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!

[Nobody_of_note]

Wow, I got the notification for that so I know what was removed.  That's a bad look.

Will be cancelling my subscription and moving away from anything Mozilla is involved in.

[nanarko]

what was removed? I didn't get the notif so I don't know

[ForgedinStars]

Hey there, I wanted to say that this is a terrible idea. A lot of people use Fire Fox to get away from AI and this would leave Fire Fox users nowhere else to go, please reconsider this choice. Thank you for your time.

[ForgedinStars]

Not to mention, AI takes so much electricity and water from the evironment, which I know is something Mozilla would NOT stand for. As a company, you preach privacy and safety for users, by adding AI you take that away.

We may just be thousands of angry, normal, every day people compared to Mozilla as a company, but we're mad because you promised something and now you are ripping it from under us like Google and Microsoft did. You guys at Mozilla said you would protect our privacy and respect it here, and it doesn't seem like you are living up to that promise. Don't add AI to Fire Fox at all, it breaks the promise you made to keep users safe.

[VildaVedo]

I honestly can’t understand this thread. Why are people complaining about AI in Firefox? Who’s forcing you to use it? I like it because the browser lets you implement AI your own way. You can even use your local model. You can enable or disable it entirely, no one is forcing anyone to use it.

If you don’t like this feature, just leave it off. It’s like any other feature in a browser or software.

This is like complaining about an adblocker: "Hey, I don’t want this adblocker in my browser." Well, great! Who’s stopping you from uninstalling it or just deciding not to use it?

This thread is frustrating and completely out of line, it could be described as "I oppose the current thing" for no real reason.

[RainbowCrows]

Ad blockers don't rely on intellectual property theft and/or environmental degradation in order to function.

AI is guilty on both counts. AI results, furthermore, are also notoriously unreliable.

Your comment is frustrating and out of line, and supports AI garbage being implemented into Firefox for no real reason.

[artemismn]

PLEASE NO AI. I’m so sick of AI, please. I switched to firefox to AVOID google’s AI. 

[VildaVedo]

Well, I like it because you can even use your local models. In the end, if you don’t like AI, just keep the checkbox off. Isn’t that simple? Just don’t enable the features you don’t want to use.

[NightmareSwag]

It's not as easy as "just don't use it bro", this machine leeches off of pre-existing fuctions and we CAN'T turn it off nor can't we not contribute to the database.

[RainbowCrows]

Tell us you don't understand how AI works without telling us you don't understand how AI works...

[razeghi71]

I can't believe all this hatred, I love this feature. It helps me better understand what I'm reading on a web page using my existing ChatGPT subscription, which has honestly saved me so much time and money. Other extensions rely on an API key from OpenAI that costs extra money over the existing subscription. This has really made my workflow 100% faster. Thanks!

[RainbowCrows]

You can't believe all the hatred because you didn't bother reading people's reasonable concerns about AI, or because you can't understand what you read even with AI "help"?

[uid65534]

For real, the amount of people that fail to understand what this experiment is even for in this thread is insane.

FFS it is even opt-in. Nobody is making anyone use this feature for those that are panicing about their tinfoil hat opinions on the topic because they don't understand how these models are offered, including local LLMs. The whole conversation here has 'ban 5G' vibes lol.

[NightmareSwag]

Perhaps you should improve your reading comprehension instead of asking a machine to do that for you.

[tuxayo]

NO. The actual software running is not libre.
Even if the training algorithm would be libre. The training data isn't.
Even if the training data would be libre. It's impossible to train with reasonable hardware for most collectives of people. We basically invented a programming paradigm that is impossible to compile for almost anyone.
And most AI usage is impossible to host with reasonable resources.

What happened to Principle 7 of The Mozilla Manifesto? https://www.mozilla.org/en-US/about/manifesto/

Most AI usage can't have the concrete benefits of libre software: massively lowering the dependence on a vendor. Whether it's about disagreement on choices or availability on the long term.

Also Principle 6. As of now, most AI usage can only end of in more centralization.

There can be compromises on these principles if having a very very strong reason. Like for translations, automated subtitles, text to speech, because it's accessibility and there are immense benefits there.

