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!

[Haagee]

I wrote a five paragraph comment, spelling out my concerns that the advocates of this feature are entirely missing the point on why so many users are deeply concerned about its implementation like this.

To which you apparently responded Just don't use it? Thank you for trying to ease my concerns - BUT THAT'S NOT HELPING.

[Justificand]

If chatbots or something of that stain of "AI" gets forced onto me with Firefox - there are always other browsers.

[DerelictDrone]

But those aren't machine learning or an LLM, they don't require an internet connection, nor do they burn your cpu and gpu alive to perform.

Any reputable use of what we called "AI" before has moved on to more specific terminology to distance itself from both the hype train of LLMS, or image generation, and their negative press, we can live with just ranking pages based on how many times you visited them to predict which page they are most likely to visit, but users are drawing the line at automatically including LLMs, that either consume significant resources to produce the barest output or require a cloud connection to perform.

Please try and remember that you're making a web browser, anything past "browsing the web" should be an extension.

[BlorgBlorgBlorg]

You know exactly what the difference between these algorithms and the LLM integration you're working on is. It would be more honest to either not respond to AI critics at all, or directly tell them you (and Mozilla) don't care about their concerns.

[Kwonunn]

You know full well that those are not the AI stuff we are concerned about. By including and promoting products from these wildly unethical AI companies Mozilla is indirectly worsening the whole thing. Please please please remove this.

[Kombee]

The fact that you are comparing LLM-driven features to non-LLM powered features such as autofill and completion suggestions shows that either A) you have no frikin clue what you're talking about. But I'm going to assume you're not an idiot and you do in fact know the difference between how autofills and suggestions work vs how "AI"/LLMs work, which leaves me with B) you're choosing to be incredibly dishonest with the user base by making comments such as this in an attempt to leverage our assumed stupidity (news flash, most of us aren't idiots and even we know the difference here) to convince us this feature is the same as existing features in implementation and protocol. It is not. You know. I know this. We know this. Stop being dishonest with us because you've already convinced yourselves this will be added and now you're doing the same thing every techbro is and play dumb to sell the userbase on a feature they didn't ask for.

Focus on more programmatic and direct engineering solutions to whatever you think "AI" will accomplish instead of using AI as a shortcut for new features.

[NosirrahSec]

We don't want this garbage at all. That is the vast majority of users' opinion.

[NosirrahSec]

No one wants this garbage.

No one cares what moronic opinion you have on it.

[aintnoway]

You're being intentionally obtuse. You know those aren't the same things.

[omlt]

You're conflating actual automation of tasks and generative AI (the plagiarism machine) and therefore providing cover to techbros and VC investors that are trying to rebrand all (useful) automation features as AI.

An algorithm IS NOT AI. And you're weaseling words here.

YOU SHOULD BE ASHAMED OF YOURSELF.

MOZILLA WASN'T FOUNDED FOR THIS, THIS IS A DISGRACE TO THE VERY CORE MISSION AND STATEMENT OF MOZILLA AND FIREFOX.

[cgranade]

I'm old enough to remember when that was called "machine learning," and wasn't doing the branding work for giant companies that do not have the Internet's best interests at heart.

[penryu]

Address bar completion / tab prediction doesn't use "AI". It uses simple substring autocompletion and tab metadata heuristics that have been around for many years, without the ethical and environmental concerns now in play wrt LLMs.

And if you specifically mean the "fuzzy" tab matching/completion that LLMs can provide, then you've found the exact features of AI that have made Google's AI Assistant a laughing stock — the least useful, least relevant, and least wanted features, that you SHOULD want to keep far away from such a fundamentally important project as Firefox.

As for translation... IFF LLMs were used in page content translation, and IFF such use were actually "privacy-preserving", then that seems an _ideal_ opportunity for an optional extension rather than integrated LLM nonsense.

This is undoubtedly a thankless job, even if you are a proponent of the feature. But your employer has solicited feedback on this "feature", so it seems strange that so many employee responses seem to be trying to convince opponents of the feature that their feedback is wrong, rather than simply collecting, considering, and acting on it.

