyond an LLM.
• 23:22
  Ejaaz: And what I mean by that is when I think of Google AI, I don't just think of Gemini.
• 23:27
  Ejaaz: I think of the amazing image gen stuff that you have.
• 23:30
  Ejaaz: I think of the amazing video outputs that you guys have. I think of the text-to-voice
• 23:36
  Ejaaz: generation that you just demoed and all those kinds of things.
• 23:39
  Ejaaz: I remember seeing this advert that appeared on my timeline.
• 23:43
  Ejaaz: And I remember thinking, wow, this must be the new GTA. Then I was like,
• 23:47
  Ejaaz: no, no, that's Florida. That's Miami.
• 23:51
  Ejaaz: No, people are doing wild stuff. That's an alien. Hang on a second. This can't be real.
• 23:55
  Ejaaz: And then I learned that it was a Google VO3 generation of an advert for Kalshi,
• 24:00
  Ejaaz: which is like this, you know, prediction markets situation.
• 24:04
  Ejaaz: And I remember thinking, how on earth have we got to AI generated video that
• 24:09
  Ejaaz: is this high quality and this high fidelity?
• 24:12
  Ejaaz: I think in my mind, VO3 has kind of like killed the VFX studio.
• 24:17
  Ejaaz: It's kind of killed a lot of Hollywood production studios as well.
• 24:23
  Ejaaz: Give me a breakdown and insight into how you built or how you guys built VO3
• 24:28
  Ejaaz: and what that means for the future of movie video production and more
• 24:33
  Logan: Yeah that's a great question i think there's something really interesting along
• 24:37
  Logan: these threads and and not to not to push back on the notion that it's killing
• 24:42
  Logan: hollywood because i think i think there is like um,
• 24:45
  Logan: I think it's an interesting conversation. The way that I have seen this play
• 24:49
  Logan: out and the great example of this, that folks have seen Flow,
• 24:53
  Logan: which is our sort of like creative video tool.
• 24:58
  Logan: And if you're using VO and you want to sort of get the most out of VO,
• 25:01
  Logan: Flow is the tool to do that.
• 25:02
  Logan: If you see lots of like the creators who are building, you know,
• 25:06
  Logan: minute long videos using VO and it's like this really cohesive story and it
• 25:10
  Logan: has like a clear visual identity, similar to what you'd get from like a.
• 25:15
  Logan: Probably not the extent of a Hollywood
• 25:17
  Logan: production, but like somebody thoughtfully choreographing a film.
• 25:21
  Logan: Flow is the product to do that. And actually interesting, like Flow was built
• 25:25
  Logan: in conjunction with filmmakers.
• 25:28
  Logan: And I think that's actually like there is, and I feel this way about vibe coding as well.
• 25:32
  Logan: And it's this thought experiment that I'm always running through in my head,
• 25:36
  Logan: which is, yes, I think AI is like raising the bar forever or it's raising the
• 25:41
  Logan: floor for everyone. We're like, now everyone can create.
• 25:44
  Logan: What does that mean for people who have expertise?
• 25:49
  Logan: And I think in most cases, what it means is actually the value of your expertise
• 25:52
  Logan: continues to go up. And like, this is my personal bet.
• 25:55
  Logan: And I don't know how much this tracks to like everyone else's worldview.
• 25:58
  Logan: My personal bet is that expertise in the world where the floor is lifted for
• 26:02
  Logan: everyone across all these dimensions is actually more important because there was something about,
• 26:08
  Logan: and I think like video production is a great example for me because I would
• 26:12
  Logan: never have been able to make a video.
• 26:14
  Logan: Like it's not in the cards, like for my skillset, my creative ability,
• 26:18
  Logan: my financial ability, like I will never be able to make a video.
• 26:21
  Logan: I can make things with VO. Um, and now I'm like a little bit closer to imagining
• 26:26
  Logan: like, okay, if I'm serious about this, I need to go out and like actually engage with people.
• 26:32
  Logan: And I've like, sort of, it's like whetted my appetite in a way that I don't
• 26:36
  Logan: think I would. It was just like too far in a way.
• 26:38
  Logan: And I think software is another example where Vibe Coding, if you were to pull
• 26:43
  Logan: a random person off the street and you start talking to them about coding and
• 26:46
  Logan: seeing C++ and deploying stuff and all this, they're like, brain turns off, not interested.
• 26:53
  Logan: I don't want to learn to code. That's not cool. It's not fun. It sounds horrible.
• 26:57
  Logan: And then Vibe Coding rolls around and it's like, oh, wait, I can actually build
• 27:00
  Logan: stuff. And like, yeah, I don't really need to understand all of the details.
• 27:04
  Logan: But there's still a limit to what I can build and who is actually well positioned
• 27:09
  Logan: to help me take the next step.
• 27:11
  Logan: Like I, you know, vibe code something. I'm like, this is awesome.
• 27:14
  Logan: I share with my friends. They all love it.
• 27:17
  Logan: I want to, you know, go build a business around this thing that I vibe coded.
• 27:21
  Logan: There's still a software engineer that needs to help make that thing actually
• 27:24
  Logan: happen. So if anything, it's like it's increasing this.
• 27:28
  Logan: I mean, on the software side, there's this infinite demand for software,
• 27:31
  Logan: and it's increasing the total addressable market of like what software engineers
• 27:36
  Logan: need to help people build.
• 27:37
  Logan: I think there'll be something similar on the video side.
• 27:40
  Logan: You know, there will be downsides to AI technology in some ways.
• 27:43
  Logan: I think there is like as the technology shift happens, there is some amount
• 27:48
  Logan: of disruption that's taking place.
• 27:50
  Logan: And like someone's workflow is being disrupted. But I do think there's this
• 27:53
  Logan: really interesting thread to pull on, which is my hope is that it actually ends
• 27:57
  Logan: up creating more opportunity for the experts and the specialists.
• 28:01
  Ejaaz: So it sounds like you're not saying VFX Studio teams are going to be replaced by software engineers,
• 28:08
  Ejaaz: but rather that team in itself will become more adept at using these AI tools
• 28:14
  Ejaaz: and products to kind of enhance their own skill set beyond what it is today. Is that right?
• 28:19
  Logan: Yeah, yeah. And I think we've seen this already play out in some ways, which is interesting.
• 28:23
  Logan: I think code is a little bit wider distribution than perhaps the VFX.
• 28:29
  Logan: And it's VFX also in a space that I'm less familiar with personally.
• 28:33
  Logan: But yeah, I think this is likely what is going to play out if I had to guess and bet.
