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!

[haingdc]

Yes. Chatbot sidebar is narrow. I also think automatically get wider feature is an improve.

[phonics8226]

BUG REPORT:

When a custom model is added and then switched using the dropdown menu at the top of the sidebar, it becomes impossible to revert back to the custom model, even after closing and reopening the sidebar.

Additionally, is it possible to have multiple custom providers? I often compare and utilize different models for various tasks to achieve better results. Maybe with "browser.ml.chat.provider.1", "browser.ml.chat.provider.2", etc.

[Mardak]

That's an interesting idea to support multiple custom providers for easy switching. Are these all local providers or maybe we can add these to the list?

[phonics8226]

Well, one is duck.ai as you can see on the screenshots, and judging by another comment you made, there are no plans to include it in the list.

Anyway, any word on the bug? Or can we expect support for multiple custom providers in some future Nightly release?

[WoefKat]

This feature looks really nice but I would really love to be able to use my own ollama server (either through the ollama API or through the ChatGPT API that my OpenWebUI frontend offers).

It'll be pretty easy to implement as the API is the same as ChatGPT's but just with a different endpoint.

[Mardak]

The current implementation is rendering a webpage in the sidebar, so exposing ollama with open-webui or other web chatbots interfaces works for now. We have been looking into directly calling inference APIs such as chat/completions llamafile and ollama already expose, and this would allow for Firefox to build its own custom response interface not necessarily a chatbot.

[WoefKat]

Thanks! I already use openwebui so this worked well! One thing I did notice is that the webpage is not provided to the chatbot unless I select text first. In that case only the title is provided, not the URL (the URL would be handy so I could use a tool for the chatbot to receive it itself into its context window).

For example, if I don't select anything it would be nice if the whole page text would be provided to the AI, or at least a URL so it can retrieve it though an automation.

[shatteredsword]

honestly mozilla is lucky that firefox doesn't really have any competition when it comes to privacy-based browsers. This would have been the end of it otherwise, and I'm not sure that they realize it.

[haingdc]

Thanks. But don't waste time for these comment like this

[Tatra]

Thanks. But don't waste time for these comment like this.

[madcaker]

Kindly keep this garbage out of my browser please. I think you're forgetting who your primary user-base is. Hint: it's not the AI techbros

[llliiimmmeee]

There is a reason it is opt-in, I swear to goodness... how hard is it to understand!

[MageBunkshelf]

Opt-in would be if this were an extension. I do not opt-in to having an AI "feature" being added to my browser.

[AbandonedRocket]

I echo all the other sentiments here that embracing this is bad form from the folks who make the only browser that I like.

I made my profile here just to post this. I do not want this to be in Firefox, plain and simple.

[AbandonedRocket]

Also, if the money people are watching this thread, I've been considering Mozilla VPN now that I have the means to pay for such a service.

I'm going to consider other options for the time being.

[WhatzitTuuyah]

Check out mullvad! Doesn't get much more private than them.

[gsuberland]

Let me put this in as uncertain terms as I can: if you put LLMs in Firefox, I will find another browser. It's an ethical nightmare that produces noxious information-shaped sludge while burning a hole in the planet, and I want no part in it.

You need to get your priorities sorted. Stop chasing fads - especially fads that consume absurd amounts of energy. Focus on reducing Firefox's energy consumption, not increasing it. Focus on core browser performance and stability.

[winterwarburton]

Good luck with finding an alternative. I can't think of any heavy hitters in the market that aren't already stuffed with it (Edge, Opera) or run by companies eager to stuff it in as soon as they can (Chrome).

[omlt]

We don't care. There's already forks of Firefox that have declared their intention to remove this **bleep**.

We, OVERWHELMINGLY, DON'T WANT THIS **bleep**.

[shom]

Hello, Firefox team! I'm a big fan of your work. What Firefox represents as far as protecting privacy on the web and upholding open web standards is critical. Local-only "AI"/Language models are a great way of bringing new technological advances and showing a path forward for embracing technology without surrendering privacy, and I applaud your work in that area.

However, integrating large language models that are proprietary and also have been built by violating rights and licenses and require user data to be shipped off device, does not conform to the stated mission of Mozilla around privacy and user rights. Perhaps Mozilla can consider offering some of this functionality as extensions, as opposed to integrating them into the core browser.

The battle of market share cannot be won by ceding the moral high ground.

[doodlemancy]

Better yet-- perhaps the people who actually want these features can simply ask their magical totally real superintelligence to build an extension for them, and save Mozilla the trouble. I hear AI is really good at programming. (the link is a PDF sorry LOL)

[SoliTheFox]

So can you not press anything and have no feature enabled on your browser. Dunno, seems really easy for me.

[aimsme]

NO PLEASE GOD NO. no one needs AI chat functionality! I don't need to be lied to by a machine guzzling huge amounts of resources to do a task it would take me 5mins tops in a good search engine to do myself. I'm so sick of companies seeing the environmental state of the planet and going 'hey you know what would be cool? adding a fun new feature that barely works and also adds a huge extra burden onto those environmental issues'

[aimsme]

also it does not matter what your AI partners have to say about privacy. I have yet to see a single AI company that actually genuinely gave a toss about privacy either of the product users, or the people online having their data scraped to use the tool (who are often one and the same group). really disappointed in this.

