Where can I find these LLM features in Firefox?

>Alternatives # There are a few projects that are forks of firefox with these features removed.

>I would think most non technical users would just use a different browser.

I would think they would list one or two of them under the “alternatives” section…?

Mozilla sure is going out of their way to alienate their remaining users. This is going to be Pocket 2.0.

Who asked for this? Who wants it? Certainly not the Linux / open-source crowd, and they're just about the only ones who are keeping Firefox alive.

If there's anybody from Mozilla or the Firefox dev team in this thread, I'd be interested to hear the thinking behind this addition.

They're all easily disabled in the GUI itself. The article is exaggerating, the closest argument is that it enables itself by default when it first updated which is fair, but they're easy to disable within the menu itself.

I understand why Mozilla have started implementing these features, it seems to have more mass market appeal than not - look at how popular “AI” powered browsers have become.

But boy does it not add extra effort removing these features every time there’s a new roll-out and it’s not done the best way IMO. I feel as if these features would go down better if Mozilla actually notified the user that they’re available and then offered whether to enable them or not (could have them enabled by default for new users). That way you’re still giving a choice, but in a more respectful manner.

If anyone is interested I’ve gutted all the more obscene stuff out of Waterfox and have instead left the useful ones such a ML translation, which is opt-in.

Related: I feel like onboarding is a lost art, more software should bring back software wizards and UI tours. Feels like you somehow have to intuitively know how something works (unlikely) or do a web search on how to use everything instead of having it shown to you nicely.

That's disappointing. If they can do EME-free builds, they can do LLM-free builds. This isn't a 'tricky issue'

https://hacks.mozilla.org/2014/05/reconciling-mozillas-missi...

The built in LLM translation is great. I'm leaving that enabled. It's useful and it's a perfect application for local LLM to be better than the equivalent corporate services.

I disable a ton in default FF and even run the unbranded versions so that it's not trialware (FF branded builds all expire when their baked in add-ons CA TLS certs expire). But the LLM translation? That's finally a good feature.

It isn't clear what browser.ml.chat and browser.ml.pageAssist are associated with in terms of features. Does anyone know? I tried disabling all shown in the write-up and local LLM translation still seems to work so I assume it's something else.

What is your favorite Firefox alternative?

There's a checkbox on whether you want to use it or not in the settings page, does this not change these settings?

I don't feel opposed to them changing the browser in principle--certainly there have been many improvements to web browsers over the years. Is privacy the concern here?

They’re opt-out and can be disabled in the settings or fully disabled in about:config. Definitely annoying but not enough to make me switch to a Chromium based browser.

This should not be considered an LLM issue. It's a user experience issue.

Mozilla has made changes that happened by default before. Often I have had to find the setting to put it back to how I wanted. I remember when it moved the URL bar to the bottom.

I don't think it is always an easy call to make. Tabs were a significant user experience improvement, but hiding it behind an opt-in would have limited it to people who knew about it.

I use Firefox as main on desktop and mobile. I have noticed messages on upgrade pointing to LLM features. I haven't engaged with them an from thereon haven't noticed any change because of them.

Saying there are reports of excessive memory or CPU use isn't terribly useful without references to those reports. One such report posted on HN was shown to have been unrelated to the LLM.

Are there any reports actually showing degradation because of LLMs rather than post hoc ergo propter hoc?

As long as it's local I think it's okay.

I also think that we in the long run will probably let machines do most tedious browsing for us-- digesting ad-ridden websites, digesting interfaces. The LLM navigates the actual web, presented to maximize revenue and maximize user engagement, time spent on the website etc., but we only see actual content, carefully arranged to be as comprehensible as possible, and if we want to communicate with somebody through a website controlled by others we formulate the message and the LLM submits it.

>I would think most non technical users would just use a different browser.

I would think most users would ignore the features they don't like? Idgi

Honestly, I haven't even noticed LLM features in the desktop version. But I found it really annoying in the mobile (iOS) app. There's a "Summarize Page" feature that's adjacent to "Find in Page." It's easy to mis-tap when you're just trying to search for a term on the page. It's also activated by the shake gesture, which can happen accidentally.

Three dots -> Settings -> Page Summaries to disable that.

Sharing my configs, hopefully they'll be helpful to someone. I don't even use Firefox but I keep the settings updated (prefs.js can be modified by Mozilla/Firefox, user.js does not change).