• 28:39
  Ejaaz: Can you help us understand how a product like VO3 gets used beyond just like
• 28:47
  Ejaaz: the major Hollywood productions?
• 28:49
  Ejaaz: Stuff, right? Because I've seen a bunch of these videos now.
• 28:52
  Ejaaz: And I'll be honest with you, Logan, it's scary how realistic this stuff is, right?
• 28:57
  Ejaaz: It's like from a high quality triple A game demo, all the way to something that
• 29:03
  Ejaaz: is shot like in an A24 film, you know, the scenes, the cuts, the changes.
• 29:09
  Ejaaz: I think it's awesome. I'm wondering whether that goes beyond entertainment in
• 29:13
  Ejaaz: any way. Do you have any thoughts or ideas there?
• 29:15
  Logan: Yeah, that is interesting. I think one of the ones that is like related to,
• 29:20
  Logan: it's sort of one skip away from video generation itself, which was Genie,
• 29:26
  Logan: which was our sort of world simulation work that was happening.
• 29:29
  Logan: I think if folks haven't seen this, go look up Genie 3 and you can see a video.
• 29:33
  Logan: It's mind blowing. It's like a fully playable game world simulation.
• 29:36
  Logan: You can like prompt on the go and this environment will change.
• 29:40
  Logan: You can control it on your keyboard similar to a game. I think that work translates
• 29:45
  Logan: actually really well to robotics, which is cool.
• 29:47
  Logan: So as you like one of the if folks aren't familiar with this,
• 29:51
  Logan: like one of the principal reasons we don't just have robots walking around everywhere.
• 29:55
  Logan: And the reason why we have LLMs that can actually do lots of useful stuff is
• 30:00
  Logan: it's this data problem, which is like there's lots of, you know,
• 30:04
  Logan: text data and other data that's like representative of the intelligence of humans
• 30:09
  Logan: and all this stuff that's available.
• 30:11
  Logan: There's actually not a lot of data that is useful for making robotics work.
• 30:14
  Logan: And I think VO could be part of, or like generally that sort of segment of video
• 30:20
  Logan: generation and this like physics understanding and all that other stuff,
• 30:24
  Logan: I think could be really helpful in actually making the long tail of robotics use cases work.
• 30:29
  Logan: Then I can finally have a robot that will fold my laundry so that I don't need
• 30:33
  Logan: to spend my time doing that.
• 30:34
  Logan: But that's my like outside of entertainment bet as far as like where that use
• 30:40
  Logan: case ends up creating value in the world.
• 30:42
  Ejaaz: With VO3, the goal is to enable humans to become a better version of themselves,
• 30:48
  Ejaaz: a 10x, 100x better version of themselves using these different tools.
• 30:51
  Ejaaz: So in the example of a VFX studio, you can now kind of like create much better movies.
• 30:56
  Ejaaz: How does that apply for Genie 3 exactly, right?
• 30:59
  Ejaaz: You gave the example of like being able to create simulated environments,
• 31:03
  Ejaaz: but that's to train these robots. That's to train these models.
• 31:07
  Ejaaz: What about us? What about the flesh humans that are out there?
• 31:10
  Ejaaz: Can you give us some examples about where this might be applied or used?
• 31:14
  Logan: Yeah, that's a good example. I mean, the robot answer is like the robots will
• 31:19
  Logan: be there to help us, which is nice.
• 31:20
  Logan: So hopefully there's a bunch of stuff that you don't want to do that you'll
• 31:25
  Logan: be able to get your robot to do.
• 31:27
  Logan: Or there's like industries that are like dangerous for humans to operate in
• 31:31
  Logan: where it's like if you can sort of do that simulation without needing to collect
• 31:35
  Logan: a bunch of human data to do those things, I could see that being super valuable.
• 31:38
  Logan: I think my initial reaction to the Genie use case, like I could see lots of,
• 31:46
  Logan: actually, the two that come to mind is like one entertainment I think will be cool.
• 31:51
  Logan: Humans want to be entertained. It's a story as all this time.
• 31:54
  Logan: I think there will be some entertainment value of a product experience like Genie.
• 31:59
  Logan: Um, I think the other one is actually back to a bunch of use cases where you'd
• 32:05
  Logan: actually want robotics to be able to do some of that work that don't yet, uh,
• 32:10
  Logan: the robot product experience, like isn't actually there.
• 32:14
  Logan: Um, this could be things like, you know, mining or like heavy industries,
• 32:19
  Logan: things like that, where like, there's actually like a safety aspect of like,
• 32:22
  Logan: how can you do these like realistic simulation training experiences, um,
• 32:27
  Logan: in order to make sure that like you're You don't have to like physically put
• 32:30
  Logan: yourself in harm's way in order to like understand the bounds or like the failure cases,
• 32:35
  Logan: like disaster recovery, things like that, where it would be you don't want to
• 32:41
  Logan: have to show up at a hurricane the first time to like really understand what
• 32:45
  Logan: the environment could be like.
• 32:47
  Logan: And like being able to do those types of simulations is interesting and building
• 32:51
  Logan: software deterministically to solve that problem would actually be really difficult
• 32:55
  Logan: and expensive and like probably isn't a large market that lots of companies are going to go after.
• 33:01
  Logan: But if you have this model that has really great world knowledge,
• 33:03
  Logan: you can throw all these random variables at it and like sort of do that type
• 33:07
  Logan: of like training and simulation.
• 33:08
  Logan: So yeah, it's perhaps an interesting use case.
• 33:12
  Logan: I don't know if there's actually a plan to use it for things like that,
• 33:15
  Logan: But those are things that come to mind.
• 33:18
  Josh: This is something I've been dying to ask you about because this is something
• 33:21
  Josh: that I've been fascinated by.
• 33:22
  Josh: When I watched the Genie 3 demo for the first time, it just kind of shattered
• 33:25
  Josh: my perception of where we were at because you see it work.
• 33:30
  Josh: And I saw this great demo where someone was painting the wall.
• 33:32
  Josh: We actually filmed an entire episode about this and it retained all of the information.
• 33:36
  Josh: And one theme, as I'm hearing you describe these things, as I'm hearing you
• 33:39
  Josh: describe VO3, Genie 3, you are building this deep understanding of the physical world.
