Now we can argue whether or not it's an appropriate amount of disk space or bandwidth to use, but that's just a reasonable practical discussion to have. Framing it around consent is unnecessarily inflammatory and makes it harder to have a discussion, not easier.
Honestly, for most features you could justifiably say its fine. I mean honestly, how large is an English dictionary? 100 KiB? That is a far cry from 4 GiB. Just taking up 4 GiB of disk space without even asking is indeed a shit move no matter how you shake it. If Microsoft Word updated and suddenly took up 4 GiB more for something like a dictionary, it might not cause as much uproar as if it were something that many people are tired of hearing about and not interested in, but I'm not sure you would find a single soul who would find that acceptable, more just tolerated, probably partly because a lot of people simply wouldn't know better.
Shipping an AI model with a browser is starting to look like sticking cameras on ALL glasses, not just smart glasses, regardless of whether anyone wants that. Saying this is fine and not unusual is clearly motivated reasoning and just normalizes the surveillance state. It's very obvious the way this ends. Browser-based models will eventually be using your computer at the edge to save corporate money in the cloud while they do ever more expensive and invasive stuff to profile you.
For me the most significant problem is the lack of consent. I assume it's just not how you want to frame the problem. Ignoring the problematic parts or behavior of some sort of behavior is a common problem in modern software, and it's actually what the article is complaining about.
So where is the line we draw where bait-and-switch goes from being acceptable to unacceptable?
MA Chapter 93A for example clearly says that businesses are prohibited from "unfair or deceptive practices" including misrepresentation or concealing defects. Where do you think the line should be?
If you market a product as a Browser and it's codebase is 10% browser related and 90% some other program... Should Google have to correctly represent that product?
Related; If you didn't like when Apple forced you to use Siri on your phone, why did you purchase a Mac? Did you not expect them to continue disrespecting your sovereignty after you let them get away with it the first couple dozen times?
Chrome is not entitled to my disk space just because I installed it and Microsoft has been excoriated for the exact same behaviour with AI.
You're right in the sense that practicality and consent are orthogonal issues. There are probably stronger arguments to complain about a feature than the disk use.
What isn't part of the software? Can they just install as much garbage they want to, as long as they claim it is part of the "browser"?
Also, scale absolutely matters. If I pull up in front of your house and say "hey, mind if I park here?" and you say yes, then I park, walk away, and 10 minutes later park a fleet of 18 wheelers in front of your house, you're going to feel like I wasn't...entirely forthcoming about what I intended.
Framing this as needing "consent" is deeply misguided. It's as silly as claiming that Google stole your credit card info without your consent. It's just part of the software. You consented to installing the software and having it autoupdate. That covers it.
Now we can argue whether or not it's an appropriate amount of physical abuse to use, but that's just a reasonable practical discussion to have. Framing it around consent is unnecessarily inflammatory and makes it harder to have a discussion, not easier.
i'd rather have a popup asking me if i want to. I don't see firefox nagging me with GBs of data for local translations (arguably the only good use for everybody right now)
People installed an html client, not a backdoor into their computer.
When I download a web browser, it is reasonable to assume a piece of software that allows me to view web pages. Not an AI model.
The correct way to handle this, is for the vendor to announce the feature, the size and capacity required, and offer an opt-in, and not an opt-out.
This is beyond dispute.
I have a 2GB mobile data plan. If I was using Chrome, then some site triggers the Prompt API, that will cause Chrome to not only wipe out my data plan, but need 2 of my data plans. I don't find this reasonable.
This is exactly a consent problem, because I'm not denying it might be a useful feature, but it should be at the user's own informed choice. The fact that Chrome developers don't appear to see this might be due to them living in a bubble where they've never had to think about the costs.
we have every right to be upset at Google's audacity to suddenly gobble up 4 fucking GB
Also from the same webiste, Claude installing spyware: https://www.thatprivacyguy.com/blog/anthropic-spyware/
but discussion about that seems to have been suppressed on HN: https://hn.algolia.com/?q=Anthropic+spyware
It is consent - and its a pattern ubiquitous in tech.
Idk a random model being part of the software is not a given as much as things are trying to be pushed.
You make a good case for much stronger laws and regulations on what such consent can legally allow.
