Not sure I understand the negativity in this thread. I’m an iPhone user and Siri drives me up a wall. Being able to converse with something that will do stuff properly without interacting with a screen (eg text someone without doing a word and punctuation perfect dictation, or find a gas stop in X minutes on a road trip) seems like a huge value add.
Because the answer isn't more automation, it's more user control. You think swapping Gemini in for Siri will solve the your problems? It won't because your problems are related to the core that both of those share.
Easier to think of AI implementation in a use case like this as just a UI/UX leap forward.
"Gemini, plot a route to [city 3 hours away] and incorporate charging locations with cost and total wait times for each"
"Gemini, find me new music you think I'll like"
I once asked a product head at Bard "so when do I get my own version that has access to all the data google has on **me**." He blanked intently for a few seconds and abruptly changed the subject.
Spooky or not, it's coming and no one wants to talk about it. Imagine if you had your own private version of Gemini that, on top of all the publicly available internet, it can tap in to what Google already has stored on *you*. "Gemini, I'm thinking about a total career change."
"Sure Dave, here are 3 ideas for a career change based on the wealth of historical data I have on you. Would you like to explore certifications, educational requirements, and strategies about how to enter them?"
Except it literally does — I just tried it, right now:
"Set an alarm for 8:30 AM tomorrow"
"Okay, I set an alarm for 8:30 AM tomorrow."
Presto, done. If your side of the argument is so bad you need to tell an outright lie, then it's pretty weak, I gotta tell you.
Hes not lying, lower your puffed up feathers.
The hit rate on this shit is low, and Im glad talking to a voice to text is the best feature you like a car.
Would you like a old copy of Sharper Image magazine to fawn over some gadgets with your happy meal?
Good for you!
Tragically, mine behaves exactly as I've stated.
Instead of assuming you're lying, I'm going to make the reasonable assumption that it's as inconsistent as many reviews paint it to be.
I think it is device apecific. On your iphone or gemini app's gemini live it wont do that. For me on samsung s24 it does when launching the gemini that has replaced google assistant.
I also think it comes from people thinking they'll lose something.
it seems this is coming with set of featurs that add more phone to a cars infotainment. Watching videos and gaming while parked. It is targeted that when you are sitting in your car waiting it to be charged you wont get bored.
The default experience most non-Pixel users have with Google's AI is those appalling AI summaries. People expect the same thing from this, rightly or wrongly. Also this sub is generally (and imo rightly) suspicious of new technology in cars, due to it often being used to spy on users.
Android Auto already can send text and find locations with additional instructions (using Google maps - Waze is more basic). I guess AA could be better but it's already very noticeably better than Siri in Carplay.
the negativity is a reaction to the prevailing mindset that texting someone while driving or requiring anything other than eyes to find a gas station are things that need to happen while driving in the first place.
ever since this sub transitioned into a current-events debate sub several years ago (coinciding more or less with the member count crossing the 1-million threshold), the prevailing attitude here is that every new technology thrown at cars is to be accepted without question.
What’s with all the negativity? I use apple carplay, and siri is a huge help. You can text, call, and play music without losing your focus on the road. It’s a huge plus.
Yeah I'm not too sure about all of this. I hope there's a way to at least make it optional. Adding chatbots to all of our car infotainment might just be a little much.
"Chatbot" implies long-context conversational back-and-forth with a persona. They're using this for commands like *"send a message to my wife, tell her I'll be home soon"* and *"find me a gas station on my route with a burger place next door"*, which both get translated into tool-use chains.
The foundation is the same, but it's a massive difference in context.
Literally the opposite is true. Chatbots are implementations of LLMs. The former implies the latter, the latter does not imply the former. Language models are stateless by default, and generally must be tuned to serve as chatbots.
This may be an upgraded experience. It likely will replace the already active "Google Assistant," which Gemini is about a hundred times better than. It does everything that GA does, but better. I highly doubt the chat bot will even be available while you drive.