[dtw_]

This is what happens when all algorithms are suddenly called "AI"...

[Saeldis]

How.... "delightfully" disingenuous.

I was going to expound on that, but honestly it's so disingenuous this is all it deserves. You should be deeply ashamed of yourself.

[dggozo]

History suggestions, is that all you got? We need AI for that?

[JHeuser]

We're not stupid. There's a clear difference between translations based on machine learning and prediction algorithms, and LLMs. We want the LLMs out. Take them out.

[doodlemancy]

Y'know, when you try to rules-lawyer about any kind of algorithm/machine learning being "artificial intelligence" and try to re-brand stuff that already existed with the buzzword of the moment... in a roundabout way, you kinda help undermine the very illusion the AI crowd is trying to perpetuate.

[IBBoard]

"AI" is just the latest marketing umbrella, and it's dishonest to try and excuse the bad stuff (GenAI and LLMs) by pointing to the good stuff (data analysis, machine learning models, or simple "ranking by similarity and visit count" metrics most of what the address bar does).

If someone wants to use GenAI and LLMs then they can choose to use an extension. Nothing about an extension increases or decreases the options on privacy-preserving AI. You then also preserve the privacy of more people by not having it in there for the majority who have zero use, and for whom GenAI is a net negative to society.

[No3]

Kindly stop putting your fingers in your ears because you got told 'No' and you don't like being told 'No'. 

(See Haagee's comment for a clear and concise breakdown about why your replies are bs so far though I doubt you'll listen to that either. Banging your own 'we are the best we know best we could never be wrong' cymbals too loudly to hear anything else.)

[LeafyGreens]

The benefit that users get from firefox is the absence of exactly this kind of garbage. You are making your product worse and removing the very reason why people use it.

People don't use firefox because they want to follow the latest trends. They use it because the competition is toxic and hostile. Firefox's raison d'etre is to provide a way for people to browse the web without having to bow to corporate overlords. If you start treating people like your competition does, firefox will cease to exist.

[dhasenan]

What we want to avoid is increasing usage of bullcrap generators, especially those built on plagiarism on an unprecedented scale. (We also want to avoid any features that might read more of our data than we've consented to, but it looks like this feature doesn't proactively send data to them without permission -- for now.)

A locally built Bayesian model to determine which URL I am likely to navigate to based on my first few keypresses in the address bar does not have those problems. Conflating them is disingenuous.

[penryu]

This is deliberately obtuse, not to mention misattributing common browser heuristics to "AI."

I'm sure you've long since tied your horse to the AI post, but please don't insult our intelligence with this sort of disingenuous posturing.

[DiogoConstantin]

I certainly don't want those "benefits" I don't want that type of feature on my browser.

How could this help privacy preserving AI alternatives, instead of all AI?

[a7h13f]

I'll be moving away from Firefox because of these changes, but thanks for being open and honest about your company's intent.

I originally switched to Firefox *because* it was less bloated, cleaner, and more privacy focused than the alternatives. I'm sad to see Firefox move away from the original design philosophy.

[omlt]

Respectfully (because you've clearly been given the **bleep**ty end of the stick with handling community response) :

NOBODY ASKED FOR THIS, NOBODY WANTS THIS.

F THE PLAGIARISM MACHINES

[GrislyBoar]

I have several questions:

1. What do you mean by "functionality?" What's the function? How does this make Firefox better?

2. Are you aware that the use of generative algorithms advertised as "AI" (I stress this because it is not true AI) make users find you and your product untrustworthy, and there are studies saying so?

3. The implementation of generative algorithms has been a failure across the board. What makes you think it won't be a failure here, too? Because as it turns out, tools that were designed for specific niche purposes, mostly relating to statistics, aren't particularly useful outside of that setting.

4. Why the insistence on it? Nobody's interested except shareholders. And even their interest is waning due to the lackluster results of these tools.

