"These 1 billion users may also be encountering Gemini in all those places, but that’s not the MAU metric. Pichai is talking only about people who are opening the Gemini app or visiting the Gemini web interface to enter a prompt or access Gemini Live. If you used Gemini only once in the past month, you are part of this cohort"
I'm pretty sure that Google replaced a shortcut for the hardware buttons on my phone to open some kind of Gemini interface in a security update and I had to change it back. And Gemini was auto-migrated from Google Assistant as the default digital assistant at a similar time(?) That alone would account for millions of users interacting with the app accidentally/unwittingly.
So as usual with these type of consumer choice/preference claims, it's sensible to be cynical. At least the Pepsi Challenge couldn't secretly swap out your water tap at home for a soda fountain, record you drinking it and claiming you prefer it to Coke. https://en.wikipedia.org/wiki/Pepsi_Challenge
That seems so impossible that it's absurd. I hear people talking about one thing. ChatGPT, nobody even mentions Gemini. If it's Gemini they're using, nobody is aware it is Gemini, and the only place that happens is on the google search page, not a Gemini app.
Though I guess it's possible all their 1B users are concentrated in India and China and I'd never know.
I'm sure my accidental clicks count for some of that. In the results page, they put an "AI Mode" button where the "All" button was, and the font doesn't have serifs on the 'I', so any time I return back from an image or video search, I end up on their Gemini page, because I don't notice the subtle change until I've already clicked on it.
Clearly they meant for people to click on it on accident, because it puts "All" in the second tab, and it otherwise doesn't make any sense to have the default tab be the second one.
The only time I use the Gemini response on purpose is when Cloudflare tells me I'm a bot and won't let me visit a web page, so I ask Gemini to scrape the relevant information for me, because bots are quickly becoming the only thing that can make it paste a Captcha. Sometimes that requires using the 'Show more' button, and it's not clear if that counts, too.
Also, I've heard they've begun rolling out replacing the Android voice assistant with Gemini, so a bunch of people are using it without trying to, even when they'd prefer the voice assistant. From what I've heard, the most common uses for voice assistants are setting reminders and controlling media players, and Gemini is much worse at that than the old Android voice assistant.
Gemini has gotten wildly worse in recent time, almost incapable of answering a question without a search, lengthy "thinking time" for trivial questions that no doubt is intended to mask waiting for inference resources
Yesterday I read that they sold a lot of TPUs to Anthropic as well as renting them out. This is odd, because I've constantly had issues with Google LLMs not having enough resources. Except for AI suggestions in Search, these work reliably.
Google did not sell those TPUs because they had excess capacity. Google sold them because it can make lots more money that way then using them for Gemini inference (the shift to sales is also indicative of some power dynamics within the company).
Same, quite often it will literally forget the thing from last message it sent me. I'd ask about car shops around me, it shows me a bunch of spots from Google Maps, cool. Then I ask 'what are their reviews' and I get hit with 'Reviews of what?'. It's still better in Android Auto than Siri is in Car Play, by a mile but random experiences in the app are very meh.
How are you accessing it? You may be using one of the limited models.
I use Gemini Antigravity daily for software design and coding, mostly using the Gemini 3.5 Flash model. It’s very competent, and quite a bit cheaper than Claude.
I was skeptical at first but these days I use Gemini on Android phone and Android Auto, search AI summary, and lately Gemma 4 a lot.
Skepticism is still needed though. The other day I asked Gemini while driving if there's a HoV lane going from 105 E to 110 S. It said yes. Actually there one to 110 N but not S. I did not act upon it because my co-passenger verified that Gemini was wrong.
Search summary is useful but not for things that require deeper technical expertise- those result in bad hallucination.
With corporate crunch on token budget, I've also started avoiding Claude and increasingly using opencode+Gemma 4, and developing some system prompts to iteratively improve the results.
I've found their models to be weaker and their inference much more unreliable, which is unexpected since they have their own TPUs and you'd think this is something they can handle. But no.
So true! Easy to make these kind of claims when you force it in front of everyone.
