Pixelated - A Gemini Vibe Check

Episode Date: April 10, 2026

Welcome to episode 95 of Pixelated, a podcast by 9to5Google. This week, Damien, Abner, and Will take a deep dive into the world of Gemini, breaking down Google's frequent UI changes, the different ways all three hosts use the service, and how it needs to evolve moving forward. Damien also walks Abner and Will through his experience using Gemini with Android Auto, and how it's more of a mixed bag than you might think. Subscribe YouTube Podcasts Pocket Casts Spotify Apple Podcasts Overcast Sponsored by Proton Unlimited: Pixelated listeners can save 30% on an annual subscription to the company's suite of privacy-friendly services by signing up using our link. Thanks to Proton Unlimited for sponsoring the podcast. Timecodes 00:00 - Intro and Gemini's UI redesigns 07:40 - How we use Gemini (and how we don't) 26:19 - Gemini for Android Auto 41:47 - Final Gemini thoughts and wrap-up Hosts Abner Li Damien Wilde Will Sattelberg Read more Gemini overlay and Gemini Live start rolling out big Android redesign [Gallery] Gemini for Android Auto is rolling out more widely, but not everyone is happy about it Listen to more 9to5 Podcasts The Sideload 9to5Mac Happy Hour Electrek Space Explored Feedback? Drop us a line at [e-mail], leave a comment on the post, or reach out to our producer. And for even more Android discussion, dive into the official 9to5Google forums!

Welcome to Pixelated episode 95. I'm your host, Will Saddleberg. This week, Damien Abner and I are going big picture to talk all about our current relationship with Gemini. From how we use Google's AI assistant and where we found it the most helpful, to our fair share of complaints and shortcomings. It's a great snapshot of where Gemini lives today as we continue towards next month's I. Plus, Damien tells us all about his not-so-awson experience using Gemini with Android Auto. It's all coming up right after this. Grab your phone and take a look at your app drawer right now.

You've probably found yourself relying on these same old apps and services since the nexus days, and what suited you best back then might not be the right choice in 2026. Luckily, Proton can help bring you into a privacy-first ecosystem built for our modern age. Proton offers practically every service you could possibly need under the sun across every platform, including email, calendar, file storage, password management, document editors, and, of course, a trusted VPN. Built with class-leading end-to-end encryption, Proton's entire suite is fast, intuitive, and secure, and export tools mean you don't need to worry about leaving data behind on those other apps. Pixelated listeners can save 30% off an annual subscription to Proton Unlimited,

which includes 500 gigabytes of cloud storage, custom email domains, a dedicated customer support team, and so, so much more. It's what's best for your current phone, not whatever you were using back in 2013. Upgrade to Proton today and save 30% on your annual subscription by heading to Proton. Dot me slash 9 to 5 Google or by clicking the link in the show notes. That's Proton.com.me slash 9 to5 Google. Thanks to Proton for sponsoring Pixel needed. So another week, another Gemini tweak, this time with regards to the UI,

looks completely different, completely different to what we've seen over the last couple of years. I'm not going to lie, Abner. I think this reminds me a little bit, there I say it, a little bit of the original Google Assistant UI. No, this is... Almost like card style?

The UI is rapidly changing and it's still rapidly iterating in a way that I think is... Okay, so let's talk about the change. So this is right now the change is with the Gemini overall there is instead of two distinct menus, two distinct sheets, one for the attachments, the plus,

and one for the tools, Google is moving to combine it into one menu. So that you just have one plus button. So I guess from Google's point of view, it's simplifying things so that you just have one thing to interact with. I think that's fine. It looks better on mobile than it does on the web where there is a weird drop-down situation. Two drop-downs in one pop-up menu,

but the mobile one looks fine. But at the same time, again, the simplicity is to why they're doing it. So users just go to one place. And I think at the end of the day, the model like in the far in the near future

the hope is probably that the model takes care of and recognizes what tools you want to use etc etc but I kind of do think

that some complexity is should be possible like two menus is not that hard the concept to grasp for users I think

I mean I like the fact that they've seen certain aspects of it simplified like when I was talking about the assistant, I think the pop-up stuff reminds me very much of the old assistant where you'd have like the weather and things like that.

But it is a strange one because it almost feels like Gemini's, like you said, is evolving at a rate at which we can't keep up with it. So there's no like almost, how would I describe it? It's almost like there's nothing to kind of pin your visual kind of memory to.

Does that make sense? Does that make sense? Like visual memory? So you kind of used to it being a certain way and then it changes three weeks later and you're like, what there? No. Yeah.