Being told repeatedly that something is the future of tech and repeating it to others does not magically make it so. These generative tools have been demonstrated over and over again that they can't perform any task that they're advertised as capable of, largely because said tasks were not what they were designed for. There are already simpler, less wasteful tools that do these jobs much better.

It baffles me that executives at *computer software companies* can display such egregious tech illiteracy all for the sake of hoping people will buy into what is essentially digital snake oil. Just because Steve Jobs had one good idea that eventually worked doesn't make everything the industry pushes the same. Yet so many are so infatuated with the idea of Steve Jobs, they're obsessed with emulating him. In my opinion, there are too many Steve Jobses in the tech industry.

What we need are more Steve Wozniaks. Not in the sense that he was a computer engineering genius, but in the fact that he was a genuinely nice, agreeable person. And he was also the one person who told Jobs "no." His insistence on putting eight expansion slots in the Apple II when Jobs didn't want them was what made Apple a success in the beginning. His tech literacy gave him authority that Jobs didn't.

Success in the tech sphere does not begin and end with successful marketing. It happens when the people who know what they're talking about, and know what the consumer truly wants, can tell executives "no" and make marketing stay in their lane.

[Empterdose]

Getting to more seamless integrations would likely be trickier as an extension

You know what's really cool? Mozilla, as the developers of the browser, can enhance the capabilities of extensions to make it possible to integrate things like this seamlessly!

Complaining that the platform you built is inadequate rings pretty hollow, particularly when the community has been telling you that over and over again since the switch to WebExtensions.

[VyriaDurav]

I want none of this. Absolutely none of this. All of it is antithetical to the principles that made me choose this browser over Chrome and it's myriad clones. The fact that not only is this something being proposed, but you have the audacity to talk about expanding it further before even checking to see if people are willing to overlook this horrendous action? Cease. I don't need any of these features and I certainly don't want to burn the world to the ground powering them. I can find my own websites, I don't need an environmental hazard and panopticon doing it for me.

[Sintanity]

An alogrithm that auto-completes a search result or suggests an open tab isn't the same as an LLM like ChatGPT and making this comparison either means you have no idea what you're talking about or is deliberately obtuse. Weird and gross behavior.

[dggozo]

those are bad features even without considering the cost of AI. I can name a tab group myself.

[MageBunkshelf]

---

@Mardak wrote:

This is currently a first integration of AI with plans for more optional functionality that don't require a sidebar interface, e.g., suggesting tabs that are relevant to your current activity or name for a tab group.

---

Jesus christ no

[AntBand]

it sounds awful like wtf

[simonjadis]

If Firefox were to ever "recommend" a tab to me based upon what I'm doing, I would use Firefox to find and download a new browser.

[malogan]

"minimal impact for those who don't want it"

That's cool. Since minimal isn't worth being concerned over, you guys have some big name financial supporters so if I stop my monthly donation to the mozilla foundation that should also be pretty minimal, yeah?

[Lenyberry]

"suggesting tabs that are relevant to your current activity or name for a tab group" sounds like a terrible, annoying feature that I actively do not want to ever be bothered by having to go figure out how to disable!

Let people use their own brains to decide what they want to do and how they want to name things, ffs.

[greenbird]

Honest question: is "what should I call this tab group" a problem you have so often that you need an outside source to help you?

[rarsneezes]

Sorry, I really don't want Firefox - and especially not any AI services - to be paying enough attention to my browsing history/tabs to suggest me new tabs. I don't care if the tabs are relavent or not - What happened to Firefox protecting my browsing data privacy?

[benlk]

If the roadmapped future integrations require access to browser APIs that aren't available to extensions, then Firefox should figure out how to make those extensions available to extensions in a safe way, rather than using this as an excuse to shoehorn unwanted AI API clients into Firefox. Leave the AI API client in an optional extension that users can install.

[verbedr]

Making it possible for all extension to benefit the same seamless integration would benefit everyone.  I would focus on that, instead of trying to integrate "an" AI into your product.

Consider this as a challenge, eat your own dogfood.

[AntBand]

we want NO impact. ZERO.