I think Mozilla's position seems perfectly reasonable. AI options for those who want them and not for those who don't. Seems fine to me. I don't have to use features I don't want. I already don't...
I think Mozilla's position seems perfectly reasonable. AI options for those who want them and not for those who don't. Seems fine to me. I don't have to use features I don't want. I already don't use lots of features of Firefox.
I am not particularly interested in AI in my browser but I know plenty of people who use it regularly for various things and would like some browser integration. Why should Firefox only cater to me and not them?
I skimmed some comments (here), and I think the common sentiment is that those people (commenting) are upset that Firefox is spending time, energy, money on AI-related anything, when they could be...
I skimmed some comments (here), and I think the common sentiment is that those people (commenting) are upset that Firefox is spending time, energy, money on AI-related anything, when they could be working on other issues or features that are more important to them. Also: the assumption among them is that most Firefox users don't want AI, and so, in adding AI-related stuff, they might be repelling and squandering what little browser market share they have.
What a lot of people in the forum thread seemed to think is that Mozilla developing this feature is taking up human time and effort that could be better spent on making Firefox better. That...
What a lot of people in the forum thread seemed to think is that Mozilla developing this feature is taking up human time and effort that could be better spent on making Firefox better. That they're trend-chasing instead of building in better privacy features, for example, that would differentiate them from all the Chromium browsers. And I think I agree; if only a small percentage of Firefox users would be interested, surely they already have their own preferred tools they use instead of whatever gets shoehorned into Firefox?
I could say the same for any of the features J Random Commenter would prefer Mozilla spent time on though. Without access to Mozilla's metrics and user testing results it's impossible to guess at...
if only a small percentage of Firefox users would be interested, surely they already have their own preferred tools they use instead of whatever gets shoehorned into Firefox?
I could say the same for any of the features J Random Commenter would prefer Mozilla spent time on though. Without access to Mozilla's metrics and user testing results it's impossible to guess at what "Firefox users" want. There are around 90 million people in that group, after all.
Take me, for example - I've been using Firefox from before it was even called Firefox - I don't care about AI, I also don't really care about privacy either. I want a fast, stable browser with good support for modern web standards and decent cross-device sync. Firefox already does that well enough for me, so I'm not bothered about them spending time on things I'm not currently interested in. If those other features help bring more people over to Firefox, great!
There's plenty of legitimate uses of AI in a browser, even if they're not for everybody. Visually impaired users can query their browser to better understand the current page context, and have it...
There's plenty of legitimate uses of AI in a browser, even if they're not for everybody.
Visually impaired users can query their browser to better understand the current page context, and have it describe images without alt tags.
The local translation feature is a more private version of Google Translate, and a great resource for non-English users or those engaging with foreign texts.
Even some of the power features like automatic tab grouping will be useful to a certain percentage of users who feel lost in a sea of tabs.
Mozilla has been careful to make these features opt-in and targeted, and I think they're doing a great job. Much of the response feels reactive, and not considerate to users other than ourselves.
All your examples have one thing in common: They are not generative GPT/LLMs. I hate that OpenAI, Anthropic etc basically get to stand next to the more useful applications of the technology and...
All your examples have one thing in common: They are not generative GPT/LLMs.
I hate that OpenAI, Anthropic etc basically get to stand next to the more useful applications of the technology and say „Yeah, I‘m with those guys.“
I would think that all examples were generative AI. Fx runs a local NMT model for translation, which may not be a literal transformer but works on the same premise of token generation. Tab...
I would think that all examples were generative AI. Fx runs a local NMT model for translation, which may not be a literal transformer but works on the same premise of token generation. Tab grouping is similarly a tiny model they distilled to focus on smaller, more categorical training. Even text descriptions are generative because they produce tokens via their CLIP training.
Essentially, they all work on a core premise of training a model via backpropagation, then running input data through to generate tokens containing relevant context and understanding.
Damn, yes. Somehow I was thinking of only summarization, where other algorithms like e.g. TextRank are less resource intensive. So „2.5 of the examples“ … sigh.
Damn, yes. Somehow I was thinking of only summarization, where other algorithms like e.g. TextRank are less resource intensive.
Translation could also be an LLM problem. It's not the fastest or lightest solution, but it does produce much better results. See Kagi Translate vs Google Translate for example.