It's inherently like inconsistent, right? Yeah, yeah, yeah. If you are rapidly iterating on a, on a UI experience or U.S., right? Like, it's going to feel, even for users that are as technically adept as I think all three of us are, it can be frustrating when like the thing that launched two weeks ago or whatever, you know, but, you know, recently is suddenly different again for like the fourth time in, you know, however many months. Yeah. It's, it's okay.

Like, even as something as simple. the shape of the prompt box. It's switching between a pill and this rounded rectangle. It's going back and forth, honestly. They haven't really settled in the UI. And I think it's kind of, they should be settling right now. But I feel like what's actually happening behind the scenes is that they're doing this immense amount of user testing and trying to get something simple. but I don't know it is changing too rapidly I feel

I think it I think it does highlight how well the was it the next gen assistant NGA I'm just remembering the NGIA next genus assistant or the next generation the Pizzle 4 assistant like how simple that was and how cool that looked

I think it's almost like there is Gemini can do so much more and it's more of a I guess it's more of a lifestyle tool isn't it rather than an assistant that does things on your advice. So maybe there is just a way of like, we don't know how to deal with this UI. We don't know what the perfect setup is. And to give Google some grace for the lack of a better term, they are

building something that can do everything. And that's all these companies are. You can do everything. How do you put that into a UI? How do you convey it to users that this can literally be used for anything? That the limit is your imagination. It is a general. generally hard UI and a UX problem, but at the same time, I kind of, like, the home page should, at a certain point, it should stay similar or the same, like Google's homepage. And I don't know, it's just moving at the pace that's so quick that I do have concerns of people's sticky, that people want familiar things. That's just a basic UI thing. I mean, I guess, I guess, the benefit is that this is effectively on potentially billions of phones around the world,

whereas chat GPT don't have that benefit of people go to the website and they see it on the website, don't they? Or they go out their way to download the application. Google's kind of doing this at scale on, as I say, billions of devices at once. And so I guess there's no real right or wrong way to do it. I mean, credit where it's due, I do think that the way that they've introduced us to the Google assistant means that there's some level of consistency there. It is this section that kind of sits above your navigation bar and then it gives you the tools which weren't available

with the assistant on top. So it kind of makes sense. They've kind of iterated a little bit. Question for all of us. How on Android, do you open the Gemini app or do you use the overlay by wrong pressing the home button or swiping the app guy? I'm an app guy. I constantly forget about the over there. No, I don't forget. I just like, it's like an app, right? And so when I, when I want to go use an app, I tend to just dive into the app experience itself. And I, I end up not thinking about Gemini any differently, which is weird because like, you know, I do use like on my pixel. I use the Google search bar. I don't, I don't hit the Google app. But it's, yeah, I don't know why that's different for me. But I, I don't use this overlay very often.

Yeah. And the overlay is the new Google assistant. That is the Google Assistant. feminine, like, that exists because of the Google Assistant, and it's kind of interesting. I do, I think my gut is, like, the most Google Assistant people who are, they use the overlay. Yeah. Yeah. I mean, I mean, the overlay launches when you, when you initiate voice controls, doesn't it? And I'm a voice control person, because, like, most of the time I'm holding a camera, I'm messing around, and voice controls are always really nice to use for that. I do think, I mean, I'm going off piece a little bit, but I do think Gemini's better at handling the really weird, like, kind of half requests, I might send.

And I think that, if there's anything that Google has definitely done better, and obviously Gemini is the key component of that, it usurps the Google Assistant for the kind of things I would do with it. And so that's good. But I still find myself, even though I do use that, going to the Gemini application and initiating a full-screen conversation and just focus is not directly. I don't know if, I would love to sit and watch. somebody use it who maybe isn't like as you alluded to before will is technically inclined and just see what they do um i would love i would also love to know google's testing testing methodology for this like yeah yeah it's a weird one isn't it no i i actually i was i was thinking

about this because i want to ask you both like yeah we're pointing out kind of obvious shortcomings to google constantly um refreshing or or iterating on its current jemini ux specifically on this screen what do you see, like, what is the argument for Google doing this, I guess? Because I'm actually kind of having, I think Google should probably slow down its public releases of redesigns for Gemini. And like, you know, internally test as much as you want. If you want to do a beta program, you know, a public beta program, that's fine. But I think at the end of the day, like normal users do need some kind of consistency that isn't quite there with Gemini right now. I think they're trying to find, I think they're just responding to user feedback.

They're just looking at the user feedback, looking at the analytics, and trying to solve pain points. But who's submitting that feedback, though? Is it regular people or is it power users? Like, I guess that's my question is, like, is Google kind of doing this to appease the 1% of Gemini users who are using it, you know, 30, 40 times a day? I don't know what is an excessive amount of gym. I imagine it's way more than that to feel. Yeah, I was gonna, I was saying that. I was like, I bet, I bet you there are, yeah.