• 33:44
  Josh: And I can't help but notice this trend like you are just starting
• 33:47
  Josh: to understand the world more and more and I could see this when
• 33:50
  Josh: it comes to making games as an example where like a lot
• 33:53
  Josh: of people were using genie 3 to just make these like not necessarily
• 33:56
  Josh: games but virtual worlds that you can walk around and interact with and I'm
• 33:59
  Josh: wondering if you could just kind of share the long-term reasoning why because
• 34:02
  Josh: clearly there's a reason there's a lot of value to it is it from being able
• 34:06
  Josh: to create maybe artificial data for robots if you can emulate the physical world
• 34:09
  Josh: you can create data to train these robots is it because it creates great experiences
• 34:12
  Josh: like perhaps we'll AAA design studios using Genie 5 to make AAA games like Grand Theft Auto.
• 34:18
  Josh: I'm curious the reasoning behind
• 34:19
  Josh: this like urge to understand the physical world and emulate it even.
• 34:24
  Logan: I had a conversation with Demis about this who's our CEO at DeepMind and someone
• 34:30
  Logan: who's been pushing on this for a long time.
• 34:32
  Logan: I think a lot of this goes back to like there's two dimensions.
• 34:35
  Logan: It goes back to like the original ethos of like why DeepMind was created and
• 34:39
  Logan: a bunch of the work the initial work that was happening in DeepMind around reinforcement learning.
• 34:44
  Logan: If folks haven't seen this, one of the challenges of, again,
• 34:48
  Logan: making AI work is that you need this flywheel of continuing to iterate and you
• 34:55
  Logan: need a reward function, which is what is the actual outcome that you're trying to achieve?
• 35:03
  Logan: And the thing that's interesting about these simulated environments is it's
• 35:07
  Logan: really easy to have a constrained environment.
• 35:10
  Logan: World, and it's really easy to also, or not maybe really easy,
• 35:15
  Logan: is overly ambitious. It's possible to define a simple reward function and then
• 35:21
  Logan: actually infinitely scale this up.
• 35:23
  Logan: And the opposite example of this, if folks have saw there was some work a very
• 35:29
  Logan: long time ago, and this is in the AI weeds, but there was this hand,
• 35:33
  Logan: this physical hand that could like robotic hand that could manipulate a Rubik's cube.
• 35:39
  Logan: And they were using AI to like help try to solve this Rubik's cube.
• 35:43
  Logan: And the, the, again, the analogy of why this, of why Genie and some of this
• 35:47
  Logan: work is so interesting is if you were to go and try to like,
• 35:50
  Logan: Hey, we need all the data to go and try to make this little hand,
• 35:54
  Logan: physical robotic hand, be able to do this.
• 35:57
  Logan: It's actually really challenging to scale that up. You need to go and build a bunch of hands.
• 36:01
  Logan: You need to like, what happens when the rubik's cube drops you need
• 36:04
  Logan: to have some system to like go and pick it back up and you just like go
• 36:07
  Logan: through the long tail of this stuff the hand probably can't run
• 36:10
  Logan: 24 hours a day like there's all these challenges with getting um the the like
• 36:15
  Logan: data in that environment to scale up um and these virtual environments don't
• 36:20
  Logan: have this problem which is if you can emulate and like self-driving cars is
• 36:24
  Logan: another example of this like again for folks who aren't familiar lots of you You know,
• 36:28
  Logan: there's lots of real world data that's involved in self-driving cars.
• 36:31
  Logan: There's also lots of simulated environments where they've built simulations of the world.
• 36:35
  Logan: And this is how they can get like a thousand X scale up of this like data understanding
• 36:41
  Logan: is by having these simulated environments.
• 36:44
  Logan: Robotics will be exactly the same. If you want robotics to work,
• 36:47
  Logan: it's almost 100% true that you're going to have to have these simulated environments
• 36:51
  Logan: where the robot can fall down the stairs a thousand times.
• 36:56
  Logan: And that's okay because it's a simulated environment and it's not actually going
• 36:59
  Logan: to fall down your stairs.
• 37:01
  Logan: So I think Genie is, there is definitely like an entertainment aspect to it.
• 37:05
  Logan: I think it's more so going to be useful for this like simulated environment
• 37:10
  Logan: to help us not have to do things in the real world and, but still have like
• 37:15
  Logan: a really good proxy of what will happen in the real world when we do them.
• 37:19
  Ejaaz: That's pretty funny. I spent the weekend watching the World Robot Olympics and
• 37:25
  Ejaaz: there was some very real fails and crashes of these robots, which is pretty funny.
• 37:32
  Ejaaz: Okay, so when I think of Genie, I think that it blows my mind because I still
• 37:38
  Ejaaz: can't get my head around how it predicts what I'm going to look at.
• 37:42
  Ejaaz: I remember seeing this demo of someone just taking a simple video of them walking and,
• 37:48
  Ejaaz: you know, it was like a rainy day on a gravel path and they stuck that into
• 37:53
  Ejaaz: Genie 3 and they could look down and see their reflection in the puddle.
• 37:58
  Ejaaz: So the physics was astoundingly accurate and astute.
• 38:02
  Ejaaz: Can you give us a basic breakdown of how this works?
• 38:05
  Ejaaz: Is this like a real engine, game engine, like happening in the background?
• 38:09
  Ejaaz: Or is there something more deeper happening? Like help us understand.
• 38:13
  Logan: My intuition, and we can gut check this with folks on the research side to make
• 38:17
  Logan: sure that I'm not, I'm not fabricating my intuition.
• 38:21
  Logan: But if folks have an intuition as far as like how next token prediction works,
• 38:25
  Logan: which is at some given, like if you're looking through a sentence of text,
• 38:29
  Logan: for each word in that sentence, there's a
• 38:32
  Logan: distribution uh between like zero
• 38:35
  Logan: and one basically of like how likely that
• 38:38
  Logan: word was to be the next word in the sequence um and
• 38:42
  Logan: if you look the and if you like look through this is like the basic
• 38:45
  Logan: principle of llms um this is why you get like the you know if you're to ask
• 38:49
  Logan: the same question multiple times um the the llm will inherently perhaps give
• 38:55
  Logan: you a different answer And that's why like small changes in the inputs to LLMs
• 38:59
  Logan: actually change this because like, again, it's this distribution.
• 39:03
  Logan: So like if you make one letter difference, it perhaps like puts you on a like
• 39:08
  Logan: a branching trajectory that looks very different than what the original output
• 39:14
  Logan: that you got from the model.
• 39:16
  Logan: Similar similar like rough approximation of this just like much more computationally
• 39:23
  Logan: difficult and i think they use a bunch of um architectural differences that
• 39:27
  Logan: sort of it's not truly next token prediction that's happening for the sort.