Translation could also be an LLM problem. It's not the fastest or lightest solution, but it does produce much better results. See Kagi Translate vs Google Translate for example.
I certainly do not. However, since Firefox pulled that stunt with selling user data ( they back pedaled ) I have since been using LibreWolf, a privacy-centric fork of Firefox.
I certainly do not.
However, since Firefox pulled that stunt with selling user data ( they back pedaled ) I have since been using LibreWolf, a privacy-centric fork of Firefox.
I'm having a hard time imagining what this 'AI window' is supposed to do. I think the existing quick access to LLMs from the sidebar and context menus makes sense even if I don't use it, but what...
I'm having a hard time imagining what this 'AI window' is supposed to do. I think the existing quick access to LLMs from the sidebar and context menus makes sense even if I don't use it, but what more could you possibly want that would warrant... that? What does it do? Is it functionally a chatbot PWA? Does it... generate the web pages for you? I don't get it.
I think Mozilla's position seems perfectly reasonable. AI options for those who want them and not for those who don't. Seems fine to me. I don't have to use features I don't want. I already don't use lots of features of Firefox.
I am not particularly interested in AI in my browser but I know plenty of people who use it regularly for various things and would like some browser integration. Why should Firefox only cater to me and not them?
I skimmed some comments (here), and I think the common sentiment is that those people (commenting) are upset that Firefox is spending time, energy, money on AI-related anything, when they could be working on other issues or features that are more important to them. Also: the assumption among them is that most Firefox users don't want AI, and so, in adding AI-related stuff, they might be repelling and squandering what little browser market share they have.
What a lot of people in the forum thread seemed to think is that Mozilla developing this feature is taking up human time and effort that could be better spent on making Firefox better. That they're trend-chasing instead of building in better privacy features, for example, that would differentiate them from all the Chromium browsers. And I think I agree; if only a small percentage of Firefox users would be interested, surely they already have their own preferred tools they use instead of whatever gets shoehorned into Firefox?
I could say the same for any of the features J Random Commenter would prefer Mozilla spent time on though. Without access to Mozilla's metrics and user testing results it's impossible to guess at what "Firefox users" want. There are around 90 million people in that group, after all.
Take me, for example - I've been using Firefox from before it was even called Firefox - I don't care about AI, I also don't really care about privacy either. I want a fast, stable browser with good support for modern web standards and decent cross-device sync. Firefox already does that well enough for me, so I'm not bothered about them spending time on things I'm not currently interested in. If those other features help bring more people over to Firefox, great!
There's plenty of legitimate uses of AI in a browser, even if they're not for everybody.
Mozilla has been careful to make these features opt-in and targeted, and I think they're doing a great job. Much of the response feels reactive, and not considerate to users other than ourselves.
All your examples have one thing in common: They are not generative GPT/LLMs.
I hate that OpenAI, Anthropic etc basically get to stand next to the more useful applications of the technology and say „Yeah, I‘m with those guys.“
Is the first example not an LLM-based example?
I would think that all examples were generative AI. Fx runs a local NMT model for translation, which may not be a literal transformer but works on the same premise of token generation. Tab grouping is similarly a tiny model they distilled to focus on smaller, more categorical training. Even text descriptions are generative because they produce tokens via their CLIP training.
Essentially, they all work on a core premise of training a model via backpropagation, then running input data through to generate tokens containing relevant context and understanding.
Damn, yes. Somehow I was thinking of only summarization, where other algorithms like e.g. TextRank are less resource intensive.
So „2.5 of the examples“ … sigh.
Translation could also be an LLM problem. It's not the fastest or lightest solution, but it does produce much better results. See Kagi Translate vs Google Translate for example.
I certainly do not.
However, since Firefox pulled that stunt with selling user data ( they back pedaled ) I have since been using LibreWolf, a privacy-centric fork of Firefox.
If you're referring to the terms and conditions update earlier this year, Mozilla never sold user data, and they were never looking to.
I'm having a hard time imagining what this 'AI window' is supposed to do. I think the existing quick access to LLMs from the sidebar and context menus makes sense even if I don't use it, but what more could you possibly want that would warrant... that? What does it do? Is it functionally a chatbot PWA? Does it... generate the web pages for you? I don't get it.
Their CEO disagrees. He wants AI to be shoved up your ass so hard.