But like, like, are they, are they kind of responding to those people? Because there is an argument that like, that those aren't necessarily the people you, you, you need to make it as mass appeal, you know, as much mass appeal as possible. But, but, but like, I might not be. It might be regular people submitting this feedback as well. Does that dumb it down though? Do you think that dumbs it down? Because I feel, I feel like, I feel like, I feel like, I feel like, I feel like, I feel like, It's okay.

Yeah, this is a thing. How do they find that medium? How did they find that halfway house? You know, we've talked with you guys, and we've talked a lot about the idea that, like, I think a lot of AI applications are stronger when they do have some set of guardrails. I do think there is an argument that, like, you should, you know, for your average user, you should kind of keep the ability to sort of, not necessarily railroad them, but certainly give them like very specific examples and options.

Well, yeah. And maybe don't open up the entire, you know, don't say like, hey, it's a playground. Here you go. This takes us to a much broader discussion. Like I said earlier, these Gemini, these chatbots, they can do anything. And that is inherently ironically constraining. So there's a Gemini app, which is Google's goal to build an AI assistant for your life. And then Google has been integrating Gemini

into every single one of its products. What? Where have you been using Gemini in like, have you been using Gemini and Google Maps? Have you been using where else do you in Google Photos? Because honestly, I'm still, I'm doing it all in the Gemini app. I rarely use the Google Photos integration. The Google Maps one,

I can see myself using it, but at the same time, I'm just using the Gemini app again. So where are you guys using outside the Gemini app? Okay, well, let me, I won't talk about Gemini for Android Auto yet. I'll save that because I'm,

I'm just going to spoil it now. I'm going to be very scathing on it because I do use Gemini in the car. A lot of the times, I'm so lazy, Gemini in Gmail is fantastic at finding specific things. I'll give you an example. So three days ago, I remembered that I'd bought a Steam key to play.

I can't remember what game. It was something. It doesn't matter what game it was. But I tried to find a Steam key in my library. And I was like, oh, I haven't redeemed it. It must be in my email. So I went to Gmail. I searched.

I couldn't really find the retail. I could not remember the retailer. It would have been about one of about 15. So then all I did was go to Gemini. I purchased a Steam key recently. I know what the game is. I mean, I can't remember it right now, but I said, oh, it's for this game. Boom, it found it.

Yes, that's not necessarily that useful. But then I can ask further questions like, oh, I, for some reason I thought, oh, how much have I spent on this website recently? Because I've been buying a lot of games on this website. Yeah. It can total them. It can pinpoint the pinpoint them. No, yes, absolutely.

And for that, it's fantastic. It is. But I don't know. I guess it speaks to how I'm definitely in that pro user category of the Gemini app. but I just do at Gemina, at Gmail in the Gemini app. But I do agree for the vast majority of people. It makes more sensible in the app like Will was rooting to. These integrations make sense.

How about you, Will? Where else are using it? Are using Gemini and Chrome? Are you using... Yeah, I mean, honestly, like... This is like a boring answer because you're looking for me outside of the app. But, like, the truth is that, like, the place I use, Gemini the most is like is within Chrome but typically it is just the you know Gemini.govogel

which is not really answer your question but like 90 plus percent of my of my AI usage comes out of very random prompts of like you know looking for like looking through my chat logs it's like there are there's like a couple very generic like photo edit things where I didn't have time to do it myself for like a, you know, a meme in a group chat or whatever. There's me looking for specific moments in a YouTube video. There's me looking up straight up, like some Pokemon information that I was having trouble finding. Oh, you could have asked me.

I had told you. Well. Having trouble finding it on Google, like very niche, uh, Pokemon info. There's like a weather prompts, uh, you know, like, there's, there's lots of like, uh, random, Like, it's really just kind of a like, if I know I'm not going to get good search results or it's a thing that Google search just can't do well anymore or maybe never could, I will turn to it. That's, that's 90% of my of my Gemini usage, which is maybe a boring answer. And Damien, your example in the pre-show or in Slack, we were talking about this. And I was like, maybe I'll get some inspiration for you guys.