> Framing it around consent is unnecessarily inflammatory and makes it harder to have a discussion, not easier.
Spoken like a Google shareholder. It’s wild to see this level of gaslighting being presented as some sort of reasonable position.
Since when is an AI part of the browser?
On top it’s another abuse of their market domination. What if users prefer other models?
Were the terms something like
https://chromium.googlesource.com/chromium/chromium/+/refs/h...
"11.1 The Software which you use may automatically download and install updates from time to time from Google. These updates are designed to improve, enhance and further develop the Services and may take the form of bug fixes, enhanced functions, new software modules and completely new versions. You agree to receive such updates (and permit Google to deliver these to you) as part of your use of the Services."
https://www.gdpreu.org/the-regulation/key-concepts/consent/
"Where should the consent request go?
Consent information must be easily identifiable by the user. It should be presented separately from any terms and conditions."
https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-re...
"You cannot rely on silence, inactivity, pre-ticked boxes, opt-out boxes, default settings or a blanket acceptance of your terms and conditions."
https://www.edpb.europa.eu/sites/default/files/files/file1/e...
"The element free implies real choice and control for data subjects. As a general rule, the GDPR prescribes that if the data subject has no real choice, feels compelled to consent or will endure negative consequences if they do not consent, then consent will not be valid.13 If consent is bundled up as a non-negotiable part of terms and conditions it is presumed not to have been freely given."
https://www.dpo-consulting.com/blog/gdpr-data-consent
"GDPR Article 7 further tightens consent. It requires clear requests (separate from general terms), a right to withdraw at any time, and documentation to demonstrate that consent was validly obtained. In short, you must prove that a person knowingly opted in. Records of data consent (who, when, how) are mandatory so that you can show regulators you followed GDPR consent requirements."
"You wanted a banana but what you got was a gorilla holding a banana and the entire jungle." - Joe Armstrong
You installed a banana with autoupdates enabled by default. Therefore you consented to installing a gorilla and an entire jungle
Anyway, joking aside, what's missing from this blog post is discussion of potential remedies for the alleged violations
It may be acceptable to Google to violate GDPR, etc. if the remedies enforced are merely a "cost of doing business" and not a threat to business success
https://developer.chrome.com/docs/ai/prompt-api
When Chrome 148 releases tomorrow, this will be the default behaviour on desktop.
To download, it should check for 22 GiB free disk space on the volume where your Chrome data dir is, and at least double the model size of free space in your tmp dir.
That said, you might be surprised to learn that some of the models from 3b-9b could probably replace 80% of the things nonvibe coders use chatgpt for.
Its a good idea to run small models locally if your computer can host them for privacy and cash saving reasons. But how can you trust Google to autoinstall one on your machine in 2026? I just couldn't do it.
I find it works fine for simple classification, translation, interpretation of images & audio. It can write longer prose, but it's pretty bad.
It can also write text in the format of a JSON schema or regexp for anything you might want to do with structured data.
Not happy about that as I would like to see more local models but that's the current state of things.
https://sendcheckit.com/blog/ai-powered-subject-line-alterna...
One option I'm leaving as default is "Use LiteRT-LM runtime for on-device model service inference." Any comment on that?
"optimization-guide-on-device-model"
- Enables optimization guide on device
"prompt-api-for-gemini-nano"
- Prompt API for Gemini Nano
- Prompt API for Gemini Nano with Multimodal Input
and deleted weights.bin and the 2025.x folder in "OptGuideOnDeviceModel"
Will report if Chrome 148 downloads the model again.
That other flag is for using a different open-source inference engine to the (from what I can tell) closed-source one that's used by default.
As much as I’m against unexpected 4GB bloat for an AI model, I’d much prefer it to install one copy, system-wide. 4GB per Windows or Linux lab machine, rather than a 4TB minimum load on our NFS server and 4GB downloads per user, per machine on our Windows labs.
2018? An estimate from 8 years ago is going to be off by a factor of 10 or so.
Not sure you'd get far with the legal arguments unless you're actually a lawyer. Too easy to misunderstand the jargon (i.e. the same reason why it's dangerous to use an LLM as your lawyer).
(As an aside, the whole thing reads to me like the style LLMs use; not saying for sure it was, just giving me those vibes).