And its not like it is that good, especially the answers that are sourced from reddit without any vetting. But I still too often go with that damn answer.
> These 1 billion users may also be encountering Gemini in all those places, but that’s not the MAU metric. Pichai is talking only about people who are opening the Gemini app or visiting the Gemini web interface to enter a prompt or access Gemini Live. If you used Gemini only once in the past month, you are part of this cohort
If you have entered Gemini to try to figure out how to turn it off, you're also part of the cohort. A blast of gmail messages from Google has announced how it's part of everything and given instructions on how to find out the little you can do about it. Really, nothing.
A company that can't make a dominating coding model well, I've no confidence in its general purpose intelligence model; as a result, I don't ask Gemini anything.
Gemini needed four assisted attempts before it admitted it doesn't know how change a certain setting of a Google product and all answers before were hallucinations.
For me that's ridiculously bad and a waste of time.
They stopped updating the official documentation and there we are: their AI doesn't /know/ anything relevant for their products.
Good luck getting proper help in the present and future.
I stopped using Copilot because it was unable to answer Windows questions, often hallucinating or regurgitating decades old info. I sense a theme here.
Somewhat tangential, but when Google doesn't participate in frontier churn, earns money and make heavy investment in infrastructure that they rent to those participating in the churn, while still they are "losing the AI race".
But when Apple doesn't participate in frontier churn, earns money, and needs to rely on Google for AI services, then this is "the Apple way", "do nothing and still win" uber-pragmatic 5D chess game.
Not my experience, I've found it extremely useful for when I need a quick answer. It's handled very specific fairly obscure questions well. I also find it quite consistent.
It is completely useless and worse than quantized gpt-oss from over a year ago. It is designed for large scale, instant responses.
I hate when co-wokers or friends say "when I searched it up" and proceed to spew out whatever the Google AI said. I would consider listen if they ran it through ChatGPT, Gemini or Claude but whatever model is used on the search results is just bullshit.
Yeah, for a lot of my Google searches I don’t go past the AI generated summary part anymore. That being said, I do a lot fewer searches now I just ask ChatGPT the vast majority of my queries. I sure as hell haven’t hit page 2 on Google search in a long time.
I'm replying to the initial reply, let me paste it in for clarity:
> You mean the Ai that unwillingly pops up on my Google searches?
Assuming this is correct... Lets check:
from TFA:
Gemini is also core to Google’s flagship search experience, with AI Mode and AI Overviews becoming increasingly hard to avoid.
Oh, correct.
Assuming you have issues in your life, please keep the attitude away from your comments, it's unnecessary.
For what it's worth and what the article is about, I personally couldn't care less about an over-glorified penile measuring contest for the fastest growing userbase growth of a product that spits out inaccurate randomness as a service touted as the next best thing since sliced bread.
Agree. It's just the builtin android and chrome uses blowing up the numbers. By these metrics android assistant and siri had reached these levels in 2015. Gemini has no brand recognition outside of the tech worker sphere
I'm sure if you make Google Meet appear every time you made a Google query, it would also be the 'fastest-growing product ever'
This comment is exactly what’s wrong with hn. Confidently wrong and at the top of the comment section lol
Read the article?
"These 1 billion users may also be encountering Gemini in all those places, but that’s not the MAU metric. Pichai is talking only about people who are opening the Gemini app or visiting the Gemini web interface to enter a prompt or access Gemini Live. If you used Gemini only once in the past month, you are part of this cohort"
AI Overviews are not accounted for.
I'm pretty sure that Google replaced a shortcut for the hardware buttons on my phone to open some kind of Gemini interface in a security update and I had to change it back. And Gemini was auto-migrated from Google Assistant as the default digital assistant at a similar time(?) That alone would account for millions of users interacting with the app accidentally/unwittingly.
So as usual with these type of consumer choice/preference claims, it's sensible to be cynical. At least the Pepsi Challenge couldn't secretly swap out your water tap at home for a soda fountain, record you drinking it and claiming you prefer it to Coke. https://en.wikipedia.org/wiki/Pepsi_Challenge
That seems so impossible that it's absurd. I hear people talking about one thing. ChatGPT, nobody even mentions Gemini. If it's Gemini they're using, nobody is aware it is Gemini, and the only place that happens is on the google search page, not a Gemini app.