[MikeCerm]

There are lots of free/libre LLMs and implementations like ollama. AI is going to happen, whether you want it or not, and if companies like Mozilla don't push for free and open-source options, then the only implementations that matter will be the proprietary ones.

As for the training data, I think there needs to be a reckoning. I'm kind of a fair-use absolutist, so I don't really where the data comes from, and I want it to come from everywhere. I believe that public libraries infringe on copyright, and are a public good because everyone deserves to have access to information. I don't believe that a writer like Stephen King is harmed if people read all of his stories and then attempt to write their own original stories based on the love of what they have read. I also don't see the difference in allowing a person to build a machine capable of reading those stories and attempting to generate new stories by incorporating new and original ideas, i.e., "prompts."

It used to require a skilled artisan to make clothing. Then looms were invented, along with other machines that make clothing so cheap that almost anyone can afford to have it. I think that's a good thing. I mean, it's not great that there is industrial waste that occurs in the process and fast fashion is not great, but I don't think we should return to a time when most people couldn't afford clothing.

AI is going to automate things that previously had to be done by hand, with massive time investments. That is all. It's still going to require skilled people to use it to create things of value. But the tools should be available to everyone, and Mozilla should help make them available as part of their mission for a free and open web.

[onionella]

In most countries public libraries pay royalties for authors per every loan. I do not understand how that could be copyright infringement...

You can be a fair-use absolutist if you want but that does not change the fact that the training data used in those "free/libre LLMs" you linked unconsensually contains copyrighted content and that by definition makes it not free and libre and open source.

[MikeCerm]

I don't know how it works in most countries, but in America you can lend a physical book for free. Ebooks are bound by different terms, but physical books have no such licensing terms, the "first-sale doctrine" exists, so there are no restrictions on what a person might do with a physical book, other than make unauthorized copies, and there are even fair-use exceptions to that, e.g., criticism, news reporting, teaching, education, scholarly research, etc.

With fair use, there are several factors that considered. In the case of these free and/or libre LLMs, their transformational nature, inherently non-commercial, and have their own academic value that outweighs the copyright concerns. The same may not be true of commercial LLMs. 

I also don't think that lot of "terms of services" licenses apply the way that copyright holders think they should. I've visited a lot of websites in my life, and with few exceptions (pay-walled websites), I've never agreed to any terms of service. Do websites publish their terms of service somewhere and claim "all rights reserved"? Maybe, but if you can get to the content without reading or agreeing those terms, then I don't see how they matter. When you broadcast your thoughts out in to the world, they're out there, and they're going to be consumed and used in ways that you cannot control.

[neutronstar87]

I don't know what's the problem with people here, they're unnecessarily skittish. It's an entirely optional feature, and it's in Nightly - it might never land. Anyways, I did try it out, and here's my review of it.

1. It makes double-clicking to select an entire sentence hard, as the "magic AI" hover button appears too close and gets clicked or somehow interferes in the process.
2. The options shouldn't expand upon merely hovering the "magic AI" button. You should need to click it.

For now, I've disabled the experiment again as it feels like it's cluttering the page. Other than that, I actually like the convenience. If people stop reacting impulsively and actually think critically, there's nothing inherently wrong with this feature. All it does is open your choice of LLM's homepage in a sidebar just as if it were a separate tab. There's no background worker or job spawning that periodically sends your webpage data or anything like that. It's functionally just a shortcut to opening a new tab and going to the page anyways.

[metalbyte]

That's true. It is completely optional. My concern primarily has to do with the resources Mozilla is investing into AI or LLM features. Firefox has real challenges and problems to overcome. Mozilla is not a wealthy organization. They have to be careful with how they invest their money and devote their money. Investing their time, money and energy into AI features.... I can't overstate how profoundly foolish that seems.

I manage a large enterprise that is abandoning Chrome, and I couldn't get our leaders or end user base to even consider Firefox because it is regarded as slow and buggy. So when I see they are investing money into AI as opposed to simply focusing on making Firefox an excellent browser, I have no kind words in me. Mozilla needs to wise up.

2024 is probably the first year in more than a decade where Google Chrome is poised to lose market share, and Firefox can't seem to attract new users because they'd rather chase fads than make a polished product.