NOTE: Some settings might block too much, edit and use as you please.

https://rentry.co/browserconfigs OR https://rentry.org/browserconfigs

I've found the "Copy text from image" menu command to be useful from time to time, and the features I'm not interested in seemed easy to turn off/dismiss/ignore?

Wondering how these chatbot links in the sidebar are any different from bookmarks.

I think that (optional) AI features are fine, but the model selection only shows the big providers by default. You can enable local LLM support under about:config -> browser.ml.chat.hideLocalhost (set to false). It will check localhost:8080, which is not a great choice for a fixed port. I'd prefer if I could just point it at any OpenAI-compatible endpoint. Either way, seems to work well when serving a local model with llama.cpp.

Firefox for Windows now have a button “Talk to OpenAI”

What is this all about? Where would I find any of these LLM features? What do they even do?

When I go to about:config and enter "browser.lm", there's nothing displayed. That's on the latest 144.0.2 release.

Is any of it somehow related to this? https://www.heise.de/en/news/One-API-for-all-Mozilla-ends-LL...

For people using NixOS, you can wrap these configs in extraPrefs in wrapFirefox firefox-unwrapped:

Example: Put this expression(using lockPref to hardcode the config values) in environment.systemPackages(assuming "with pkgs"):

  (wrapFirefox firefox-unwrapped {

        extraPrefs =
        (
          ''
          lockPref("browser.ml.enable", false);
          lockPref("browser.ml.chat.enabled", false);
          lockPref("browser.ml.chat.hideFromLabs", true);
          lockPref("browser.ml.chat.hideLabsShortcuts", true);
          lockPref("browser.ml.chat.page", false);
          lockPref("browser.ml.chat.page.footerBadge", false);
          lockPref("browser.ml.chat.page.menuBadge", false);
          lockPref("browser.ml.chat.menu", false);
          lockPref("browser.ml.linkPreview.enabled", false);
          lockPref("browser.ml.pageAssist.enabled", false);
          lockPref("browser.tabs.groups.smart.enabled", false);
          lockPref("browser.tabs.groups.smart.userEnable", false);
          lockPref("extensions.ml.enabled", false);
          ''
        );

      })

It's important to remember that if people think of Firefox as a way to escape LLMs being pushed on them in Chrome, they might accidentally attract users from Chrome. It's the only possible reason why this wouldn't be a plugin. Why else would a user-focused browser consciously come up with strategies that attack user preferences?

You just don't understand the eleven-dimensional chess that you have to play to get from 30% marketshare to 2% marketshare. They have to think they have a winning strategy, judging by the way they talk to everybody who criticizes their decisions.

Mozilla seems to be an organization that is metrics driven to a degree that would make Robert MacNamarra blush.

I think this is why they keep shoving new features at users whether they want them or not, making them incredibly difficult to disable, rather than presenting an option try something new, or even making opting out of features easy and intuitive.

The latter two would lead to fewer users of the feature, which means it risks being removed for not being used by most users.

I'm not sure if I ever used any LLM features from Firefox besides exploring, but so far it seems great they're trying local approach instead of sending it to some "partner".

In the comment section here I see lot of people complaining about the fact it's enabled by default as well as some concerns about resource usage. Could someone experienced in desktop app architecture explain if disabling them functionalities makes Firefox that much faster or using less resouces? I'd assume that those functionalities are kind of loaded on demand?

"AI" became such a keyword that seem to instantly give either positive or negative response, it's also an advertised feature of every second app with many of them just forcing AI into you just because of hype. This doesn't seem to be a case in Firefox - so I highly disagree with the title - the features are there but they don't go into your way if you don't want them, therefore it's easy to just use it, only when needed

deleated from most recent linux install on a new to me desktop. vile I just want to work(create stuff), there is nothing that any "service" can do for me. the relentles attempts to get in the way of sending and recieving information is mindbogling, and I am going to be getting an analog telephone adapter, so that I can start useing fax, over cellular data, and making paper plans ,taking pictures, and converting to pdf's to avoid the wasted time and cost of software solutions the surreal part is that the LLM's have found me, and are recomending me to people looking for custom metal, so much for a low profile there, and somehow my number got out on indeed, and I got a wave of people calling, applying for "the job" but bitterly complaining that the app they downloaded as part of that, wasn't working right. fine, ok, whatever.

but not on my machine