• 39:30
  Ejaaz: Of like pixels colors bunch of other things yeah
• 39:33
  Logan: Exactly yeah so it's like you can like roughly map
• 39:36
  Logan: the mental model of like as the as a model
• 39:38
  Logan: looks down or as like the figure looks down
• 39:41
  Logan: in some in some environment like again it has
• 39:44
  Logan: all this like context of the state of the world but then it also knows like
• 39:49
  Logan: what are the pixels that are preceding it etc etc it like loosely is doing this
• 39:54
  Logan: like next next pixel prediction you could you could sort of approximate with
• 39:58
  Logan: um that's happening at the at the genie level which is which is an interesting way to think about it.
• 40:04
  Josh: So ijaz one of the things you were mentioning was that um it's happening much
• 40:07
  Josh: faster right and it's happening presumably much cheaper because now i heard this crazy stat.
• 40:11
  Josh: You're at like 500, hundreds of trillions of tokens per month that is being
• 40:14
  Josh: pushed out by Gemini. It's unbelievable.
• 40:16
  Josh: And I want to get into the kind of infrastructure that enables this because
• 40:20
  Josh: Gemini is feeling faster, but it's also feeling better. And it's also getting cheaper.
• 40:26
  Josh: And behind you earlier in the show, you mentioned you have a TPU.
• 40:30
  Josh: I understand TPUs are part of this solution. And I want you to kind of just
• 40:34
  Josh: walk us through how this is happening.
• 40:36
  Josh: How are we getting these quality and improvements across the board?
• 40:39
  Josh: And what type of hardware or software is enabling that to happen.
• 40:42
  Logan: I think like one, you have to give credit to like all of these infrastructure
• 40:46
  Logan: teams across Google that are making this happen.
• 40:48
  Logan: If you think, and I think about this a lot, like what is Google's differentiated advantage?
• 40:54
  Logan: What does our expertise lend us well to do in the ecosystem?
• 40:58
  Logan: What are the things we shouldn't do because of that? What are the things we
• 41:00
  Logan: should do because of that? It's something I think about as somebody who builds products.
• 41:04
  Logan: One of the things that I always come back to is our infrastructure.
• 41:08
  Logan: And like the thing Google has been able to do time and time again.
• 41:12
  Logan: Scale up multiple products to billions
• 41:14
  Logan: of users, have them work with high reliability, et cetera, et cetera.
• 41:19
  Logan: And that's like a uniquely difficult problem. It's a even more difficult problem
• 41:24
  Logan: to do in the age of AI where like the software is not deterministic.
• 41:28
  Logan: The sort of compute footprint required to do these things is really difficult.
• 41:33
  Logan: The models are a little bit tricky and finicky to work with sometimes.
• 41:37
  Logan: So again, like our infrastructure teams have done an incredible job making that scale up.
• 41:42
  Logan: I think the stat was IO 2024, we were doing roughly 50 trillion tokens a month.
• 41:49
  Logan: IO 2025, I think it was like 480 trillion tokens a month, if I remember correctly.
• 41:56
  Logan: And just a month or two later, and this was in the conversation I had with Demis,
• 42:01
  Logan: we crossed a quadrillion tokens, which comes after a trillion,
• 42:06
  Logan: if you're not haven't thought about numbers higher than a trillion before
• 42:09
  Logan: it's what comes after a trillion um and there's
• 42:12
  Logan: no slowdown in sight and like i think this is just a great
• 42:15
  Logan: reminder of like um so many
• 42:17
  Logan: of these ai like markets and
• 42:21
  Logan: product ecosystems is still so early and there's
• 42:24
  Logan: this massive expansion i think about in my own life like how
• 42:27
  Logan: much ai do i really have in my life helping me like not
• 42:30
  Logan: really that much on the margin it's like you know maybe
• 42:33
  Logan: tens of millions of tokens a
• 42:35
  Logan: month maximum and like you think
• 42:38
  Logan: about a future where there's like billions of tokens being spent on a monthly
• 42:42
  Logan: basis in order to help you in whatever you're doing in your professional life
• 42:45
  Logan: and your work and your personal life whatever it is there's we're still so early
• 42:50
  Logan: and tpus are a core part of that because it allows us to like um control every
• 42:56
  Logan: layer of the hardware and software.
• 42:59
  Logan: Delivery all the way to the actual like silicon that the model is running on
• 43:04
  Logan: and we can do a bunch of optimizations and customizations,
• 43:08
  Logan: other people can't do because they don't actually control the hardware itself
• 43:11
  Logan: and there's some good examples of the things that this enables
• 43:14
  Logan: one of them is um you know we've been at the Pareto
• 43:17
  Logan: frontier from a cost performance perspective for
• 43:20
  Logan: a very long time and again if folks aren't familiar the Pareto frontier is this
• 43:23
  Logan: like trade-off of cost and intelligence and you want to be on the highest intelligence
• 43:28
  Logan: lowest cost um and we've been sitting on that for for you know basically the
• 43:33
  Logan: entirety of the Gemini life cycle um so far which is really important so people
• 43:37
  Logan: get a ton of value from the Gemini models.
• 43:40
  Logan: Another example of this is long context. Again, if folks are familiar,
• 43:44
  Logan: there's a limit on like how many tokens you can pass to a model at a given time.
• 43:49
  Logan: Gemini's had a million or 2 million token context windows since the initial
• 43:54
  Logan: launch of Gemini, which has been awesome.
• 43:56
  Logan: And there's a bunch of research showing we could scale that all the way up to
• 43:59
  Logan: 10 million if we wanted to.
• 44:01
  Logan: And that is like a core infrastructure enabled thing.
• 44:05
  Logan: Like research, There's a lot of like really important research to make that
• 44:08
  Logan: work and make that possible.
• 44:10
  Logan: But it's also really difficult on the infrastructure side. And you have to be
• 44:13
  Logan: willing to do that work and pay that price.
• 44:16
  Logan: And it's a beautiful outcome for us because we have the infrastructure teams
• 44:19
  Logan: that have the expertise to do this.
• 44:21
  Josh: Okay, Logan, one quadrillion tokens. That's a big number.
• 44:25
  Josh: We need to talk about this for a little bit because that is an outrageously,
• 44:28
  Josh: mind-bendingly big number.
• 44:30
  Josh: And when I hear you say that number, I think I'm reminded of Jevin's Paradox
• 44:34
  Josh: for people who don't know. it's increased technological efficiency in using
• 44:37
  Josh: a resource which can lead to higher total consumption of that resource.
• 44:40
  Josh: So clearly with these cool new TPUs, this vertically integrated stack you've
• 44:43
  Josh: built, you are able to generate tokens much more cheaply and produce a lot more
• 44:47
  Josh: of them. Hence the one quadrillion tokens.