I have gigabit internet (125 MB/s). This would imply when I'm downloading something I'm using 18 to 45 kW of electricity. Completely bonkers.
Additionally, the cited number also conflates wired internet (low power consumption) with mobile internet (higher), even though this model is only being downloaded to Chrome Desktop AFAICT.
EDIT: got the maths very wrong with some other estimates, deleted them.
Quitting chrome these days is the easiest thing to do. The writing is on the way. You don't control the browser on your network, google does. ANd for better or worse, google's priority is AI at this time.
Sysadmins should take notice.
If the network is ~65% chrome and thus deemed painful, take the gradual approach. Do not push chrome on new devices or users. Watch that problem slowly go away.
There’s nothing stopping Google Chrome from doing something similar except, I suspect, Google knows or feels it will result in many fewer installs of its bloatware.
So no, I don't think it's a weird trend at all that people start describing software as "silently" doing things when trust in automatic updates of software (a thing that software silently does) has deservedly gone down the drain in the last few years.
A 4 GiB model has nothing to do with the functionality of a web browser. It is something forced on users without their consent.
Of course that's what we get for giving the benefit of doubt to the company that insisted on learning the wrong things from the Google Buzz fiasco.
Delete Chrome's silent 4 GB AI model file and AI
In Chrome, go to: chrome://flags
Search for and Disable these:
Enables optimization guide on device
Prompt API for Gemini Nano
AI Mode
Click the Settings (gear icon).
Go to AI Innovations and uncheck Enable AI assistance.
sudo rm -rf ~/.config/google-chrome/OptGuideOnDeviceModel
sudo rm -rf ~/.config/googlechrome/Default/OptGuideOnDeviceModel
sudo touch ~/.config/google-chrome/OptGuideOnDeviceModel
sudo chmod 400 ~/.config/google-chrome/OptGuideOnDeviceModel
sudo touch ~/.config/google-chrome/Default/OptGuideOnDeviceModel
sudo chmod 400 ~/.config/google-chrome/Default/OptGuideOnDeviceModel
sudo chown root:root ~/.config/google-chrome/OptGuideOnDeviceModel
sudo chown root:root ~/.config/google-chrome/Default/OptGuideOnDeviceModel
ls -l ~/.config/google-chrome/OptGuideOnDeviceModel
ls -l ~/.config/google-chrome/Default/OptGuideOnDeviceModelhttps://www.reddit.com/r/chrome/comments/1t536x6/psa_chrome_...
"On-device AI" can be disabled. At least, it is in Chrome on Linux Desktop.
Gemma 4 E4B is a much better model, but it's too large to simply download and run everywhere.
IMHO, this is jumping the gun. Google's going through a lot of effort to release a model that will give everyone a very poor first impression of what on-device models are capable of, souring it for everyone for a long time afterwards. It would be better to wait until a smaller, better model ships before doing this.
It's good to have something to work with if these Web APIs are going to be part of a standard. I suppose this means that ALL the browser vendors are likely to implement something
Mozilla has taken a strong stand against the prompt api.
I wonder what that will do for the competition between hosted genai and local models...
You're not even the customer when it comes to Google.
They simply read your mails, how would you expect there to be anything resembling decency in a company like that? It is the ad business.
Bad thing is that people still use gmail.
Not everyone has access to the same infrastructure you have.
An xBox game can be 50+ gigs. Millions of gamers. Fire up the presses!
I'm not at all saying nothing matters so we shouldn't care. I just disagree about the utility of calling out specific things out of proportion to their place in the climate crisis. Tackle AI, yes, and fast fashion and cars, and ... that one change to Chrome? I guess if that's where you want to put your energy, Sisyphus.
So make your own judgement, but this seem pretty significant to me.
[1]: https://www.aboutchromebooks.com/global-chrome-user-base/ [2]: https://www.iea.org/commentaries/the-carbon-footprint-of-str... [3]: https://www.anthesisgroup.com/insights/what-exactly-is-1-ton...
2.5 million downloads of 4 GB are 10 PB of traffic.
I think there are be a lot more than 2.5 million Chrome users in the world.
Sorry folks, your low bandwidth situation is not, in fact, a climate change emergency.