Though I guess it's possible all their 1B users are concentrated in India and China and I'd never know.
Well, maybe they lie, who knows. But "hearing people talking" is rather poor counterevidence.
I'm sure my accidental clicks count for some of that. In the results page, they put an "AI Mode" button where the "All" button was, and the font doesn't have serifs on the 'I', so any time I return back from an image or video search, I end up on their Gemini page, because I don't notice the subtle change until I've already clicked on it.
Clearly they meant for people to click on it on accident, because it puts "All" in the second tab, and it otherwise doesn't make any sense to have the default tab be the second one.
The only time I use the Gemini response on purpose is when Cloudflare tells me I'm a bot and won't let me visit a web page, so I ask Gemini to scrape the relevant information for me, because bots are quickly becoming the only thing that can make it paste a Captcha. Sometimes that requires using the 'Show more' button, and it's not clear if that counts, too.
Also, I've heard they've begun rolling out replacing the Android voice assistant with Gemini, so a bunch of people are using it without trying to, even when they'd prefer the voice assistant. From what I've heard, the most common uses for voice assistants are setting reminders and controlling media players, and Gemini is much worse at that than the old Android voice assistant.
Gemini has gotten wildly worse in recent time, almost incapable of answering a question without a search, lengthy "thinking time" for trivial questions that no doubt is intended to mask waiting for inference resources
Yesterday I read that they sold a lot of TPUs to Anthropic as well as renting them out. This is odd, because I've constantly had issues with Google LLMs not having enough resources. Except for AI suggestions in Search, these work reliably.
Google did not sell those TPUs because they had excess capacity. Google sold them because it can make lots more money that way then using them for Gemini inference (the shift to sales is also indicative of some power dynamics within the company).
Same, quite often it will literally forget the thing from last message it sent me. I'd ask about car shops around me, it shows me a bunch of spots from Google Maps, cool. Then I ask 'what are their reviews' and I get hit with 'Reviews of what?'. It's still better in Android Auto than Siri is in Car Play, by a mile but random experiences in the app are very meh.
How are you accessing it? You may be using one of the limited models.
I use Gemini Antigravity daily for software design and coding, mostly using the Gemini 3.5 Flash model. It’s very competent, and quite a bit cheaper than Claude.
Then hand out free annual pro subscriptions with phone data plans in India. I wonder how many of those are behind the 1B.
That's how I got mine, and I'm not in India. Cost like 20 bucks or sth. No warranty it will last a full year, but it's been over a quarter now.
I was skeptical at first but these days I use Gemini on Android phone and Android Auto, search AI summary, and lately Gemma 4 a lot.
Skepticism is still needed though. The other day I asked Gemini while driving if there's a HoV lane going from 105 E to 110 S. It said yes. Actually there one to 110 N but not S. I did not act upon it because my co-passenger verified that Gemini was wrong.
Search summary is useful but not for things that require deeper technical expertise- those result in bad hallucination.
With corporate crunch on token budget, I've also started avoiding Claude and increasingly using opencode+Gemma 4, and developing some system prompts to iteratively improve the results.
So high time to shut it down, right ?
- eventually they ll beat both openAI and Anthropic because I can literally hear google saying this in bane's voice
- "Ah so you think data is your ally? you merely scraped data, i was born in it, moulded by it, did not spare a single user until I was already a man"
I've found their models to be weaker and their inference much more unreliable, which is unexpected since they have their own TPUs and you'd think this is something they can handle. But no.
I get the impression that they'd prefer to sell as much of their TPU capacity as possible.
Which is definitely a risk-minimisation approach, which might leave them in a better place if/when this all collapses.
This is Microsoft bundling Internet Explorer with Windows all over again.
So true! Easy to make these kind of claims when you force it in front of everyone.
And its not like it is that good, especially the answers that are sourced from reddit without any vetting. But I still too often go with that damn answer.