• 44:49
  Josh: Do you see this trend continuing? Is there going to be a continued need to just
• 44:53
  Josh: produce more tokens or will it eventually be a battle to produce smarter tokens?
• 44:58
  Josh: I guess the question I'm asking is the quality of the token more important than
• 45:03
  Josh: the amount of the tokens?
• 45:04
  Josh: And do you see a limit in which the quantity of the tokens starts to like kind
• 45:09
  Josh: of go off of a cliff in terms of how valuable it is?
• 45:12
  Logan: Yeah, I could buy that story. And some of this is and it's something that's
• 45:17
  Logan: actually super top of mind for our teams on the like Gemini model side is around
• 45:21
  Logan: this whole idea of like thinking efficiency,
• 45:23
  Logan: which is like ideally you want
• 45:25
  Logan: to get to the best answer using the limited amount of thoughts possible.
• 45:28
  Logan: Same thing with humans like ideally like you're the example
• 45:32
  Logan: of like you're taking a test you want to as you know
• 45:34
  Logan: the shortest number of mental hops possible to get you to the answer.
• 45:37
  Logan: Of whatever the question was is ideally what you want you don't want
• 45:40
  Logan: to have to just like think for an hour to answer one question
• 45:43
  Logan: um and there's yeah there's a bunch
• 45:46
  Logan: of odd parallels in that world to like models and
• 45:49
  Logan: and humans doing this approach um so i
• 45:52
  Logan: do think thinking efficiency is top of mind you don't want to just
• 45:54
  Logan: like use tokens for the sake of tokens um i
• 45:58
  Logan: think even if we were to like 10x reduce the
• 46:01
  Logan: number of tokens required which would be like awesome and
• 46:04
  Logan: would be like a great innovation the models are like much more
• 46:07
  Logan: token efficient i do think there's like a um a
• 46:11
  Logan: pretty low ceiling to how far that will be able to go
• 46:14
  Logan: specifically because of this like next
• 46:17
  Logan: token prediction paradigm of like how the models
• 46:21
  Logan: actually approach solving problems using using
• 46:25
  Logan: like the token as a unit so it's
• 46:28
  Logan: not clear to me that you'll be able to just like you know a thousand x reduce
• 46:32
  Logan: the amount of tokens required to solve a problem i think it probably looks much
• 46:35
  Logan: more like 10x or something like that and then the there'll be a 10x reduction
• 46:39
  Logan: in the number of tokens required to solve a problem and there'll be a 10 000
• 46:42
  Logan: x increase in the total amount of ai and and sort of token consumption in the world.
• 46:47
  Logan: So I think you probably, even if we made that reduction happen,
• 46:51
  Logan: I think the graph still looks like it's going up and to the right for the most part.
• 46:55
  Josh: It still keeps going. There is no wall. We have virtual data to train models
• 46:59
  Josh: on. We have tons of new tokens coming into play.
• 47:02
  Josh: There's another question I wanted to ask, which is just a personal question
• 47:06
  Josh: for you, which is a feature that, because I find when a lot of people leave
• 47:10
  Josh: comments on the show and they talk about their experience with AI,
• 47:13
  Josh: a lot of them are just using like ChatGPT on their app or they have Grok on their phone.
• 47:17
  Josh: And I think Gemini kind of has some underrated features that don't quite get enough attention.
• 47:23
  Josh: So what I'd like for you to do is maybe just highlight one or two of the features
• 47:26
  Josh: you shipped recently that you think is criminally underrated.
• 47:29
  Josh: What should people try out that you think not enough people are using?
• 47:32
  Logan: I think the one that continues to surprise me the most is deep research.
• 47:37
  Logan: I think deep research is just like a, is the North Star for building an AI product experience.
• 47:42
  Logan: And if folks aren't familiar with this, so you can show up with,
• 47:45
  Logan: yeah, it's so, you can show up with like a pretty ill-defined question that's
• 47:50
  Logan: like very open and vague.
• 47:51
  Logan: And the model will traverse essentially across the internet hundreds or thousands
• 47:58
  Logan: of different web pages, try to accumulate enough context and then come back to you with initially,
• 48:05
  Logan: basically like a research report, could be like a 40 page report in some cases that I've seen.
• 48:10
  Logan: You might hear a 40 page report and say, that's not very useful to me because
• 48:13
  Logan: I'm not going to read 40 pages.
• 48:15
  Logan: And I'd say you and me are exactly the same because I'm not reading 40 pages either.
• 48:19
  Logan: There's a beautiful feature, again, if you've used Notebook LM,
• 48:22
  Logan: this audio overviews feature.
• 48:25
  Logan: The same thing actually exists inside of the Gemini app with deep research,
• 48:29
  Logan: which you can just press that button and then get like a, you know,
• 48:32
  Logan: 10, 15 minute podcast that sort of goes through and explains all the different
• 48:36
  Logan: research that's happened.
• 48:37
  Logan: You can, you know, listen to that on your commute or something like that or
• 48:40
  Logan: on a walk and not need to read 40 pages, which is awesome.
• 48:43
  Logan: The part of this that makes it such an interesting experience to me is,
• 48:47
  Logan: I don't know if other people have felt this before, but most AI products back
• 48:51
  Logan: to that, Josh, that like blank slate problem or that like empty chat box problem.
• 48:57
  Logan: You, you as the user of the product have to put in so much work in order to get useful stuff.
• 49:03
  Logan: I talk to people all the time who are like, yeah, I use these models and like,
• 49:06
  Logan: they're just not useful for me. And like, actually what's happening behind the
• 49:09
  Logan: scenes is the models are super capable.
• 49:11
  Logan: They're really useful. It just requires that you give the models enough context.
• 49:15
  Logan: And I think deep research, there's this new emerging like prompt engineering
• 49:19
  Logan: 2.0 is this context engineering problem where it's like, how do you get in the
• 49:24
  Logan: right information so that the model can make a decision on behalf of the user?
• 49:28
  Logan: And I think deep research is this really nice balance of going and doing this
• 49:33
  Logan: context engineering for you,
• 49:34
  Logan: bringing all that context into the window of the model, and then
• 49:37
  Logan: being able to answer what your original question was and principally showing
• 49:42
  Logan: you this like proof of work up front I think about this proof of work concept
• 49:46
  Logan: in AI all the time which is I have so much more trust in deep research because
• 49:51
  Logan: as soon as I kick off that query it's like boom it's already at like 50 web
• 49:55
  Logan: pages I'm like great because.
• 49:57
  Logan: I was never going to visit 50 webpages. Like there's pretty much nothing that I'm researching.
