I do think dipping your toes into the future is worth it. If it turns out the LLM is trying to kill us by cancelling our meetings and emailing people that we're crazy that would suck. But I don't think this is any more dangerous than giving people a browser in the first place. They have already done enough to shoot themselves in the foot enough.
This whole backlash to firefox wanting to introduce AI feels a little knee-jerky. We don't know if firefox might want to roll out their own locally hosted LLM model that then they plug into.. and if so, if would cut down on the majority of the knee jerk complaints. I think people want AI in the browser, they just don't want it to be the big-corp hosted AI...
[Update]: as I posted below, sample use cases would include translation, article summarization, asking questions from a long wiki page... and maybe with some agents built-in as well: parallelizing a form filling/ecom task, having the agent transcribe/translate an audio/video in real time, etc
I don't. And the whole idea of Firefox's marketing is that it won't force things on me. Ofc course om frustrated. My core browser should serve pages and manage said pages. Anything else should be an option.
I'm beyond tired of being told my preferences, especially by people with incentives to extract money out of me.
I guess it's nice for non-technical people who don't know how to use `about:config` but beyond that I don't really see the need. Hopefully adding that extra layer of indirection doesn't mean the users will have to wait too long for security patches.
> Machine learning technologies like the Bergamot translation project offer real, tangible utility. Bergamot is transparent in what it does (translate text locally, period), auditable (you can inspect the model and its behavior), and has clear, limited scope, even if the internal neural network logic isn’t strictly deterministic.
This really weakens the point of the post. It strikes me as a: we just don't like those AIs. Bergamot's model's behavior is no more or less auditable or a black box than an LLM's behavior. If you really want to go dig into a Llama 7B model, you definitely can. Even Bergamot's underlying model has an option to be transformer-based: https://marian-nmt.github.io/docs/
The premise of non-corporate AI is respectable but I don't understand the hate for LLMs. Local inference is laudable, but being close-minded about solutions is not interesting.
My take is that I'm ok with anything a company wants to do with their product EXCEPT when they make it opt out or non-opt-outable.
Firefox could have an entire section dedicated to torturing digital puppies built into the platform and... Ok, well, that's too far, but they could have a costco warehouse full of AI crap and I wouldn't mind at all as long as it was off by default and preferably not even downloaded to the system unless I went in and chose to download it.
I know respecting user preference doesn't line their pockets but neither does chasing users down and shoving services they never asked for and explicitly do not want into their faces.
> Large language models are something else entirely*. They are black boxes. You cannot audit them. You cannot truly understand what they do with your data. You cannot verify their behaviour. And Mozilla wants to put them at the heart of the browser and that doesn't sit well.
Am I being overly critical here or is this kind of a silly position to have right after talking about how neural machine translation is okay? Many of Firefox's LLM features like summarization afaik are powered by local models (hell even Chrome has local model options). It's weird to say neural translation is not a black box but LLMs are somehow black boxes that we cannot hope to understand what they do with the data, especially when viewed a bit fuzzily LLMs are scaled up versions of an architecture that was originally used for neural translation. Neural translation also has unverifiable behavior in the same sense.
I could interpret some of the data talk as talking about non local models but this very much seems like a more general criticism of LLMs as a whole when talking about Firefox features. Moreover, some of the critiques like verifiability of outputs and unlimited scope still don't make sense in this context. Browser LLM features except for explicitly AI browsers like Comet have so far had some scoping to their behavior, either in very narrow scopes like translation or summarization. The broadest scope I can think of is the side panels that show up which allow you to ask about a web page with context. Even then, I do not see what is inherently problematic about such scoping since the output behavior is confined to the side panel.
On Windows Mozilla can't even handle disabling hardware acceleration, a.k.a. the GPU, from its settings page. Sure you can toggle the button but it doesn't work as verified in the task manager. What hope is there that they can be trusted to disable AI then? It's a feature that I'd never want enabled. When that "feature" comes out users will be forced to find a fork without the feature.
Did Firefox already add AI into Tabs? Today I just got my first 'Tab Grouping' and it says "Nightly uses AI to read your Open Tabs". That's the worst way to do grouping ever... just group hierarchically based on where it opened from...
How do you disable the telemetry in Waterfox? It looks like they get their funding because they partnered with an Ad company. Do I just need to change the default search?
I was a FF driver for ages and now making the switch to Chrome based browser simple because it's faster and websites are all tested against Chrome / Safari. I see both of these issues manifest IRL on a weekly basis. Why do I want to burn up CPU cycles and second using FF when Chromium is literally faster.
With this, people will come here and the go. I mean consider the example of many GNU/Linux users I know who use GNU/Linux (or for them Linux means Ubuntu) system and can ask them to try out Waterfox. But, about installation - can't we have .deb? I know we can easily install from tarball and then setup the .desktop file and then adjust the icon to properly display, and what not...But, Can we make a bit simpler to try?
> Waterfox won't include them. The browser's job is to serve you, not think for you... Waterfox will not include LLMs. Full stop. At least and most definitely not in their current form or for the foreseeable future.