From TFA:
> These 1 billion users may also be encountering Gemini in all those places, but that’s not the MAU metric. Pichai is talking only about people who are opening the Gemini app or visiting the Gemini web interface to enter a prompt or access Gemini Live. If you used Gemini only once in the past month, you are part of this cohort
If you have entered Gemini to try to figure out how to turn it off, you're also part of the cohort. A blast of gmail messages from Google has announced how it's part of everything and given instructions on how to find out the little you can do about it. Really, nothing.
A company that can't make a dominating coding model well, I've no confidence in its general purpose intelligence model; as a result, I don't ask Gemini anything.
what percentage are willing users making a choice to adopt?
Gemini needed four assisted attempts before it admitted it doesn't know how change a certain setting of a Google product and all answers before were hallucinations.
For me that's ridiculously bad and a waste of time.
They stopped updating the official documentation and there we are: their AI doesn't /know/ anything relevant for their products. Good luck getting proper help in the present and future.
I stopped using Copilot because it was unable to answer Windows questions, often hallucinating or regurgitating decades old info. I sense a theme here.
theyre sunsetting the google assistant that can "take me home"
Shouldn't it be Google Plus? ;)
Somewhat tangential, but when Google doesn't participate in frontier churn, earns money and make heavy investment in infrastructure that they rent to those participating in the churn, while still they are "losing the AI race".
But when Apple doesn't participate in frontier churn, earns money, and needs to rely on Google for AI services, then this is "the Apple way", "do nothing and still win" uber-pragmatic 5D chess game.
You mean the Ai that unwillingly pops up on my Google searches?
Assuming that's the one, it has given me many wrong answers lately and a sarcastic response calling it out makes it spit out a correct answer.
Pushing it for an explanation on every instance, it says that it ends up not relying on the query of the user to generate the answer. (?)
I've noticed that adding the line "check your answer before answering" gets it to replace English words got Chinese characters..
Funky stuff these flagship products.
Not my experience, I've found it extremely useful for when I need a quick answer. It's handled very specific fairly obscure questions well. I also find it quite consistent.
It hallucinates about 50% of the time. [0].
It is completely useless and worse than quantized gpt-oss from over a year ago. It is designed for large scale, instant responses.
I hate when co-wokers or friends say "when I searched it up" and proceed to spew out whatever the Google AI said. I would consider listen if they ran it through ChatGPT, Gemini or Claude but whatever model is used on the search results is just bullshit.
[0] https://oumi.ai/blog/oumis-study-finds-50-of-ai-overviews
> It is designed for large scale, instant responses.
Who wins when someone has to make 3 or 4 prompts to get an answer as opposed to just the initial prompt?
Seems likr studies like this need to be made at regular intervals to keep tabs on model functional, especially public facing ones..
Yeah, for a lot of my Google searches I don’t go past the AI generated summary part anymore. That being said, I do a lot fewer searches now I just ask ChatGPT the vast majority of my queries. I sure as hell haven’t hit page 2 on Google search in a long time.
Why are you "assuming" when reading the fscking article would tell you you're wrong?
How about you actually read the topic in question before putting a bitch in?
I'm replying to the initial reply, let me paste it in for clarity:
> You mean the Ai that unwillingly pops up on my Google searches?
Assuming this is correct... Lets check:
from TFA:
Gemini is also core to Google’s flagship search experience, with AI Mode and AI Overviews becoming increasingly hard to avoid.
Oh, correct.
Assuming you have issues in your life, please keep the attitude away from your comments, it's unnecessary.
For what it's worth and what the article is about, I personally couldn't care less about an over-glorified penile measuring contest for the fastest growing userbase growth of a product that spits out inaccurate randomness as a service touted as the next best thing since sliced bread.
AI Overviews are not accounted for in that number.
Conflated numbers. Show me counts of returning users and session lengths.
Agree. It's just the builtin android and chrome uses blowing up the numbers. By these metrics android assistant and siri had reached these levels in 2015. Gemini has no brand recognition outside of the tech worker sphere
My god, do people really lost the ability to read such short article?
No, built-in Android and Chrome are not blowing these numbers.
[dead]