[1] Used since forever by the Tobacco & Pharmaceutical, Fossil Fuels & Climate, Food & Diet Industries.
is about 60 gramms of co2 per user?
For one, not everyone in this world lives on high bandwidth unmetered connections. In Germany, you got a lot of people still running on 16 MBit/s ADSL, that's half an hour worth of full load just for AI garbage. With the average 50 MBit/s, it's still 10 minutes. For those running on hotspots - be it their phone with often enough 10 GB or less on your average data plan or train hotspots that cut you off after 200MB - the situation is similarly dire.
The other thing is storage. I got a nominally 256GB MacBook Air. Of these 256 GB, easily 50GB are already gone for macOS itself, swap, Recovery and everything that macOS doesn't store as part of the immutable partition (such as, you guessed it, its own AI models). Taking up 2% of the disk space without consent is definitely Not Cool.
When Firefox does it, it sparks outrage across the internet, with entire forums filled with people vowing to leave Firefox forever and switching to something like Waterfor or Ilp/Zorp/Floop instead.
As a result, searching for experiences other people had with Firefox makes it sound like hell on earth, while people have little more to say about Chrome other than "Google gonna Google, but it's fast at least".
It's how you get things like "Browser monocultures are an issue, so don't use Chrome (Blink), use Brave (Chromium (Blink)) instead!" said in earnest.
Curious if Google plans to allow other browsers doing that too.
99.99% I do not need Chromium but when I do, it's worth the ~200MB of used space.
you would think google is not stupid enough to mess with gcp account holders
With Librewolf I can get proper WebGL, full UBo -with the AI blocklist too to avoid all the slop- and Bypass Paywall Clean from Giflic or whatever was called. Yeah, eh, y local newspaper won't mainly get adverts' money but the rest of local company ads show up well even with UBo/BPC, so they get some money after all.
On RAM usage, Librewolf it's far lighter on the long term and it doesn't ping back as Firefox, and many times less than Chrom* based browsers where, I repeat, Chrome based browsers don't allow UBo any more even if installed from their Github repo enforcing some about:flags variables related to legacy extension support.
The web today without UBo it's unmanageable. Popus, more than the ones from 2003, malware disguised as ads even on mainstream, safe sites, and all of these running zillions of cookies and trackers converting your -otherwise perfectly usable- old amd64 Celeron machine with 2GB of RAM into some crawling Pentium III with 256MB of RAM. With LibreWolf and UBo I could even test Yandex Maps with Prypiat and the like and InstantStreetView too. No slowdowns, no OpenGL >= 3.3/Vulkan video card required, and no need to own a 8GB machine.
HN developers there without UBo if they depend on the web for documentation they are bit screwed if they use Chrom* based browsers, sorry. Half of the resources for their machines coudn't be used, you know for IDE's, compilers, virtual machines/containers and whatnot. And, yes, I know about ZRAM under GNU/Linux, and just imagine how many tasks would anyone accomplish with a ZRAM compressed chunk (~1/3 of the physical RAM), a light desktop environment as Lumina/LXQT and a non-Chrom* browser blocking all pests. Up to 3X more tasks in the same machine. No need to waste money on upgrades, and compilng cycles are cut down for the good.
For as long as the funding for Firefox continues, it remains a viable option. And despite all their bad decisions of late, they still give users the ability to configure or disable user hostile components.
Their funding model is a risk, but I've been using Firefox and librewolf forever and I'd argue it's a much better option than chrome or edge, especially with a handful of plugins. A risk is still better than the actual realization of the risk.
Question from last November, even referring to macOS, by @paulirish: https://superuser.com/q/1930445/can-i-delete-the-chromes-opt...
With policy setting, debug url, docs in the answers.
One search away.
# From one's $HOME dir:
rm -fr ./.config/google-chrome/OptGuideOnDeviceModel
mkdir -p ./.config/google-chrome/OptGuideOnDeviceModel
touch ./.config/google-chrome/OptGuideOnDeviceModel/weights.bin
chmod 0400 ./.config/google-chrome/OptGuideOnDeviceModel/weights.bin
chmod 0500 ./.config/google-chrome/OptGuideOnDeviceModel
This has, so far, kept Chrome from (re)installing that file on my system.
Hypothetically the parts involving weights.bin aren't needed so long as the containing directory is not writable.