I mean, the way you were describing it, I hadn't, you know, I knew you could pull up information pretty easily with Gemini. I guess I hadn't been thinking about that you could then be like, by the way, total all this together. Because like, you know, going through my receipts, it's tax, it's literally tax season. It's that it's a, as we recording this, we were a week away from tax day in the U.S. And, you know, it means like I manually went through and tracked all of my expenses right throughout various websites and then going through Gmail. And like, I hadn't considered that I probably could have used Jemai to at least dig up a lot of that information. And part of that is, you know, I would, I would probably want to double check it anyway because you don't want to turn hallucination-filled information into the IRS. That's not one

place you want it. Yeah, but it does. This is a good thing about, I feel like, I feel like, this is, that AI is moving so fast into, well, I mean, it slowed a lot recently, but in terms of like from where we were three or four years ago to where we are now is completely different stratosphere. With AI, I think a lot of what Google, and the others have, and I've said this on a podcast before, they've not communicated the real value in what it can do. Instead, it's more fearmongering from certain sides of things that it will take this job, it will do this, it will do that,

which is a genuine threat. But I think in terms of like day to day, a lot of people could potentially utilize certain tools. Like I've been using anti-gravity recently to, I mean, I've programmed before, but I'm a very lazy program. I just don't like it. Usually if I have a vision for something, I'm making small applications to save time in certain areas.

so auto watermarking tools, maybe certain things where it compresses images for me, rather than using a tool that I would potentially have to pay a fee for. Like you can do these things with specific tools, but you're not going to necessarily do them in the Gemini app. So I do think for me, the real value of Gemini is its integrations elsewhere, using it with X code and anti-gravity to build certain things at saving time on my PC or whatever it is,

or like I can build potentially a component for After Effects that automates animation tools, rather me manually doing it. Like, no, that does not look sexy on stage. It doesn't look sexy on an advert potentially unless it's something really, really cool. Sadly, I've given you bad examples, but those are the kind of things where it's integrated

into the products that people are never going to find out about unless they have a prior knowledge of it. I do think that's why Google is probably going to, emphasize the integrations for things like Gmail, like photos, like you mentioned, Nabner. I don't use it in those products probably as much as I should, because I do, like I say, I can be quite scattergun with certain things that I have. Like I say, coming back to it at a later date and searching for things is really useful. But then you could potentially build an application that does it for you if you really wanted to, if you were really technically inclined.

I think probably that's where there's a lot of innovation going on with regards to AI the last six months, and I changed my opinion quite drastically on it, that there are genuine use cases that don't, like, potentially damage the products I'm already using. They can enhance it, enhance it, sorry. Yeah, I mean, it probably gets at a wider point. The way Gemini works is, it has to be a one-size-fits-all solution for everything, because that's what AI is, isn't it? It's applied. So maybe, maybe getting out of the website, Gemini, at Google.com or the application and trying other integrations might be a way and having you to get more from it.

Okay. To what you said, Damien, about how, like, yeah, it's not sex. The use cases that are useful to you to each person are not sexy and not broadly applicable. Like I've said before, AI is the hardest thing to market. It is the hardest thing. You can give a thousand examples and maybe one sticks to a person. And in advertisements, like we've talked about, like, the pixel 9, pixel 10 ads about Gemini a few years ago. And they were like taking a picture of what's in your fridge.

That's the lowest common denominator they ended up at. And that's not, that didn't resonate with me. But there are, like, again, there are genuine uses. I do think for most people, and that this reflects why Google is adding Gemini so hard to so many features, I think that's to so many apps. I think that's why they're pushing so hard. At the same time, like, Will, you mentioned using Gemini to find something in the YouTube video. That was with a Gemini app and not with the, like, there is that ask feature in YouTube that basically has that same level of functionality.

But again, that's something that I don't use. I find myself taking the link and putting it. it into Gemini app. So it is, I do think for most, for the wider audience, how they'll use Gemini is in these apps is when people integrate them in, like in photos, in Google Drive for the tax examples. But I don't know, like for the power user thing, I think as a power user, as somebody in this field, I know that all the advancements are going in. into the Gemini app

as what's the latest models are being put there versus in these other application specifically in these applications and I do think that most people

wouldn't care about what model is being used at the end of the day but I it's that gap is interesting and I don't think it's something that gets resolved I have so many thoughts

and I feel like we're to lose track of them I wonder if we're reaching this point where the built-in app experiences, the Gemini experience is built into YouTube or Gmail, all of these places that we're talking about, should start to, I'm not saying Google should hide that they're powered by Gemini or whatever, but maybe we are kind of, and maybe this is something that gets brought up at I.O. But we're reaching the point where these can just be features in the app instead of being like, we are layering AI on top of this, like build it as a traditional feature into the app

where you go like, oh, yes. It's getting there, but I still think there is that, like, separation of, like, this is the Gmail experience, I remember, but also there's this layer of Gemini over it that I can choose to interact with if I want, rather than seeing a cool feature that, you know, 10 minutes ever you use it, you're like, oh, and also that was powered by, you know, Google's Gemini LLM.