My current tablet can barely switch between waze and google maps while playing a podcast on pocketcast. There's already negative headroom for features.
but that doesn’t run android auto, your phone does, and it’s only going to be supported on phones that support gemini (which have more than enough processing power to multitask), the tablet just displays it
From the article:
>Today at Google I/O, a developer conference that takes place every year in the San Francisco Bay Area, the Mountain View, California–based search giant announced a significant upgrade to both Android Auto **and Android Automotive OS**.
Android Automotive being what I know I have, but hey, assume I don't and I said something else just so you can look smart.
Woah, did you READ the article or listen to the Android show where this was announced yesterday?
You aren't supposed to do that here! Just take your down votes and like it!
I get curious while I'm driving. When I see something, I immediately have questions. Right now, I have to go through the hassle of opening it through my phone, but this seems useful.
Why? it’s conversational and leagues above the old assistant on mobile, much less whatever apple is doing with siri. Great natural language handling & works well to plan an itinerary etc.
You're getting downvoted because reddit hates AI, but I also have been waiting for Gemini to hit android auto.
Sometimes I wanted to search things up while I'm in the car and the standard assistant is useless for that. This would be great.
I use it daily at work as programmer. It’s convenient for initial search/boilerplate code/random debugging, especially when I don’t know exactly what I want.
Basically it works well for ambiguous questions at the expense of sometimes producing BS in the answer and being overly verbose.
Asking a car to increase the A/C to 75 degrees is not an ambiguous question, and I don’t want any ambiguity while operating a 2 ton crowd missile.
This is for android auto, not android automotive, it can’t control climate and the like by design, just maps, music, calendar, etc. It’s the analogue to siri and the replace for google assistant
I've been using Google Assistant for 5+ years to set reminders, I'll tell my phone "remind me at 5pm to do XYZ." Assistant never had a problem, but I've only had it work with Gemini once. I'm honestly all for AI but I think it's a really poor implementation
that’s weird, i’ve had the opposite, assistant could rarely understand me and gemini has picked it up each time, good with switching languages too
Sucks that it’s hit or miss, still, i think it’s good they integrate it as long as there is an option to default back then
Switching between English and Polish has been the main difference for me. Assistant was borderline useless, Gemini tends to select the correct language most of the time.
I have had Gemini on my S25 now for months, and honestly, it is great.
I can tell it things throughout the day, and it just remembers, it can access my camera and help me with things in real time.
I dont see why you wouldnt want it in your car? Way better than most voice assistant systems in cars that barely ever work.
No idea either. I’m convinced the people hating it haven’t used it, because to me there are genuinely zero drawbacks.
Either that or just the standard “ai bad”
All the negativity! Personally, I can't wait to switch from Google Assistant to Gemini.
Whenever I say "Navigate to X" it seems to get it wrong.
Gemini can only be better, surely.
Imagine being at the Bay Area and specifically Silicon Valley, it’s all people talk about it.
“So how are you gonna integrate AI to your photography career?” Like, full stop, please I don’t even wanna think about AI anymore. I hate it so much. They all think it’s the savior of the world in here.
The thing that irks me is that this is a primarily US push, so while we make shit shitty over here and tech bros cheer it on like people are shitting on the first smartphones (or whatever bs reason they think up), most criticisms get ignored because of ego, it has yet to be close to profitable, and plenty of other countries are already starting to surpass our industries in quality.
So we’re essentially, absolutely killing our future in the global market, but at least a few guys got rich from making the US’s industries worse, and hey, you can view being behind on innovation as just more money to earn catching up. Fixing all of the “vibe coding” as new job market opportunities done for cheap, probably mostly done abroad, but new jobs.
There’s a bright side to being the worst, where else to go but up? Not like we’re going to let better products come here like with the Chinese EVs while we struggle to decide on either fully backing renewables or continuing to boost up oil companies
I have gotten something similar to this so many times when asking either Google or Siri to play white noise. There was a time where about 60% of the time, Google would play a band or song named white noise, it was really friggin annoying.
Not soon enough. Friggin Google assistant on Android Auto has developed brain damage in the past 6 months or so. 90 percent of the things I ask anymore end up "I don't understand" to simple straightforward questions that if I wasn't driving could be found by a simple Google search within 39 seconds.
