Share your feedback on the AI services experiment in Nightly

[asafko]

Hi folks, 

In the next few days, we will start the Nightly experiment which provides easy access to AI services from the sidebar. This functionality is entirely optional, and it’s there to see if it’s a helpful addition to Firefox. It is not built into any core functionality and needs to be turned on by you to see it. 

If you want to try the experiment, activate it via Nightly Settings > Firefox Labs (please see full instructions here). 

We’d love to hear your feedback once you try out the feature, and we’re open to all your ideas and thoughts, whether it’s small tweaks to the current experience or big, creative suggestions that could boost your productivity and make accessing your favorite tools and services in Firefox even easier.

Thanks so much for helping us improve Firefox!

[midgy]

This is such a great comment! I'm really glad to see some additional sources on this topic.

[arazil]

This should be an optional extension (at best) instead of a core feature of Firefox.

In addition, Mozilla would do well to remember that privacy conscious users make up a majority of the Firefox power user demographic. That user demographic is very skeptical of the current generative AI hype cycle and are drawn to products (such as Firefox) that limit their exposure to the technology. That demographic also contains some of the project's greatest advocates and promoters. Adding generative AI technologies without proper due diligence and community consultation will drive that valuable demographic to competing products and will further harm Firefox perception and market share.

[arazil]

In addition to my previous comment, I would also like to add:

Adding proprietary service support to Firefox should be done through the Firefox extension store and maintained by the vendor. These services should never be a part of the basic Firefox experience. In addition, Mozilla should not be diverting their limited resources to supporting proprietary platforms without (at least) some form of financial compensation.

[ellie23]

please no christ maybe my favourite thing about firefox is that it’s the one place that isn’t trying to shove ai down my throat don’t take that away if you want to ruin the climate more just set some tyres on fire at hq or something

[ellie23]

literally made a goddamn account to tell you how much i don’t want ai in firefox it makes the user experience worse it scrapes data people didn’t consent to give and it wastes water and energy everybody loses. bad choice for a browser a lot of people (including me) use bc they feel it’s more ethical/transparent than the alternatives 

[BurdNerd]

Please do not. "AI" as it exists now is a serious problem and not good. Do not go the way of Chrome and Google, I beg of you.

[BurdNerd]

You can literally count how many new profiles were made to tell you "please no" to determine actual interest (or more accurately, DISinterest) there is in having AI in any version of Firefox.

[CEB98]

As a long time user I absolutely do not want this. I use Firefox to try to avoid this kind of thing. This kind of AI is a pointless fad that brings no added utility to anything. Please, I'm so tired of trying to avoid this **bleep**.

[Dandelion44]

This is 3rd reply that tells "No AI" within in just 4 minutes, It would be extreme funny if firefox still adds it despite the hate they got (and still getting) from community. I have over 1000 notifications sent for just this thread (Kudos, new posts on this thread, etc)

[Dreamweaver]

I use Firefox for privacy.  AI has a reputation for data scraping.  Your plans are clearly very unpopular.  If you continue despite all the opposition and decide to add AI to our computers against our will, I am going to have to assume that your intent is to scrape my data for your own purposes, and I will have to use a different browser for my own safety.

[Dreamweaver]

LLM chatbots are built by nonconsensually scraping data.  Mozilla is demonstrating that it is willing to nonconsensually scrape data and deliver that content to other people.  Mozilla can no longer be trusted. 

[-sb]

Hi folks! First, thank you for working on the closest thing to a "good" browser that is currently available.

I would like to ask you - to beg even - to please drop all the LLM nonsense. LLM's do more harm than good. LLMs are built on stolen material without regard to the creators. LLMs are more confident in being wrong than a republican politician.

You know what we *do* want? A fast, stable, and SAFE browser. We want a browser that can detect and PROTECT us from LLMs and their dangerously incorrect content. We want something that will hold our tabs and not spy on us. That's it.

Can we please, PLEASE just have that?

Yours for the present,

Sean Boyer

[Asj]

Well said.

[cassanova]

I don't understand the hate for this since it's an entirely optional feature and not turned on by default. This sounds like the exact way to implement AI.

[Raventazmily]

What use case justifies the environmental impact? When large language models are inherently based off stolen data, that lies in its outputs unpredictably, when it is so much a turn off for artists and writers and people who actually know about the field?