Environmental analysis for operations? Not a fan of thinking in such terms.
> For users on capped mobile data plans, particularly in regions where smartphone-as-only-internet is dominant (much of Africa, much of South and Southeast Asia, most of Latin America), 4 GB of unrequested download is on the order of a month's data allowance, vapourised by Chrome on the user's behalf. Google has not, to my knowledge, published any analysis of the welfare impact of this on the populations whose internet access is metered.
THIS is a valid concern. Otherwise I'm not buying into "ask for consent because of dependency X". Users don't like questions/consents.
However OS (at least windows) has an way to set network connection as a metered so software can make informed decisions. Also Android has "Data Saver" function which should also be honored by software.
Why not? It's about 60 000 London - New York City flights by the way (https://www.theguardian.com/environment/ng-interactive/2019/...). And what's the benefit again?
Are they against washing machines too? Or are they just grandfathered in?
The oceans are boiling [0], marine life is dying [1]. Land close to the water will be land under water soon [2]. The ice caps are melting and setting free all sorts of diseases. [3]
Large parts of our planet on fire all the time now, here's one from Australia from this year [4], but I'm sure you've read about wildfires in Australia last year, California every year, Greece last year etc etc.
What you're proposing is nothing short of a death cult. It's either degrowth or we all die, sacrificed at the altar of capitalism.
[0] https://www.theguardian.com/environment/2026/jan/09/profound...
[1] https://www.nature.com/articles/s41559-026-03013-5
[2] https://www.nature.com/articles/s43247-025-02299-w
[3] https://www.unep.org/news-and-stories/story/could-microbes-l...
[4] https://phys.org/news/2026-01-australia-declares-state-disas...?
Unfortunately, that automation is unreliable. It doesn't work across operating systems - Windows laptops won't enable data-saver mode when connected to iPhones and macOS laptops won't when connected to Android phones, and neither will enable it when connected to, say, public transport wifi.
And even if the OS has the information, websites can't reliably use it either. Firefox and Safari both don't implement the NetworkInformation API [1].
[1] https://developer.mozilla.org/en-US/docs/Web/API/NetworkInfo...
All of your history, trivially searchable. Imagine the waste heat generated by the browser bar conducting thousands of non-consensual searches every time you type.
https://web.archive.org/web/20260505052217/https://www.thatp...
https://archive.ph/sM7O5 (missing images and styling, but the content all seems to be there)
The good point in this article is about how the "AI" features in Chrome all use Google's cloud API and not a local model. That's true and some of it should be local. ("AI mode" uses the Web index, so it fundamentally cannot be local, but there are features that could be.)
Or Firefox of course.
https://adsm.dev/posts/prompt-api/#which-browsers-support-th...
languagemodel should be an OS service..
It reminds me of the "dialup warnings" common 2 decades ago on huge pages (often containing many images). Yes, bandwidth and storage has gotten cheaper, but the unwanted waste should still be called out. I'm not even anti-AI, having waited several hours recently to get some local models to experiment with, but that's because I wanted to and made the decision to use that bandwidth.
https://chromeenterprise.google/policies/#GenAILocalFoundati...
The prompt API can be tested here: https://chrome.dev/web-ai-demos/prompt-api-playground/
It would be really helpful if there was a way to download the model to a central location, so multiple users on a single system could easily share it.
Brave has always just worked for me and seems light on memory usage. Dunno why anyone would use chrome.
A lighter Brave.
Works without Javascript, no CAPTCHA, no DDoS, no geoblocking, etc.
https://web.archive.org/web/20260504192142if_/https://www.th...
So if you see this as just a new feature that provides some on-device AI, it's a bit, so what? A new feature? The last GT7 or Flight Sim patch was bigger than this, what's the big deal, etc.
However, that's not really what's going on. It theory Chrome gives you a local LLM that can provide local AI powered features. In practice, everything gets sent to the cloud anyway so the local LLM seems mostly to exist as a disguise for that, which is shady AF.
As others have pointed out, the solution is https://www.firefox.com/. And whilst it's been trendy on HN for several years to slag off Firefox and Mozilla, I went back to Firefox as my daily driver several years ago, and Chrome's high-handed enforcement of Manifest V3 extensions (meaning no full fat uBlock Origin) has only served to cement that decision.