Actually, Gemini is a very... good example of that. Sorry, Gmail is a very good example of that. Originally, the idea for integration was a Gemini side panel where there is the Gemini Sparkle on the right side of the screen and you can open a panel. With that, there's still the case for most workspace apps like Drive and docs and whatever. But the newest wave as announced in January is they're basically putting AI overviews in the search box and that's... It's one step closer, basically, I would say, to to what I'm, I'm really just envisioning. It's like, it's like, oh, we've launched, you know, you get that press release where

I'm not trying to say Google should downplay Gemini, but like it, you know, you get that press release where it's a really cool feature being added to Gemini and or added to Gmail. I did it too in July or whatever, right, like a few months from now. And then like it's not until like paragraph five where they start talking about that it's powered by Gemini. Like I, I think we're getting to that future where it's like Google has its models in place and can now just like integrate them directly. But as you said, Abner, you're going to be dealing with older technology a lot of the time

or like that will slow down implementing, you know, future changes to these models, which Google is obviously pushing as fast as it probably can, which is sort of the whole AI race, right? I wonder if that's the other reason that like I, you know, and broadly we, but certainly I default to the dedicated application experience, whether on mobile or on the web, is because it kind of pushes all of that data into one place and you can return to it and you can add on top of it in a way that, like, you maybe can't if you use like YouTube's. I actually, I don't think I've used YouTube's. I haven't used Gemini in YouTube. So I don't know how if you can return to those conversations ever, but like if you go back to the video, but like, and correct me that you guys

can tell me. But I don't think so either. It's nice to be able to go back and see that information and be like, oh, hey, I actually have like a follow-up question two days later, especially if you're working on, you know, a project of sorts, right? Instead of just asking for entertainment purposes. Yeah. Yeah. It's an ecosystem, isn't it? The Gemini ecosystem is growing. And I think that Google, it would be, I mean, Google is trying to emphasize it. And I understand why they're doing that because you have to, because everyone's in it, like you said, the AI race and everybody needs to know that Gemini's at the forefront of it. I think for most people, they don't necessarily care.

They just use the products that work well. And I think that the long game has got to be these products work well. If it's Gemini doing them behind the scenes, that's all that matters to Google, as opposed to the person buying the product or using the product. I mean, let me get into some teething problems I've had recently using Gemini and underdata because this is one of those things that I think when I'm, I mean, it wasn't marketed as such. I saw last year, I know, we managed to see a demo of this running on an automotive.

I think it's Google built in now in cars, terrible name, should have stayed at automotive. And then Android Auto, which I think is probably the de facto experience most people will have with Android in the car. I genuinely think that the way that this was shown off and what we actually have now is such a drastic step down, that it's actually really, I've been so disappointed after using it for the past. probably three weeks to three weeks. And it kind of doesn't mirror my experience using Gemina on my phone or on a desktop because I think I kind of expected it

to the stage we're at currently with the service that it would be completely seamless. Like the mobile experience on a bigger screen, in my car, I can do the kind of things that I'm used to. And while it promises those, it just doesn't, it's one of the few times where I think it hasn't delivered. It's a little bit like the early days

of when Bard, was previewed and we were kind of like, what the hell are you going to do with this? Why is this useful? It feels like that. And it doesn't help that the assistant is still hanging around in cars. Like, what have you been using? Give some examples. Okay. Let me give you a terrible example. So when you get into a scroll hole on Wikipedia, you know, you know the kind of thing. You do your research. You search, you hear something on the radio. You hear something in a podcast. You know, I need to learn about that right now. Well, when I'm in the car, I can't do that safely while behind the wheel. So the assistant's terrible at that, but Gemini is very good at pulling of

information from Wikipedia pages, I can delve into the subject, I'm one of then kind of nerds. So I see, I hear about a sports player, let's take, for example, Trent Alexander Arnold, any soccer fans, football fans out there, and know what that is. I wanted to know how many games he played for Real Madrid recently, asked the question, Gemini gives me an answer, I try and get into a back and forth with it, Gemini doesn't want to do that in the car. Usually when I'm at home, amazing, we've gone live, I'm having a flowing conversation, I'm asking follow-up question after follow-up question. For whatever reason, it just does not want to do it in the vehicle. And that instantly has put, like, it's increasing my road rage in the vehicle.

I think when I saw it in person last year, it was a sort of on Rails demo, but a not on Rails demo, we could kind of go off piece a little bit. And it felt like this is the future of cars. And it understandably is. I don't think Apple are going to be able to compete with this. The ability to have a flowing conversation with what effectively is a super part assistant and get these questions like maybe when you're driving, I've tried to do it while I'm driving as well.