[neutronstar87]

The feature as it stands absolutely would NOT require any significant amount of engineering resources. Unless they're planning to make some big AI moves - like having their own chatbot service or deeper integration with AI which indeed will be problematic, it wouldn't divert Mozilla's finances away from its main goals. And the mere presence of it shouldn't bother users. Everyone who's even a little bit concerned about privacy ends up hardening the Firefox browser anyways by going into about:config and disabling a bunch of stuff. You can just consider this another one of the Pocket-like features that you end up disabling.

[TJax]

It doesn't matter what this feature does people don't like it because it is built off of a tech fan that has a huge negative impact on our environment, that functions only by stealing whatever had been fed to it, that doesn't actually function the way most people want it to, and that many experts in the field are warning against using. On top of that it's just a waste of money and time that could be spent doing something that truly benefits users. Every company is scrambling right now to fit an AI feature somewhere it's a bad move for their consumers and their business.

[zy]

Not sure if it's worth it, you really seem to have made your choice, but let's do it anyway.

Would it be okay to add an optional feature that allows the user to easily access websites selling guns? I mean, this is optional and disabled by default, so the people that don't wanna see guns won't. I think you wouldn't do it, because guns probably don't fit in your values.

An other thing that is supposed to not fit in your values is disrespecting privacy. And guess what: generative AI does that. They train their AIs using data of people that weren't asked if they wanted to. This is atrociously against privacy. And you are associating yourself with that.

The fact I'm not the first one explaining that and the disrespect your have to us in your answer to Frisk make me think you understand the problem and just don't care? At this point, I don't think even removing this feature, fixing that other privacy option that is enabled by default and apologizing publicly would fix the trust you lost from us. But if privacy is still one of the values of Mozilla, that's what you would do anyway.

Hoping there will still be people fighting for our privacy like Mozilla used to do.

[Nobody_of_note]

Sorry to say you are wasting your time.  This is how Mozilla is monetizing going forward and they simply don't care what we think.

There are enough people who don't care or worse think AI in its current wasteful state is a great idea that they drown out those of us who know better.

I had a short conversation with some of their staff when I cancelled my donations after this announcement and they were clear it was just the way it was going to be.

[Manchester100]

So 'feedback' is somewhat of a misnomer.

You might be right. Reading what is said at the head of the thread more closely "...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" doesn't imply any reflection on backward steps even if they are to correct what users see as a misstep.
I hope we are wrong.

[Nobody_of_note]

I really do too, but I'm not waiting around to see if an organization pivoting to focus on money instead of principals is going to do the right thing.

History shows they never go back.

[Carusus]

It's not showing up when you click show in sidebar in settings

[Francesco17]

Go to settings, enable the feature in Firefox Lab and then you will find it in the sidebar by switching between bookmarks and chatbots

[Carusus]

How do I switch between bookmarks and chatbots? In Opera AI is imply a tab option

[Carusus]

OK. I found a solution. Open ChatGPT and then right click on the tab and click on Add to Essentials.

That keeps it in the tab bar, but somehow I suspect it's not supposed to work like that...

[duskord]

Can you also provide Claude?

[Mardak]

You can configure a custom provider to Anthropic Claude for easy access to chat while on another tab. Currently the context menu passing in a prompt doesn't work, but we'll work with them for better compatibility.

 

[feoh]

Thanks for the response! Would the custom provider feature allow me to use something like ollama with the ai chat sidebar?

[Mardak]

The latest Nightly 130 (20240712095045) includes Anthropic Claude in the list and supports passing in prompts with the context menu when logged in.

 

[wutongtaiwan]

I want kimi chat.This is chinese chat bot

[wutongtaiwan]

It would be even better if AI could be used for search, enabling an open-source AI model locally, and when searching, the AI aggregates the search results and answers them in the form of a human conversation, just like a chatbot

[Mardak]

Are you suggesting the chatbot to look at the search result page content or would you expect Firefox/chatbot visit some linked pages for you? There are AI services that do this like https://perplexity.ai that you could configure as a custom provider and also supports passing in your custom prompts.

[wutongtaiwan]

I want the chatbot to look at the content of the search results page, summarize it based on the content, and answer in human language

[Mardak]

Something like this?

[wutongtaiwan]

yes