It's mostly been great. The only downside is that some sites don't work properly on Firefox, and I'm 99.999% sure that's not Firefox's fault.
For example, Paypal's post-login verification step breaks so every time I want to buy something using Paypal I have to switch to Chrome. And, no, disabling uBlock Origin and other extensions on Paypal doesn't help - I've done this already. Seriously, Paypal, it's been months: will you please just fix signing in and paying on Firefox, please?
And many sites will assume you're a bot first and ask questions later if you hit them with anything other than Chrome or Safari... which is also extremely lame and scummy.
If Chrome had installed 4GB for some other tooling that most people don't need, would anyone care? My operating system installs with a million default packages that I don't need. Users install applications with optional features all the time. Applications install additional tooling so that they'll function all the time.
To the other point: of course Claude Desktop modifies the browser--that's how it works. Most apps install integrations with existing apps. Often apps install a whole collection of plugins, even for things the user doesn't use, so they're available if the user does start using the other apps.
The fact that this happens to be AI-related is a moot point. The environment concern is utter nonsense. They're not using everyone's browser to power AI for others as some kind of shared collective resource. 4GB is not a lot of data in the grand scheme of things (beyond general application bloat). I have more than 4GB worth of ads shoved in my face every month.
The legal argument is facile as well. When you install any application, its terms of service cover functional updates and additions. You don't have to explicitly consent to all of them.
Other than the size of it, I don't have any problem with anything this article is mentioning.
This is a huge nothingburger that only caught peoples' attention because of the irrelevant mention of AI.
Of course I would. It’s already the largest application on my computer, and I only keep it around for when a site doesn’t render right in Firefox.
This kind of size increase clearly pushes it over the line for me and it’s getting uninstalled.
Have you seen SSD prices recently?
He even boasts about it on twitter.
Meanwhile car batteries are draining faster. Costing the consumer electricity & battery cycle life.
Google Chrome just exists to make Google money at your expense, to sell your data and deplete your battery.
npm install …
Here chrome is just installing things behind your back, whether you really want it or not.
I haven't touched a Chrome browser in a very long time and I just hope that other vendors don't take a similar route.
Can it be the basis for nano-openclaw?
Can I use it to run a Karpathy optimization loop?
A local Gemini Nano might be useful!
> the features that do use the local model (Help-Me-Write in <textarea>, tab-group AI suggestions, smart paste, page summary) are buried in textarea-context menus and tab-group right-click menus
on android aicore: mediatek, qualcomm, aosp vendors, and google will pull down models you cant touch
> The roughly 4 GB × N devices of disk-storage cost, sustained, on user hardware. SSDs have a per-GB embodied carbon cost of approximately 0.16 kg CO2e per GB of NAND manufactured [18]
The estimated environmental aspect of the download also seems like an overblown point, noted for sensationalism. There are always hand-wavy numbers involved and I had to look no further than the quote above to find evidence of this. The reference for [18], "The dirty secret of SSDs: embodied carbon", incorrectly links to "Toward Carbon-Aware Networking" and makes no mention of the environmental cost of SSDs. After looking up "The Dirty Secret of SSDs: Embodied Carbon" myself, I was able to see the same methodologies as I was expecting used [1].
> We conducted an analysis encompassing 94 Life Cycle Assessment (LCA) reports, which collectively quantify the embodied cost of SSDs. Owing to the scarcity of direct and up-to-date LCA studies focused specifically on SSDs. We compiled a dataset comprising LCA reports pertaining to Server, Workstation, Desktop, Laptop, and Chromebook products, all of which feature SSDs
All these studies rely on metrics extrapolated from layered assumptions and end up being used by those who try to use them as objective numbers.
Which apparently means it'll never install btw, even if I were to run Chrome. Another comment said they check for 22GB free space
4gb definitely isn’t a negligible amount of space on most people’s devices.
The quite successful it would seem MacBook Neo has 256GB of storage in its base configuration.
A MacBook Air and a basic sub $1000 Dell laptop starts at 512GB.
> To put 4gb in context, I currently have 2 tabs open that nearly take up 4gb.
You are conflating disk and memory.
> The fact Chrome also has a way to disable this makes it kind of a nothingburger in my opinion.