And this doesn't work. I'm near a, we just had Tim Hortons has come to the UK in the last, I don't know, two years. there's one nearby to my parents. You missed the good era. Yeah, yeah. I'm going to tell you. It used to be good. I promise you that.

I know there's one on the route to my parents' house. Yeah. I have to deviate slightly to go to it. I asked, I had mapped already. It was already running because I like to know when I'm going to arrive. I can drive there without directions. I asked for a kind of a,

the quickest way to get Tim Hortons on the way to my parents. Gemini is asking me, trying to ask me follow up questions. It's just completely unrelated. doesn't work, stops the navigation completely. I'm like, what is going on? The assistant obviously wouldn't be able to handle that, but I know it's limitations.

I think, because I know Gemini can do more, there's a level of frustration that I'm getting here behind the wheel, which is just, it's just so irritating. And even little things like the UI, like I've had instances where I've tried to reply to a text message. And yeah, again, the assistant was not good at this. It was fine. You had to be very precise about what you said.

It would take it verbatim. Gemini is pretty good at understanding vocal ticks that you might have, like you say, um, or whatever it is. Or it will then put, uh, sorry, it will format it correctly to send a message. But it'll cut off halfway through. And Gemini is really good at it. Normally when I'm talking to Gemini Live, I can talk for five minutes and it doesn't need to respond and it knows.

Whereas in the car, it'll cut off halfway through. I have to do several messages back to back. and that kind of, again, for me, defeats the purpose. It's supposed to make it safer for driving. Yeah, there's some little things there that definitely need work. I know they'll get fixed, but it has left a really sour taste in my mouth, having seen such an impressive working demo over a year ago. And then this was delayed as well, so I don't know what the problem is.

Probably because the too busy working on phone UI rather than making it work in a car. And the rollout's been done. I could talk, I could literally, it's getting me aggravated thinking about it right now. So I will okay I don't even know where to start I had to go look I remembered I was like when I was at Android please at one point I wrote I remember writing an article about

how I was like Gemini needs to come to car soon because assistant is not only bad and when you're driving it can be like dangerous it can be it can be useless to the point of being dangerous where you you know the average driver not me

but the average driver might be that sounds like I'm being okay I really do not use my phone in cars when I'm driving. But the average driver might be tempted. I mean, God knows how many people use their phone while driving might be tempted to be like, you know what? I tried using it hands free. I'm just grabbing my phone.

Screw this, right? And like, I have spent, you know, first of all, I wrote that article almost two years ago. It was in June of 2024. I just found it. It took me a while to find because of that. Like, that's how long we've been waiting for this. Secondly, like, it's, if, if Gemini and Cars is opening. up, you know, even more danger. It's ending your navigation, which can be fine if you're, if you

kind of know where you are or you're on like, you know, city roads. But if you're, if you're going fast down a highway and you, and you're like coming up on your exit or you're in busy traffic, you know, I remember, I remember I was driving. The last time I drove to New York City, uh, when we were driving home, we were on one of the bridges leaving the city. I don't remember which one and maps crashed, Google Maps and Android Auto crashed as I was like approaching like three different exits and I didn't know which one to take. And like it's things like this that like reminds me of that where I'm like that was like one of like the most like concerning moments I've had because it was deep, it's heavy traffic. It's New York City. Um, you know,

while driving of like, oh, if I pick wrong here, I might be like, I might be adding 45 minutes onto an already eight hour drive. Um, it, it's things like that. And, the promise to me of Gemini and cars, the reason I was like, well, this is actually where I want Gemini is because it's got improved. Its ability to understand language is so much better than assistant that I'm like, that is where I want that tool more than anywhere is in cars, but then I need it to give, I don't want it to be just like standard Gemini and the way that it seems to be offering you, Damien. I want it to be kind of paired back, kind of limited, and maybe you can, you know, activate a bypass mode or something. But like, I don't want the car just like talking to me for like 90 seconds unless I really, unless I tell it to, right? Like I want like, I sort of want the paired down assistant experience with improved language processing, which was sort of what I thought Google was trying to build here. And in fact, I think I had even asked them if they were planning on, uh, limiting Gemini's, uh, you know, responses response length after they announced this to, to,

years ago and they said yes. It doesn't totally sound like they did that. But, you know, I don't understand why you would offer this like, especially after two years, this like buggy early experience that might not be ready for what is like statistically a very dangerous task that we all do every day, right? Like you can't screw around with this. I'm frustrated to hear your experience having not used it yet, Damien. Like, it makes me not want to use it. I mean, another thing that annoys me is, I don't even know if I've seen this yet. I've spoken to a couple of you about it

and no one can really confirm me the way, it's not the case, but you know the Gemini Live call thing that has changed on phone, which looks great, by the way, the little pill. I think it looks fantastic. The Gemini Live UI isn't appearing for me in the car, and I'm going to put that down to

potentially just a geolocation thing. It could just be UK, whatever, but it just shows that, You know the kind of like the, how do I describe it? Wave form? Yeah, it doesn't even show the waveform. It just shows a little line that we've had from the assistant. So if I'm not, if I'm, I mean, it says ask Gemini, which is fine.