[doodlemancy]

This! This! This! "Um well this company says they protect your privacy" is the most baby brained stuff, like... you'd have to be born yesterday to believe that after what we've already seen, from AI and from the entire history of corporations' online behavior in general. And if you were born yesterday, you shouldn't be working for Mozilla, child labor is illegal.

[terriest]

I'm confused as to the purpose of this.  It doesn't appear to run locally, while being only fractionally better than my local models.  It makes firefox heavier, which removes the main reason I prefer it to Chrome or Edge (I'm mostly running FF on 4GB boxes).  So it doesn't add value, and chips away at your existing market niche.

This really seems like jumping off a bridge because everyone else is doing it.  I'd prefer you focus on being a browser.  For me, this and the advertisement thing in version 128 have really soured me to your brand.  I do find I'm drifting back to Edge as you're losing your market differentiation.

[terriest]

Stuff it, I'm going full indie and trying Lynx and will optimize my page for it.

[terriest]

Wow that was a menace to install and does not like window scaling.  Does display my most used pages surprisingly well though.

I should add, I'm the one spreading FF within my organization, and I'm nominally an 'AI' researcher.  I think your decision to implement LLMs is a bad one, instead you should be focusing on differentiation in UX and maintaining a light core.

[StefiStarlite]

It's disappointing to see this feature made it out of the Nightly phase.

At best, this should have remained an optional extension NOT bundled with the browser at all. At worst, this is another action that results in myself and many other Firefox users to continue to lose faith in Mozilla's ability to actually make good on its promise to focus on privacy. It not only tells users that Mozilla only seeks to appease those falling for the inflated benefits of such features (regardless of the proven track record of copyright disregard and accuracy) but it is massively environmentally irresponsible given the unreasonably high energy usage LLM systems require.

I was really hoping the Nightly experiment would be just that, an experiment. Once I saw that these features were added to the production release, I immediately stopped using the mobile app and fully intend to remove the desktop app in favor of a different browser solution.

[Qhaz]

If you find another good browser (in the event that Firefox continues this path of self-destruction), do share!

[Asj]

which version of Firefox has it so I know not to update to it?

[essia]

I'm going to break the unanimity here. Most concerns on AI privacy are valid, but don't take into account (or maybe don't trust) the Mozilla AI initiative, based on Mozilla's values for privacy. AI doesn't have to be doubtful, misleading, untrustworthy. It can be trained on reliable and honestly acquired data, and ran locally with llamafiles. I disagree with the use of big tech models in the browser, but I think Mozilla AI can come up with great ideas, all based on Mozilla's core values.

[SkipfordJ]

While this may be true at the current time, there are no safeguards in place for this to be kept ethical in the long term. How many changes in the managerial chain in a single engineering team would it take for that to collapse? And with all the data gathered with the trust of Mozilla's core values? It's a precarious thing.

The only way to be assured, in this case, that this trust isn't breached is to not use it.

[cyan1]

AI is also wildly inefficient using huge amounts of water (for cooling) and electricity. I also don't trust Mozilla, or any AI program itself. There are many examples of AI programs doing something they were not supposed to do or going against company policy. So even IF Mozilla doesn't want their AI to violate privacy it might do it anyway.

Finally just because an AI's training data is accurate, that doesn't mean when the AI chews it up and spits it back out, it will remain accurate. Say I write an article debunking a conspiracy, and explain the contents of the conspiracy, the ai (not capable of actual thought or reading comprehension) might regurgitate that conspiracy theory to someone and not the surrounding debunk.

[comradecrow]

I would much rather this be an official extension, so I would not have to do the work of removing it myself, or switching browsers again.

[soraka]

even if it became an official extension, i would uninstall firefox.

[jahtnamas]

actually the extension route would be funny because they would see the low install rates, realize it was a huge waste of time and take it down, head in their hands in abject defeat

[CyberneticFlesh]

Stop wasting time implementing useless features detrimental to the environment. Hard no on this one.

[FriedPipis]

The environmental impact is exaggerated.
https://blog.giovanh.com/blog/2024/08/18/is-ai-eating-all-the-energy-part-1-of-2/

[tuxayo]

That article doesn't seems to go in that direction.

> For the most part, the energy issues are caused by the AI “arms race” and how irresponsibly corporations are pushing their AI products on the market. Even with operational efficiency ruled out as a cause, AI is causing two killer energy problems: waste and externalities.

> It’s simply not true that the technology behind AI is particularly energy-intensive. The technology isn’t insatiable, the corporations deploying it are. The thing with an insatiable appetite for growth at all cost is unregulated capitalism.

We are here in the arms race and irresponsible push to AI for everything. The environmental impact is acknowledged.
We can send that article to someone blaming someone else for using an LLM (like chatgpt) but it's doesn't look relevant in the context of mozilla.