I, for one, quite like this idea. To tell you the truth, I can't tell you how many times I've tried to tell Google Assistant to pull up a location or play X song or Y podcast, and it just completely fails, causing me to take my eyes off the road to do what I need it to do. All the tin-foil "Oh ma gah data breach" goobers have clearly never heard of services like Cloaked. And funnily enough how much more data do you think Gemini will take vs assistant? Our phones already listen to us lol
You guys have no idea how many times current assistant gets the music I want played wrong. Gemini's voice analyzing capabilities are much much better. This is good news not everything related to ai is bad news
Am I the only one who's excited to be able to do more complex tasks than what Google assistant can do in my car? Seems like Gemini is strictly better than assistant at this point, can't remember the last time I asked my phone a question that it couldn't answer in a way I was happy with. That happens ALL the time on Google home and in my car.
So this about 25% of interactions with my Google Home at the moment:
"Okay Google, add Yellow Mustard to the shopping list."
"Okay, I added those two things...."
- Yellow
- Mustard
---
*I set a timer, the timer goes off*
"Okay Google, cancel the timer."
"There are currently no active timers..."
"Okay Google, STOP the timer...."
Google Assistant has actually gotten worse since Covid. It used to respond to commands and execute them in well under a second, it could even process compound commands like it was nobody's business (Hey Google, set the light to 50% and 2700k). Now days it can barely control a single light bulb without thinking for 10 seconds, and Gemini seems to be a massive downgrade from even that as far as commands on my phone are concerned.
It is like bigger screen than smart phones. It is also comming with a bunch of updates targeted at EV drivers. Apps to use or watch videos while waiting for their car get charged.
You think it was bad when touch screens took away all your buttons, wait until they start making cars that need you to talk to them to make them do anything, and it's all they give you because it's so high tech and luxurious and surveys show it's what people want.
The only thing I use Google's assistant for is playing music, and last I checked Gemini would just show Google results for whatever I wanted to listen to. Fingers crossed they manage to fix that before it's forced.... Maybe if I'm lucky it'll only take twice as long as assistant to start.
It could be good for things like opening up apps or asking questions while on the go but I feel like it could do more harm than good.
Although one benefit could be a locally running gemini model which can give you information even when offline. That would be truly useful when like camping for example.
I'm sure they'll run local models eventually, but we're not even close to that happening yet practically speaking. Gemini Nano is barely coherent for tool use and is too low-throughput on phone hardware — it's much easier to shunt requests to an API.
Qwen's great, but it isn't magic, and the economics still do not favour loading local models into memory every time you want to interact with an LLM. It's much easier to hit an API from almost every single angle.
If you want an offline model for when you're out camping (and for whatever reason, if you don't want to download a Wikipedia archive) you can already download Android clients for that, you don't need Google's permission.
Anything with a reasonable level of knowledge and low risk of hallucinating isn’t going to be running on device anytime soon
What they have currently on device (gemini nano) is what runs summarize, smart reply, etc. Every usable chatbot is cloud
Well it will but it would be a better source of information than nothing lol. Also gemini nano is incredibly mid and there are better local running models such as the recently released qwen 3 4b.
It arguably wouldn’t be a better source than nothing because it would routinely give out misleading info. Qwen is impressive, but you aren’t running it at 10 t/s on *most* android phones.
I doubt google wants to deal with the liability on that anytime soon, but It would be interesting if someone made a local app and side loaded it
I've been messing with local models on my desktop GPU and have been pretty impressed with general Q&A interactions (Ex. "write me an FFT algorithm in Rust" and it did. Slowly. With my entire GPU RAM filled.). The limits are there but it sure seems like you could run that class of model in something like a car with the right hardware acceleration, assuming you can get an automotive temperature grade high-end GPU or NPU which may not exist.
...Of course the incentives are to keep it cloudy so they can collect subscription revenue and/or user data, and keep the car itself cheap. There are some actual benefits like always having an up-to-date model and training data too of course. But I'd be far more excited if AI tech was more local and less corporate.