> If AI browsers dominate and then falter, if users discover they want something simpler and more trustworthy, Waterfox will still be here, marching patiently along.
This is basically their train of thought: provide something different for people who truly need it. There's nothing to criticize about.
However, let's don't forget that other browsers can remove/disable AI features just as fast as they add them. If Waterfox wants to be *more than just an alternative* (a.k.a. be a competitor), they needs discover what people actually need and optimize heavily on that. But this is hard to do because people don't show their true motives.
Maybe one day, it turned out that people do just want an AI that "think for them". That would be awkward, to say the least.
how is adding ai chat different than asking search engine? I think mozilla wants to make sure that it gets some cut for sending queries to ai similar to their existing revenue model where they get cut for sending it to google. Similar to SE's users should have a choice to use any ai or not.
if kagi can make a search engine that charges users, why dont we have a 1$/month open source browser whose code can be verified but people pay to use monthly?
When they say "AI browsers are proliferating." and "Their lunch is being eaten by AI browsers." what does that mean? What's an "AI Browser", and are they really gaining significant market share? For what?
I found this (1) that suggests that several "AI Browsers" exist, which is "proliferating" in a sense.
46 comments
[ 0.28 ms ] story [ 69.5 ms ] threadMozilla appoints new CEO Anthony Enzor-Demeo
https://news.ycombinator.com/item?id=46288491
[Update]: as I posted below, sample use cases would include translation, article summarization, asking questions from a long wiki page... and maybe with some agents built-in as well: parallelizing a form filling/ecom task, having the agent transcribe/translate an audio/video in real time, etc
I don't. And the whole idea of Firefox's marketing is that it won't force things on me. Ofc course om frustrated. My core browser should serve pages and manage said pages. Anything else should be an option.
I'm beyond tired of being told my preferences, especially by people with incentives to extract money out of me.
This really weakens the point of the post. It strikes me as a: we just don't like those AIs. Bergamot's model's behavior is no more or less auditable or a black box than an LLM's behavior. If you really want to go dig into a Llama 7B model, you definitely can. Even Bergamot's underlying model has an option to be transformer-based: https://marian-nmt.github.io/docs/
The premise of non-corporate AI is respectable but I don't understand the hate for LLMs. Local inference is laudable, but being close-minded about solutions is not interesting.
Firefox could have an entire section dedicated to torturing digital puppies built into the platform and... Ok, well, that's too far, but they could have a costco warehouse full of AI crap and I wouldn't mind at all as long as it was off by default and preferably not even downloaded to the system unless I went in and chose to download it.
I know respecting user preference doesn't line their pockets but neither does chasing users down and shoving services they never asked for and explicitly do not want into their faces.
Am I being overly critical here or is this kind of a silly position to have right after talking about how neural machine translation is okay? Many of Firefox's LLM features like summarization afaik are powered by local models (hell even Chrome has local model options). It's weird to say neural translation is not a black box but LLMs are somehow black boxes that we cannot hope to understand what they do with the data, especially when viewed a bit fuzzily LLMs are scaled up versions of an architecture that was originally used for neural translation. Neural translation also has unverifiable behavior in the same sense.
I could interpret some of the data talk as talking about non local models but this very much seems like a more general criticism of LLMs as a whole when talking about Firefox features. Moreover, some of the critiques like verifiability of outputs and unlimited scope still don't make sense in this context. Browser LLM features except for explicitly AI browsers like Comet have so far had some scoping to their behavior, either in very narrow scopes like translation or summarization. The broadest scope I can think of is the side panels that show up which allow you to ask about a web page with context. Even then, I do not see what is inherently problematic about such scoping since the output behavior is confined to the side panel.
https://mozilla.github.io/policy-templates/#generativeai
https://mozilla.github.io/policy-templates/#preferences
https://searchfox.org/firefox-main/source/browser/app/profil...
https://searchfox.org/firefox-main/source/modules/libpref/in...
> Alphabet themselves reportedly see the writing on the wall, developing what appears to be a new browser separate from Chrome.
at this point it’s more so a sandbox runtime bordering an OS, but okay
The black box objection disqualifies Widevine.
> If AI browsers dominate and then falter, if users discover they want something simpler and more trustworthy, Waterfox will still be here, marching patiently along.
This is basically their train of thought: provide something different for people who truly need it. There's nothing to criticize about.
However, let's don't forget that other browsers can remove/disable AI features just as fast as they add them. If Waterfox wants to be *more than just an alternative* (a.k.a. be a competitor), they needs discover what people actually need and optimize heavily on that. But this is hard to do because people don't show their true motives.
Maybe one day, it turned out that people do just want an AI that "think for them". That would be awkward, to say the least.
When they say "AI browsers are proliferating." and "Their lunch is being eaten by AI browsers." what does that mean? What's an "AI Browser", and are they really gaining significant market share? For what?
I found this (1) that suggests that several "AI Browsers" exist, which is "proliferating" in a sense.
1) https://www.waterfox.com/blog/no-ai-here-response-to-mozilla...