• 50:02
  Logan: I could be going and buying a car and I'm going to go and look at less than
• 50:05
  Logan: 50 webpages for that thing or a house.
• 50:07
  Logan: I'm looking at less than 50 webpages. Like I'm just, it's not in the car.
• 50:11
  Logan: At least this is maybe personal to me and other people are doing more research. I don't know.
• 50:15
  Logan: But so automatically I'm like in awe with how much more work this thing is doing.
• 50:20
  Logan: And I think there's, this is, again, this is the North Star from an AI product
• 50:24
  Logan: experience standpoint. and there's so few products that have like made that
• 50:29
  Logan: experience work and just every time I go back to deep research I'm reminded
• 50:33
  Logan: of this and that team uh crushed it so.
• 50:35
  Ejaaz: And it's not just deep research from a LLM context that is so fascinating about
• 50:42
  Ejaaz: Google AI um you guys have created some of the most fascinating tools to advance
• 50:48
  Ejaaz: science and I don't think you guys get enough flowers for what you guys have built.
• 50:54
  Ejaaz: Some of my favorites, AlphaFold 3 is crazy.
• 50:58
  Ejaaz: So, you know, this is this model
• 51:00
  Ejaaz: that can predict what certain molecular structures are going to look like.
• 51:04
  Ejaaz: And this could be applied to so many different industries, the most obvious
• 51:08
  Ejaaz: being drug design, creating cheaper,
• 51:10
  Ejaaz: more effective, curable drugs for a variety of different diseases.
• 51:15
  Ejaaz: And then I was thinking about that random model that you guys launched,
• 51:20
  Ejaaz: where apparently we could translate what dolphins were saying to us and vice versa um
• 51:26
  Ejaaz: Kind of stepping back from all of these examples, can you help me understand
• 51:30
  Ejaaz: what is Google's obsession with AI and science and why you think it's such an
• 51:35
  Ejaaz: important area to focus on?
• 51:36
  Ejaaz: Are we at a point now where we can advance science to infinity or where are we right now?
• 51:42
  Ejaaz: Are we at our chat GPT moment or do we have more to go?
• 51:45
  Logan: I'll start with a couple of cheeky answers, which Demis, who is the only foundation
• 51:51
  Logan: model lab CEO to have a Nobel prize, uh, in the science domain,
• 51:56
  Logan: which is, uh, for him, for him, chemistry, um,
• 51:59
  Logan: had this comment, which is actually really true.
• 52:02
  Logan: There's lots of people talking about this, like impact of AI on science and humanity.
• 52:07
  Logan: Um, and there's very few, uh, if not only one, um, being deep mind research
• 52:13
  Logan: lab, that's like actually doing the science work.
• 52:16
  Logan: And I think it's this like great example of like deep mind and just being in
• 52:20
  Logan: the like culture and DNA, DNA of like Demis is a scientist,
• 52:23
  Logan: all of these folks around, um, around DeepMind are scientists and they like
• 52:27
  Logan: want to push the science and, and push what's possible in this,
• 52:31
  Logan: this future of discovery,
• 52:32
  Logan: um, using our models.
• 52:34
  Logan: And I was in London a couple of weeks ago meeting with Pushmid who leads our
• 52:38
  Logan: science team and hearing about sort of like the breadth of the science that's happening.
• 52:43
  Logan: Um, and how like Dolphin Gemma is like a great, like kind of like funny example,
• 52:48
  Logan: cause it's, it's not super applicable in a lot of cases, but it's interesting
• 52:51
  Logan: to think about, um, alpha fold, like if, if folks haven't, um.
• 52:57
  Logan: Watched the movie, the thinking game, it's about sort of the early days of,
• 53:00
  Logan: of, uh, Google deep mind.
• 53:02
  Logan: And, um, they're talking about like folding proteins and why this is such an interesting space.
• 53:09
  Logan: And I'm not a, not a scientist, um, but the, to like hit on the point really
• 53:15
  Logan: quickly of like why Alpha Fold is so interesting. The the
• 53:18
  Logan: historical like context is humans to fold
• 53:22
  Logan: a single protein would take many humans millions
• 53:25
  Logan: of dollars and it would take on the
• 53:28
  Logan: order of like five years in order to fold a single protein the original impetus
• 53:33
  Logan: and like why demas won the nobel piece the nobel prize for this in chemistry
• 53:36
  Logan: was because deep mind was able to figure out using uh reinforcement learning
• 53:42
  Logan: and other techniques They folded every protein in the known universe,
• 53:46
  Logan: millions of proteins, released them publicly,
• 53:49
  Logan: made them available to everyone.
• 53:51
  Logan: And it was like, you know, dramatically accelerated the advancement of like
• 53:58
  Logan: human medicine and a bunch of other domains and disciplines.
• 54:01
  Logan: And now actually with isomorphic labs, which is part of DeepMind,
• 54:04
  Logan: like actually pursuing some of the breakthroughs that they found and like actually
• 54:08
  Logan: doing drug discovery and things like that.
• 54:10
  Logan: So like overnight, you see that hundreds of thousands of human years and hundreds
• 54:16
  Logan: of millions of dollars of like research and development costs saved through a single innovation.
• 54:22
  Logan: And I think we're going to continue to see that like acceleration of new stuff happening.
• 54:26
  Logan: A recent example of this, Alpha Evolve,
• 54:30
  Logan: which was our sort of like geospatial model that came out and being able to
• 54:37
  Logan: like fuse together all of this, the Google Earth engine with AI and this understanding of the world.
• 54:44
  Logan: Like it's just so much cool science and so much is possible when you sort of
• 54:49
  Logan: layer on the AI capability in all these disciplines. So I think to answer the
• 54:54
  Logan: question, I think we're going to see this acceleration of science progress.
• 54:58
  Logan: I think DeepMind is going to continue to be at the forefront of this,
• 55:00
  Logan: which is really exciting.
• 55:02
  Logan: And the cool thing for even for people who aren't in science is all of that
• 55:07
  Logan: innovation and like the research breakthroughs that happen, it feeds back to
• 55:11
  Logan: the mainline Gemini model.
• 55:13
  Logan: Like we had a bunch of research work about doing proofs for math.
• 55:17
  Logan: And it's like, oh, that's not very interesting at the face value. But like that research.
• 55:22
  Logan: Fuels back into the mainline gemini model it makes
• 55:25
  Logan: it better at reasoning it makes it better able to like understand these like
• 55:29
  Logan: really long and difficult problems um which then benefits like every like agent
• 55:33
  Logan: use case that exists uh because the models are better at reasoning through all
• 55:37
  Logan: these like difficult problem domains so there is this like really cool research to reality,
• 55:43
  Logan: science to like practical impact flywheel that happens at deep mind as.