What would I do with it?
It's not like you can actually just go off of whatever answer it gives you without double checking it, since it's wrong so often.
I'm all for it.
The Android Auto suggestions are ridiculous. Half the time it will suggest me a destination I looked up on maps three months ago and the other one will be my home, when I haven't even left the street I live on yet.
I work remotely and if I get in my car around 4:45 there is a 99% chance I'm heading to the gym. The gym is never a suggested destination.
Google should know my routine and tell me what traffic is like.
116 Comments
tsar73@reddit
AwardImmediate720@reddit
Discovered_This_Week@reddit
su1ac0@reddit
ZombiePope@reddit
Recoil42@reddit
KirbyQK@reddit
Recoil42@reddit
Tw0Rails@reddit
ZombiePope@reddit
ionabio@reddit
PolarWater@reddit
PolarWater@reddit
ionabio@reddit
PurpEL@reddit
Thomas_633_Mk2@reddit
egowritingcheques@reddit
AmericanExcellence@reddit
Pestothepenguin69420@reddit
AdmiralZassman@reddit
94746382926@reddit
Tw0Rails@reddit
Brooklynspartan@reddit
Euler007@reddit
TransGeorgeCostanza@reddit (OP)
DiligentRow4825@reddit
Recoil42@reddit
gumol@reddit
Recoil42@reddit
OvONettspend@reddit
Recoil42@reddit
strongmanass@reddit
L9-HY8R1D@reddit
AwardImmediate720@reddit
Recoil42@reddit
TransGeorgeCostanza@reddit (OP)
SuaveBolo@reddit
Euler007@reddit
hi_im_bored13@reddit
Euler007@reddit
hi_im_bored13@reddit
Euler007@reddit
hi_im_bored13@reddit
Euler007@reddit
FriedEldenRings@reddit
Euler007@reddit
SubjectRevenues@reddit
Good_Air_7192@reddit
Whitewind101@reddit
XCCO@reddit
RobinsShaman@reddit
linyeraworking@reddit
Good_Air_7192@reddit
Virtual_Substance_36@reddit
hi_im_bored13@reddit
Snoo_31120@reddit
L9-HY8R1D@reddit
xlb250@reddit
hi_im_bored13@reddit
AndroidMercury@reddit
hi_im_bored13@reddit
pzduniak@reddit
Juicyjackson@reddit
hi_im_bored13@reddit
garrulinae@reddit
sasip39@reddit
WhatAreYouSaying05@reddit
L9-HY8R1D@reddit
REphotographer916@reddit
NecroCannon@reddit
undockeddock@reddit
userjack6880@reddit
Dignam3@reddit
L9-HY8R1D@reddit
SubjectRevenues@reddit
Jaiden051@reddit
L9-HY8R1D@reddit
reegz@reddit
thescreensavers@reddit
gonzo_gat0r@reddit
L9-HY8R1D@reddit
Yigsss@reddit
Icecum@reddit
xxBrun0xx@reddit
mazi710@reddit
Clover-kun@reddit
ionabio@reddit
bwoah_gimmethedrink@reddit
Individual_Serve_505@reddit
ionabio@reddit
MumpsyDaisy@reddit
PolarWater@reddit
Skippy989@reddit
turb0_encapsulator@reddit
undockeddock@reddit
FiveAccountsDeep@reddit
PM_your_Tigers@reddit
Pro-editor-1105@reddit
Recoil42@reddit
Pro-editor-1105@reddit
Recoil42@reddit
hi_im_bored13@reddit
Pro-editor-1105@reddit
hi_im_bored13@reddit
Pro-editor-1105@reddit
natesully33@reddit
turboash78@reddit
Conspicuous_Ruse@reddit
2Drogdar2Furious@reddit
Dirty_Dragons@reddit
KingKontinuum@reddit
Good_Air_7192@reddit
KingKontinuum@reddit
Good_Air_7192@reddit
jamesgilboy@reddit
KingKontinuum@reddit