There’s a reason they picked an opt-out model for this, and not an opt-in approach.
But I also see the point in it. We recently did a hackathon, and I considered relying on Gemma 4 for privacy considerations. The local model could interpret the user’s natural language request and derive less privacy revealing requests to form based on that.
But then, a web app that shows people a loading screen while it downloads a 4GB model probably wouldn’t be a best-selling UX.
I never conflated anything. I said it's a neglible amount of space for current hardware, which I still believe.
If anything, the fact that I think the amount of space is acceptable for the amount of ram a modern laptop has exaggerates the point.
> There’s a reason they picked an opt-out model for this, and not an opt-in approach.
That's the approach they take for most of their features.
> But then, a web app that shows people a loading screen while it downloads a 4GB model probably wouldn’t be a best-selling UX.
Which seems to be the motivation of having these local models embedded in the browser's available resources: https://developer.chrome.com/docs/ai/prompt-api
..will tell you everything you need to know - including model state, file path, device capabilities, etc.
And there's a single button to uninstall the model.
There is also the ability to load a model from a central location, as suggested by another commenter here, although I haven't tested it yet.
The official chrome.dev Prompt API Playground linked in the thread doesn't work.
Chatgpt made a me tiny chrome extension to test the prompt and summariser api's when they announced last year - my laptop wasn't capable the time but these newer models are obviously smaller and more efficient, so it has sprung into life.
Full prompt and code is on pastebin `7Ja3ATHZ` if anyone wants to test quickly. It summarises the current page and brainstorms app ideas based on the summary.
https://chrome.dev/web-ai-demos/summarization-api-playground...
and you can watch it generate the output token by token in the Event Logs:
chrome://on-device-internals/
I blame the kids these days (waggles fist), downloading their Pokiman shows at 4-5gb a pop! No respect for their disk space limited elders.
I'm actually gonna have to uninstall Chrome from a few machines tonight.
If you don't like be treated like anything but human, you should seriously consider replacing chrome with ungoogled chromium or other browsers.
Javascript running on a page can use a feature that requires a model to be downloaded.
I have pages that use it, or other LLM models via LiteRT or HuggingFace transformers.js.
I try to warn the user, but that is my responsibility as a page author. I like that this is enabling the web platform to remain competitive.
The author is pulling a long bow by trying to claim this is some GDPR violation. Have they ever used the web? There are inefficient sites everywhere, with autoplaying video etc.
4GB isn't nothing, but if a page wants to use it then hopefully it is useful to the user!
The logic around not providing access to model version to prevent fingerprinting is laughable when the suggestion to counteract fingerprinting from prompting is the model should only update when user agent string updates. Just put the damn API behind a explicit user permission.
That is actually an interesting question. I would think that any given amount of processing is at least slightly more efficient when running on large-scale cloud hardware, but I have no idea whether it's a few percent or 75%. There's also some overhead, so maybe there's no gain for smaller jobs.
Google should have asked.
Oh my god thats terrible I hope you continue this article in this mode and dont pivot to some unsubstantiated bs claim that makes absolutely no sense...
>At a billion-device scale the climate costs are insane.
sigh.
Imagine if everyone on the planet start using a memory hogging, cpu chugging browser application what a terrible hazard that would be for the climate.
Oh and it might have an AI component in it.
This claim is worse than the AI in data centers boiling the earth claims.
We can measure carbon released down to the watt. If you have an issue with people using power, shut up and talk to your government about carbon taxation/moving to alternative power sources. trying to shame some power users, quite arbitrarily isn't just senseless its self defeating. Its a measurement problem, the second people start getting shaky measurements of what their neighbors are doing, they start trying to shift the blame.
The file might be 4GB but the transfer sure as heck wasn't, so what are we even talking about? How much data is actually transferred? Can someone just grab that weights.bin file and zip it up with max compression and report a more realistic number that we can do the math with, if the number is even worth doing the math for?
OH MY GOODNESS, this is the WORST headline.
If Google Chrome comes with an AI model, and you install Chrome of your own free will, you just gave consent.
The "climate costs" are happening whether or not the AI is there. Sure, maybe it makes the hardware work a bit harder, but like, come on. I'm still using my computer anyway. YOU are the one costing the climate, not Google. You're the one turning the "On" button on.