• 55:47
  Ejaaz: A former biologist this warms my heart. This is amazing to see this get applied at such scale.
• 55:53
  Ejaaz: Okay, we can't talk about Google AI without talking about search.
• 55:58
  Ejaaz: This is your bread and butter, right?
• 56:00
  Ejaaz: However, I've personally noticed a trend shift in my habits.
• 56:05
  Ejaaz: I've used a computer for decades now, and I've always used Google search to
• 56:10
  Ejaaz: find things, Google Chrome, whatever it might be.
• 56:12
  Ejaaz: But I've now started to cheat on this feature.
• 56:16
  Ejaaz: I have started using LLMs directly to do all my searching for me,
• 56:20
  Ejaaz: to get all my sources for me.
• 56:22
  Ejaaz: And you've got to be thinking about this slogan, right? Is this eating the search business?
• 56:27
  Ejaaz: Is this aiding the search business? Or are we creating a whole different form
• 56:31
  Ejaaz: factor here? What are your thoughts?
• 56:33
  Logan: There's an interesting form factor discussion. I think on one hand, the AI...
• 56:38
  Logan: Sort of answer market is definitely distinctly different, it feels like,
• 56:42
  Logan: than the search market to a certain degree.
• 56:44
  Logan: Like, I think we've seen lots of AI products reach hundreds of millions of users
• 56:48
  Logan: and, you know, search continues to be a great business and there's billions of people using it and.
• 56:52
  Ejaaz: All that stuff.
• 56:53
  Logan: There's also this interesting question, which is like, what's the obligation
• 56:56
  Logan: of Google in this moment of this platform shift and all this innovation that's happening?
• 57:02
  Logan: And I, You know, as somebody who doesn't work on search, but,
• 57:06
  Logan: you know, is a fan of all the work that's happening inside of Google and has
• 57:11
  Logan: empathy for folks building these products, it is really interesting.
• 57:14
  Logan: And like my perspective has always been that search actually has this,
• 57:18
  Logan: you know, as the front door to the internet has this stewardship position that
• 57:22
  Logan: makes it so that they actually can't disrupt themselves for the right reasons
• 57:27
  Logan: at the same pace that that sort of,
• 57:29
  Logan: you know, small players in the market are able to do.
• 57:33
  Logan: And my assertion has always been that like, actually, this is the best thing
• 57:37
  Logan: for the world, the best thing for the world and for the Internet and for this
• 57:40
  Logan: entire economy that that Google has enabled through the Internet and bringing
• 57:45
  Logan: people to websites and all this stuff doesn't benefit by like,
• 57:48
  Logan: you know, day one of the LLM revolution happening.
• 57:52
  Logan: All of a sudden it's like a fully llm
• 57:55
  Logan: powered search product and like feels and looks completely different
• 57:58
  Logan: not only i think whether that throw you know users who
• 58:02
  Logan: are still trying to figure out like how do i use this technology what is the
• 58:05
  Logan: way that i should be engaging with it um what are the things that it works well
• 58:09
  Logan: for and it doesn't work for not only to throw those people into a bad um perspective
• 58:13
  Logan: from like a user uh from a user journey but i think it also has impacts on like
• 58:18
  Logan: people who rely on Google from a business perspective.
• 58:21
  Logan: So I think you've seen this sort of like gradual transition and like lots of
• 58:26
  Logan: shots on goal and lots of experiments happening on the search side.
• 58:30
  Logan: And I think we're now getting to the place where like they have confidence that
• 58:33
  Logan: they could do this in a way that is going to be super positive for the ecosystem
• 58:37
  Logan: and is going to create lots of value for people who are going and using these products.
• 58:43
  Logan: Like the understanding of AI technology has increased the adoption
• 58:47
  Logan: and the models have gotten better and hallucinations have gone down and all
• 58:51
  Logan: this stuff and I think there'll be also some like uniquely search things that
• 58:57
  Logan: like only search can do and I've spent a bunch of time with folks on the search
• 59:02
  Logan: team like Robbie Stein as an example who leads all the AI stuff in search and,
• 59:06
  Logan: There's all of this infrastructure that search has built, which as you think
• 59:11
  Logan: about this age of AI, where the ability to generate content,
• 59:16
  Logan: which actually like looks somewhat plausible.
• 59:19
  Logan: Has basically gone to zero. Like it's very easy to do that.
• 59:23
  Logan: Great search is actually more, it's like this premium is like more important than ever.
• 59:27
  Logan: There's going to be a million X or a thousand X or whatever than X number of
• 59:32
  Logan: like growth and content on the internet.
• 59:34
  Logan: How do you actually get people to
• 59:37
  Logan: the most relevant content from people who have authority who
• 59:40
  Logan: have you know expertise and all this stuff it's a
• 59:43
  Logan: really difficult problem and it's like it is the problem of
• 59:45
  Logan: the decade that like search has been solving for the
• 59:48
  Logan: last 20 years and is now a more important problem than ever um so i'm i've never
• 59:53
  Logan: been more excited for for the search team and like i think they've never had
• 59:57
  Logan: a bigger challenge ahead of them as they try to like figure out how to make
• 1:00:00
  Logan: you know these internet scale systems that they build continue to scale up to
• 1:00:04
  Logan: solve this next generation of problems,
• 1:00:06
  Logan: while also becoming this frontier AI product experience where billions of people
• 1:00:11
  Logan: are experiencing AI for the first time in a different way than they've done.
• 1:00:15
  Logan: There's so many interesting use cases too,
• 1:00:19
  Logan: even around like image search is a great example of this like new sort of,
• 1:00:26
  Logan: it's like one of the fastest growing ways in which people are using search now
• 1:00:29
  Logan: is showing up with an image and asking questions about it.
• 1:00:32
  Logan: And just like the way people had just traditionally used search is already changed.
• 1:00:38
  Logan: It's like different than it was five years ago or even two years ago.
• 1:00:42
  Logan: And I think we're gonna continue to see that happen. I think search as the product
• 1:00:46
  Logan: you see today will evolve to have things like multi-line text input fields as
• 1:00:52
  Logan: sort of user questions change and all that stuff.
• 1:00:55
  Logan: So there's so much cool stuff on the horizon for search that I'm really excited.
• 1:01:00
  Josh: Yeah, as I'm hearing you describe all of these cool new things,
• 1:01:03
  Josh: particularly funneling into a single model, like the science breakthroughs are unbelievable.
• 1:01:08
  Josh: And I think that's what gets me personally really excited, like Ejaz,
• 1:01:10
  Josh: is this is actually going to help people.