And I can get a back and forth conversation going. But it doesn't show the little call UI that you get for Gemini Live. So if I'm driving along and suddenly Gemini just stopped talking, I'm like, oh, I'm going to have to initiate you again. And then I have to go and press the assistant. button, either on my steering wheel or on the screen, or use the wake word, which I'm not going to say, it's these little things. It's like, it's almost like, I don't know, maybe I'm being harsh, but I had such high expectations. I think that's a problem. I had such high expectations. And seeing it where it is now versus what I saw a year ago, I'm like, what have you, like,

what has, has regulators stepped in maybe? I don't know. It could, it could genuinely be that. It could be the safety in cars, things and have diminished it a little bit. But, when it works, it's still really good. Like, again, I can sometimes get a mini conversation going and all of a sudden it just stops. Yeah. This is why I'm not that big on voice as the way to use Gemini. I type everything out. It's, from the technical side, these voice, Gemini Live, the model that powers Gemini

life is not the state of the not it's a hard problem to do to do the voice real-time voice interaction latency etc etc that's not that's on the same level as where the text stuff is the text stuff is so that's one thing but it's I don't know it's still I this just feels like such an old issue to rehash but talking with four I still prefer text and that's So I probably don't encounter these cases that you are seeing with Android Auto. That being said, of course, it has to be in a car. It has to be voice. You're not typing with a keyboard or anything like that.

But yeah, it is frustrating. Yeah, it is. I know it will develop. I know it will improve over time, and that's the main thing. But I'm seeing people with the similar sort of frustrations in the car. But I guess everything's going to have to eat in problems. I don't. But yeah,

but okay, okay. I appreciate the devil, like, kind of being devil's advocate for Google, but it's been two years. This is,

and launching in cars specifically, it's just not a place where you don't want to be mostly ready. And, you know, even more than like,

is Google's current experience dangerous? Let's set that aside and say, eh, probably mostly no, because if it closes your navigation

or something, you can pull over. Let's, let's set that to the side. Google gets one chance to make a first impression with Gemini and And if this is its first impression, that's bad because it's going to, again, push people to be like,

you know what? I tried hands free. I'm going back to just using my phone. Like, dangerously, do not do that to be clear. But like, those are the, and I'm not saying anybody's right to do that, but that's the social circumstances that Google is going to build if it doesn't iron out some of these bugs pretty quickly. Yeah. Yeah, I don't doubt that they will. I think the feedback is going to be thing and it's rolling out more widely now so maybe it will improve drastically it could just be that i was one of those early rollouts and it isn't working necessarily on my car head unit so while i've had a little bit of a disappointing experience i know that it will improve i know that it will reach where it was previously um i'm going to go but i'm probably going to continue to put it

through its paces over the next few weeks anyway and then hopefully things will have changed because like i say it works it's probably like 60% at the time it works how it should um and it maybe doesn't help it knows time it works every time yes yes exactly I'm, God damn, I missed that. But yeah, I think it doesn't help for me. My experience seems to be better when I'm parked than it does when I'm actually driving. So there's a little bit diminution, diminution experience there.

So, yeah. So let's move. So I am curious about, you use Gemini Live on your phone, right? I don't use Gemini Live. I have not found the utility for Gemini Live in my life, Mostly because, Abner, I don't, I don't, I find that, you know, live is best for talking and I don't, I don't talk. Yeah. Use voice commands.

Like, so I just, I'm not, I'm not a Gemini Live user at all. And I'm not going to pretend to be. It's just, I haven't found space for it in my life. Damien? Yes, I use Gemini Live way too frequently. Um, I think I've talked about it in the pod before that I get caught using it by my partner. And she's like, who the hell he's talking to? Um, I like it, especially when I'm cooking is good.

Um, it's so funny. you and I are like polar opposites. When I'm when I'm cooking, I'm like usually listening to a podcast like truly like. Yeah. It's interesting. No. Yeah.