I don't even know why headlines like these are taken seriously.
On the one hand, Waymo seems to have a better safety record than Tesla does. That's not nothing. For someone nominally in charge of SpaceX like Elon is, it's a red flag
On the other, Google does things like this with Chrome, and also they arguably censor. It's irritating
You can also ask why the US government fails to protect the users. Corporate dictatorship at its finest.
God, I'm SICK of this AI slop style. After ingesting terabytes of pirated books you'd expect a little bit more variety in it's writing.
>After ingesting terabytes of pirated books you'd expect a little bit more variety in it's writing.
I think it's the result of post-training. The original base model most likely had a less slopy style. This style is what AI companies think is a good style (they specifically train for it).
Maybe manually point to a local endpoint with one of the multiple 100s of GBs of models that I already downloaded? Even better
Don't even know what the model is for... but as things are, one of the alternatives would probably be send high amounts of user data to google which is worse. I would personally pay to have a 4GB download and keep my data/privacy
So this is an odd take from a "privacy guy"
If you're okay with 4GB being added, where would you draw a line? What if it downloaded a 40GB file? 400GB?
Lately Firefox has been getting better, but I still prefer Chrome for almost all my needs, so I stick to it. This barely even makes a difference to me. If it was 400GB however it would make a difference to me, and I would make more of an effort to switch to something else.
The thing about these kind of arguments is that any economic activity or any sort of action involves some load on climate. The magnitudes are important.
In this case: a single hamburger does the same amount of emissions as 50 such downloads. What’s really the point of this kind of virtue signalling?
Hamburger is usually held up as a grotesque example in climate talk and can't be consumed with a clear conscious so are downloads insanely worse than we thought or is a hamburger not even in the same realm of climate damage as usually claimed.
cloud: FUCK YOU GOOGLE
These days, I just navigate away from anything that demands I use Chrome "for best results." One of the sites for a local utility company does this, so instead I just call monthly and pay or manage my service by phone. I'm old enough to remember when that was the preferred way after mailing personal cheques went the way of the dodo, so it does not feel that inconvenient to me, but I can see where it might for other people. Still, nobody said the fight to regaining our agency online would be easy. Or convenient.
I hate how much companies don't care about efficiency or their customers. It's like windows 11 requiring like 2 more GB of RAM just to see your desktop, what an upgrade, yuck.
I think the only time I've ever had to use Chrome instead of Firefox was because of some USB device thing that worked inside Chrome. Otherwise everything just works in Firefox.
I can recommend the following extensions:
- Youtube Enhancer
- DuckDuckGo Privacy Essentials
- Cookie Auto Decline (a MUST for Europeans)
- Slop Evader
- No Gender (a MUST for Germans)
Its a totally different browsing experience than what most people have.
I recently watched my kiddo looking something up with Edge on her laptop. I had to interfere and install Firefox. It was ridicolous!!! The amount of spam on the screen. How people can cope with this is beyond me. Especially if the solution doesn't cost anything. Just Firefox + some free extensions.
edit: because people asked about the No Gender extension:
Germany didn't have “gendered” language, until it was introduced some years ago.
Imagine the sentence: The teachers explain to their pupiles that the managers work only for the shareholders.
in regular German, it would translate to:
Die Lehrer erklärten den Schülern, dass die Manager ausschliesslich für die Anteilhaber arbeiten.
In gendered German, it became:
Die Lehrer:innen erklärten den Schüler:innen, dass die Manager:innen ausschliesslich für die Anteilhaber:innen arbeiten.
For me, it ruins the reading experience.
Gosh most of the time when I read people complain about firefox, it gives me the impression they have not even used firefox.
Chrome doesn't do that. I literally can't use Firefox anywhere I don't have a power socket.
My laptop also becomes a toaster.
[1] https://www.jwz.org/blog/2024/10/mozillas-ceo-doubles-down-o... [2] https://blog.mozilla.org/en/mozilla/leadership/mozillas-next...
This is satire, obviously.
Yes, 4GB of unintended traffic can absolutely wreck someone's finances.
Hell, a few hours of browsing Tiktok or Youtube will take more bandwidth.
This is fake engagement bait.