[szakib]

Please don't ruin the last reasonable browser with this bull**bleep**.

[szakib]

It's lovely to know that my comments are censored.
Is "BS" okay to use?

[pacohope]

I'm sorry it's so passionately negative, but that's the nature of unasked-for AI features. It's well-documented that consumers are put off by AI features. (example1, example2) Balanced against the overwhelming negative sentiment, Mozilla has put forward some really tepid and uninteresting benefits and use cases:

• Summarize the excerpt and make it easier to scan and understand at a glance. It's pretty well established that LLMs don't do this. Studies have shown LLMs are bad at summarising. Shouldn't Mozilla make sure it actually does what they think it does before they put it in the browser and tell users that it does something?
• Simplify language. We find this feature handy for answering the typical kids’ “why” questions. That seems like an exceptionally weak motivation for spending extraordinarily scarce developer resources modifying the core browser. See above where it doesn't summarise accurately or correctly. The LLM is quite likely to make something incorrect while making it "simpler."
• Ask the chatbot to test your knowledge and memory of the excerpt. Seriously? I won't even dignify that with a counterargument.

I will continue to use Firefox. I still like it better than the alternatives. But when this feature makes it into an ESR version, I'm going to do whatever is required to disable it and I will never look back. I will never wonder whether that feature might have helped me.

[LueDelou124]

Bless you for providing the sources, i lost track of mine and searching proved inefficient. largely because of, you guessed it, generative ai results flooding the search engines!

[StarfighterVix]

I tried it, and within a minute it was giving me medical advice. I don't object to AI as much as many people do, but IT SHOULD NEVER BE RELIED ON FOR ANYTHING EVEN A LITTLE IMPORTANT. There WILL be people who ask it for advice on important things, there WILL be people who follow that advice, and that WILL lead to them making terrible mistakes.

Don't be part of it.

Thanks for making it opt-in and being open to feedback!

[LueDelou124]

precisely one of my concerns!! by putting this thing in, it gives the impression that it can be trusted, and a percentage of users will follow that trust and believe the ai when it starts giving bad information. this isn't just a bad idea, it is outright dangerous.

[oDaniel]

I feel like the side menu could use some improvement at the moment. I’ve been hiding and reopening it using Ctrl+H and then switching to the AI, but even if there were a hotkey specifically for this, it might not be the most effective solution. A  button on the toolbar could work better.

Also, while it may not make that big of a difference, I really don't know, it would be nice if the side menu was in a special container that prevents it from accessing any information in your browser that you prefer to keep private. Maybe this is already the case, and I’m just not aware.

Considering all of the user feedback so far, it might be worth exploring alternative implementations. It could be an add-on. While it doesn’t bother me personally, I understand that it’s a concern for many others.

I genuinely hope this can be implemented in a way that satisfies users and helps avoid negative discussions on platforms like Twitter and Reddit.

[oDaniel]

I also see it as an "easy access to AI services from the sidebar," as you said. There's the "Show queries on text selection" feature, which displays a shortcut for queries when you select text. Since Firefox sends the text, page title, and queries to the chatbot, it may be a little too much for a lot of people. I know that Pocket is disabled in a lot of Firefox hardenings, so you know how people feel.

I use AI for some small things, and having it in a more handy way is really cool. As I said before, if you can make it more secure—maybe if the Multi-Account Containers add-on had a special type of container for this kind of stuff and the add-on was also implemented with this feature—it could be really nice.

Maybe you could make the "Show queries on text selection" feature more suitable for this private browser. A little bit. At the end of the day, you're just making it easier to access. People will go to the websites if they want; you're just making these people's lives more convenient.

Bing's Copilot could be useful, but I don't know if you can implement it.

It will really be what people would be mad about, but if you could easily send the page as a PDF to the AI, it could be useful. Like, you already have the "Ask me" shortcut there. So, if you could send an article easily to the AI to have a quick exam, it would be nice.

TTS AI can also be interesting. It's a more fluid voice. Again, it would be up to the user's choice to have this. Sell some privacy for convenience. As long as you don't force it, it can be really great.

I think that's it. Have a great day!

[patchworker]

if firefox users want to be lied to i'll do it without wasting a bottle's worth of water per query.

[NeonC]

I wouldn't mind being lied to. Why does the sun get replaced by the moon at night?