It's for me, for me it's, it is. It's for me it is the conversation element that's useful because like say, when my hands are busy like chopping things or cutting things, I can ask questions. But to preface that, I think because I was a very early adopter of the Google Home and the Google Home

and the Google Home Mini, my brain is hardwired to do that. And what I will say is that I've wanted, when the assistant has kind of degraded over time, it's kind of corresponded with my usage of Gemini increasing. So, because Gemini can handle the kind of queries that I want where I'll ask it to play a certain radio station or I'll ask it to do things because the assistant was so streamlined to, this is the voice request, can't deviate from it. Yeah.

For me, I've always wanted to be able to deviate from it. So, yeah, I think for me, Gemini is a useful product data. I couldn't understand for a lot of people. They have to find a way to integrate it, but Gemini Live will probably be my most used Google product of the Gemini era anyway. Interesting.

Yeah, for me, Gemini Live doesn't have enough phone control capabilities yet for me to really use it. It can, yeah, it's not, it's not, can't, I'm still waiting for phone control. That's at the end of the day. And not there yet with the Astra stuff. Yeah.

Hopefully this year. Yeah, definitely. I think that kind of sums us up on Gemini. I think we talked about Gemini for a long period. That's probably the longest in a pod we've done for a while. I think we'll probably see some really cool stuff at IO and then be blown away and be like, oh, this is amazing. And Will will be like, I don't know how to integrate this into my life. I'm going to continue being a little done. Wait a minute. But it's it's it really is like I'm still a

little bit waiting for the the it's not that I haven't found shortcuts with with certain tools or or things that where I'm like oh this like minor thing that used to take me 10 minutes now takes me two minutes or something right like I those exist for me but it I have not found the like you know there was pre the pre-Gemini era in the post-Gemini era for my mobile phone usage and I still kind of, it's not that I'm, I don't, I obviously have the app, but I still kind of feel like, yeah, this is just an elevated version of assistant to me for the most part, like, which I might be true for a lot of people, but it's, it's, I got to build, maybe it's just building up that, that muscle memory that you have Damien in particular. Yeah. Yeah. So a few

weeks ago, I wrote something about the, the mundane absurdity of having Gemini about your life, about updating it on things. It's so hard to talk about this because it's, it's, it's, it's, it's a personal assistant, putting the personal assistant in assistant. Like for the specific use cases, like just updating it's about something that you're tracking in your life and then like giving it, giving Gemini an update about what's the latest. And then it giving you updated guidance, let's say. That is the end goal of all of this. Like, in a way, thinking of it as an extension of Google Assistant is almost of the Google Assistant we got versus the actual vision 10 years ago.

It's a great disservice because I still think most people's mental model of all this stuff is based on giving it a structured prompt like you were saying, Damien, and then getting a response that may or may not be right. You can use this as a personal assistant, like to keep a track. Let's say fitness goals in your life. You give it a, you say you want to do this, you have a plan to do a marathon and you give it like your split times or something like that. And then it builds up, it gives you a plan. You can, that is something you genuinely could not have done at Google Assistant. But again, it goes back to the idea that people have been conditioned for, 10 years of these dumb assistants.

I think that's what I'm calling them now. And it is so hard to break. I think in the past year or two, Gemini, chat, GPT, Kla, they've definitely broken that. And there is the power user audience and above the developers. They have absolutely broken that.

And they know it's going to be done for so much more. And the issue is getting other people, the general, the general, everybody else to find out that this can be used as so much more than a dumb assistant. So again, it honestly falls on Google to start breaking things out to really explicitly tell people this is a system for your life. And they're starting to get there with past IOs, but we'll see whether this is this coming IEO in the future,

whether those thoughts being so explicit about it. Yeah, and we've got that to look forward to, haven't we are you? So, I mean, what are we, nearly a month away? They're five weeks away. I'm sure we'll learn more in time. I don't doubt that Gemini will be at the forefront again like it was last year. Maybe we'll hopefully get a bit more Android as well because we like talking about Android more so than we're talking about Gemini.

I'm going to be completely honest, but it is what it is. Let's convene there, guys. We will talk more about Gemini in the coming weeks, talk more Gemini, talk more I.O., sorry, and everything else in between. But I want to say thanks for joining me. It's really interesting. I think it's hopefully an illuminating chat for anyone out there who doesn't know how to use Gemini and is wondering why it keeps changing so drastically and how to integrate it more in their lives.

But maybe you've learned something as well, Will. But yeah, thanks for joining me. And I'll speak to you soon. Bye. Bye. Thanks for listening to Pixelated, a 9-to-5 Google podcast. If you enjoyed the show, we ask that you rate and review it on the podcast platform of your choice and help spread the word by sharing the show with friends or on social media.