• 1:01:13
  Josh: Like this is going to make a difference in people's lives. Right now,
• 1:01:15
  Josh: it's a productive thing. It's a fun thing. It's a creative thing.
• 1:01:17
  Josh: There's a lot of tools. But then there's also the science part and a lot of
• 1:01:20
  Josh: this all funneling down to one amazing model.
• 1:01:22
  Josh: I think it leaves us in a really exciting place to wrap up this conversation.
• 1:01:25
  Josh: So Logan, thank you so much for coming and sharing all of this,
• 1:01:28
  Josh: sharing the news about the new model, sharing all of the updates and progress
• 1:01:31
  Josh: that you're making everywhere else. I really enjoyed the conversation.
• 1:01:34
  Josh: For you, you also have a podcast called Around the Problems.
• 1:01:38
  Josh: Is there anything you want to leave listeners with to go check it out or to
• 1:01:41
  Josh: check out the new AI studio or the new AI model?
• 1:01:44
  Josh: Let us know what you have interesting going on in your life.
• 1:01:46
  Logan: I love seeing feedback about AI studio. So if you have things that don't work
• 1:01:50
  Logan: that you wish worked, even for both of you, please send them to me.
• 1:01:53
  Logan: Would love to make things better. for the new model as well.
• 1:01:56
  Logan: Like if there's, I think this is like still, this is still early days of what
• 1:02:01
  Logan: this model is going to be capable of.
• 1:02:02
  Logan: So if folks have feedback on like edge cases or use cases that don't work well,
• 1:02:07
  Logan: please reach out to our team.
• 1:02:09
  Logan: Send us examples on Axe or Twitter or the like.
• 1:02:12
  Logan: Would love to help make some of those use cases come to life.
• 1:02:16
  Logan: And I appreciate both of you for all the thoughtful questions and for the conversation.
• 1:02:20
  Logan: This was a ton of fun. We got to do it again sometime.
• 1:02:23
  Josh: Awesome. Yeah, we'd love to. Anytime. time please come and join us we really
• 1:02:26
  Josh: enjoyed the conversation so thank you so much for watching for the people who
• 1:02:29
  Josh: enjoyed please don't forget to uh to like share it with your friends and do
• 1:02:31
  Josh: all the good things and we'll be back again for another episode soon thank you so much
• 1:02:39
  Josh: i have a fun little bonus for those of you still listening all the way to the
• 1:02:42
  Josh: end the real fans when we were first going to record with logan we actually
• 1:02:46
  Josh: had no idea that he would break the exclusive news of nano banana on our show
• 1:02:49
  Josh: it was super cool so we wanted to kind of restructure the episode to prioritize that at the front,
• 1:02:53
  Josh: we did record a separate intro where I said, hey, Google makes some really good stuff.
• 1:02:58
  Josh: In fact, you guys have an 80 something percent chance of being the best model
• 1:03:01
  Josh: in the world by the end of this month.
• 1:03:03
  Josh: Can you explain to us why, why Google is so amazing at what they do?
• 1:03:06
  Josh: And this was the answer to the question. So here's a nice little nugget for
• 1:03:09
  Josh: the end to take you out of the episode.
• 1:03:11
  Josh: Thanks for listening. I really hope you enjoyed and we'll see you guys in the next.
• 1:03:13
  Logan: My general worldview of like why Google is in such a good place for AI right now.
• 1:03:18
  Logan: There's many layers of this, depending on sort of what vantage point you want to look at.
• 1:03:22
  Logan: I think on one hand, it's like, I think search is this like incredible part
• 1:03:28
  Logan: of this story, which I think people have historically looked at Google search
• 1:03:31
  Logan: as this legacy Google product.
• 1:03:33
  Logan: And I think search is going through this transition and is actually like today
• 1:03:38
  Logan: actually just announced as we're recording this earlier, that AI mode is rolling
• 1:03:41
  Logan: out to 180 plus countries.
• 1:03:44
  Logan: English supported right now, and hopefully other languages in the future and is a great example of.
• 1:03:50
  Logan: Ai overviews and ai overviews sort of double
• 1:03:53
  Logan: clicking into ai mode being this
• 1:03:56
  Logan: product that actually like for many
• 1:03:59
  Logan: people around for for billions of people around the
• 1:04:02
  Logan: world is the first ai product experience that they
• 1:04:05
  Logan: actually touch um and i think there's like something
• 1:04:08
  Logan: really interesting where like google has been on this
• 1:04:10
  Logan: mission of like deploying ai and like you know there's some
• 1:04:14
  Logan: you know the the some naysayers on
• 1:04:17
  Logan: twitter will be like you know google created the transformer and then
• 1:04:20
  Logan: did nothing with it and it's actually uh very far from
• 1:04:22
  Logan: the truth which is search has been this like transformer which
• 1:04:25
  Logan: is the architecture that powers language models and uh
• 1:04:29
  Logan: and gemini um has been powering that experience with this
• 1:04:32
  Logan: technology for the last like seven years the product experience maybe
• 1:04:35
  Logan: looks slightly different today than it did then um but google's been an ai first
• 1:04:40
  Logan: company uh for as long as i can remember um basically as long as ai has existed
• 1:04:44
  Logan: that's been the case and now we're seeing more and more of these product surfaces
• 1:04:48
  Logan: like become these frontier AI products
• 1:04:50
  Logan: as sort of Google builds the infrastructure to make that the case.
• 1:04:55
  Logan: I think people also forget like it's not easy logistically to deploy AI to billions
• 1:05:00
  Logan: of people around the world. And now as you look at like, I think Google has
• 1:05:03
  Logan: like five or six billion plus user products.
• 1:05:08
  Logan: So the challenges of like, even just making a small AI product work today,
• 1:05:12
  Logan: if anyone's played around with stuff or tried vibe coding something like it's not easy,
• 1:05:15
  Logan: doing that at the billion user scale is also very difficult um
• 1:05:19
  Logan: so i i continue to be more and more bullish and
• 1:05:22
  Logan: like part of the thing that allows us to do that billion user scale
• 1:05:25
  Logan: deployment is the whole infrastructure story like if you're watching on video
• 1:05:28
  Logan: i don't know if you can see but i have a couple of tpus sitting behind me um
• 1:05:32
  Logan: yeah and like that tpu advantage which is our sort of equivalent to gpus um
• 1:05:37
  Logan: is something that i think is going to continue to play out so there's there's
• 1:05:40
  Logan: so many things that I get excited about,
• 1:05:43
  Logan: and the future is looking very bright.