[Lithish]

When I learned Mozilla embedded AI into Firefox, I was FURIOUS. I'm STILL furious. I've been a Firefox user since 2011, but if Mozilla doesn't remove the LLM slop from Firefox, I WILL find an alternative, and so will everyone I know. The steep environmental and ethical costs associated with so-called "AI" are in no way shape or form justified by the programs' underwhelming utility, and I'm disheartened to see Firefox jumping on this overhyped, underbaked, morally repugnant bandwagon.

In short: GET RID OF IT.

[rolingmetal]

You could consider switching to Vivaldi. They have taken a stance against using "ai" in the browser.

It's Chromium based, and the Chromium browser dominance can be seen as an issue.

[whybecauseiwill]

Please do not implement this. It goes against everything Firefox should stand for. There is no way using mass scraping of data promotes privacy. Generative ai has no ethical form nor practical use as it is full of stolen work and presents incorrect information as fact. Please please please do not add ai in any form even if we can opt out it is terrible all around. Don't fall for the trend of unethical tech be better! 

[rolingmetal]

Chatbots are horrible, and cause high blood pressure.

But an LLM, nor calling it "ai", summarizing search results can be usefull.

[brytning]

I've been a dedicated Firefox user for over a decade. Not interested in any AI features!

[rolingmetal]

Consider switching to Vivaldi. They are against integrating "ai" in their browser.

[kabba]

Generative AI is built on the ongoing theft of creators’ art and writing without their knowledge, consent, or compensation. I am appalled that Firefox, a so-called ‘ethical’ browser, is considering amplifying AI and its harms, and becoming complicit in it.

[kabba]

Beyond the ethical angle, AI uses catastrophic amounts of water and energy - many times greater than a simple search. Amid an escalating environmental crisis, promoting and facilitating the usage of AI is doubly unethical.

And, as so many others have mentioned, the constant integration of AI tools into every. single. app with no off button or opt-out reeks of predatory business practice - trying to artificially inflate demand for a product no-one wants, before everyone realises it has extremely limited utility. And it’s annoying beyond belief! I love Firefox because it allows me to browse in peace, to curate my internet experience, and to avoid the predatory ads and click-mes that Chrome and other mainstream browsers have en**bleep**tified into. Why are you trying to ruin that??

[belle]

Hard pass, the entire reason I'm on Firefox is because it doesn't do **bleep** like this.

[hadtoregister]

Yeah, this implies that something fundamental has changed. No service stays trustworthy forever, so if the other shoe is dropping... welp, it was a good run.

[TieflingKisser]

Absolutely NOT! The people who are drawn to using Firefox are largely the same people who are looking to get away from this stuff in other browsers. Generative "AI" is based on intellectual theft and data harvesting, exactly why I moved away from Chrome in the first place! I would hate to have to find another browser but I refuse to deal with that garbage.

[TieflingKisser]

Frankly, given how overwhelming the vitriol against AI has been for months, I find it very worrying that Firefox would even consider adding such a thing.

[Skribbly]

I notice many hate comments. What they don't understand is that the unethical part of the AI isn't the AI itself, but the companies that host it. That's why I prefer to use local AI.

Can support be added for Ollama? That currently seems to be the easiest local LLM service to set up, so I'm sure many users would appreciate it.

Even if you end up integrating your own version of Ollama into the browser (another GUI has done this), I would prefer this to be optional, similar to the optional local language translation. I'd like to use the Ollama instance I already have set up.

I'm alright right now, as the Page Assist browser extension handles this behaviour, but it would be nice to have this built in to the browser for more streamlined design and optimization.

EDIT:

I agree with others that this should be entirely optional, knowing that the default would likely be one of the unethical company hosted LLMs.

I think that this should be an opt-in type thing, meaning that it is off by default.

[Skribbly]

Given the majority of people mistrusting this type of service, this might not be fitting enough to the target audience of Firefox.

I think it would be a good idea to keep it isolated to a browser extension. This would minimize dislike.

And for those who use Firefox for it's security/privacy, you might prefer the modifications Librewolf has. It just optimizes the default settings for security and privacy, along with adding easy access to some security options in about:config. It also seems to include Firefox Labs options.

[Skribbly]

They marked the original post with 'JustNo', and the negative response rate is very high.

Given this information, I think they will likely listen to the almost unanimous agreement from the community, and exclude this feature.