The pathological need to find something to use LLMs for is so bizzare.
It’s like the opposite of classic ML, relatively tiny special purpose models trained for something critical, out of desperation, because it just can’t be done well conventionally.
But this:
AI-enhanced tab groups. Powered by a local AI model, these groups identify related tabs and suggest names for them. There is even a “Suggest more tabs for group” button that users can click to get recommendations.
Take out the word AI.
Enhanced tab groups. Powered by a local algorithm, these groups identify related tabs and suggest names for them. There is even a “Suggest more tabs for group” button that users can click to get recommendations.
If this feature took, say, a gigabyte of RAM and a bunch of CPU, it would be laughed out. But somehow it ships because it has the word AI in it? That makes no sense.
I am a massive local LLM advocate. I like “generative” ML, within reason and ethics. But this is just stupid.
When I’m browsing around with multiple tabs open, the last thing I want is something to start moving them around and messing my flow up. This is a solution looking for a problem.
The pathological need to find something to use LLMs for is so bizzare.
Venture capital dumped so much money into the tech without understanding the full scope of what it was capable of. Now they’re so in so deep that they desperately NEED to find something profitable it can do, otherwise they’ll lose the farm.
It makes a lot more sense when you realize that the Mozilla corporation is a for profit run by the same techno-fascist aggrandizing bait-and-switch narcissists as the rest of SV.
I’ve been saying it for years, but I will never donate to Firefox until it is freed from the shackles of a for profit corporation that can use your donation for any profit motive it sees fit; not even related to Firefox.
I don’t care. It’s a corrosive force that causes them to pay for over priced CEO’s and integrate services that nobody cares about into Firefox (like pocket) or that runs against their principles (container VPN’s exclusive to Mozillas for-profit VPN).
90% of their cash flow comes from google to be the default search engine - they are probably trying to open up alternative routes of funding to reduce the risk, since it’s not guaranteed that the money will keep coming due to the current lawsuit.
Like why should including a feature with “AI” in it get them VC money?
Spoken like someone who’s never interacted with Silicon Valley VCs… just imagine someone with tons of a money, a moderately competent business background, and very little understanding of even the basics of technology that you and I take for granted. And then make them stupid and greedy.
“AI? Yes please! Here’s some money, I’ve heard of Firefox so I know you’re good for it.” It’s not really any more complicated than that, I don’t think.
They could contribute to some existing local inference effort, do actually useful dev work, and slap their brand on it. It would both be cheaper and “look” better to VCs.
Yeah. There would be a way to do it that I feel like might potentially be useful. The described method (doing clustering instead of just having a similarity threshold to group tabs together, vectorizing the entire tab title through a whole fucking network instead of just tokenizing it and calling two tabs similar if they have uncommon tokens that are within a certain similarity level) really sounds to me like people who have no real idea what they’re doing, just being “ML experts” all over the codebase and fucking things up, and probably walking away very proud of themselves while helping themselves to bunches and bunches of the Mozilla Foundation’s Google-money.
It’s like the opposite of classic ML, relatively tiny special purpose models trained for something critical, out of desperation, because it just can’t be done well conventionally.
Here i disagree. ML is using high dimensional statistics. There exist many problems, which are by their nature problems of high dimensional statistics.
If you have for an example an engineering problem, it can make sense to use an ML approach, to find patterns in the relationship between input conditions and output results. Based on this patterns you have an idea, where you need to focus in the physical theory for understanding and optimizing it.
Another example for “generative AI” i have seen is creating models of hearts. So by feeding it the MRI scans of hundreds of real hearts, millions of models for probable heart shapes can be created and the interaction with medical equipment can be studied on them. This isn’t a “desperate” approach. It is a smart approach.
The recognition of the pattern is done by the machine learning. That is the core concept of machine learning.
For the interpretation you need to use your domain knowledge. Machine learning together with knowledge in the domain analyzed can be a very powerful combination.
Another example in research i have heard about recently, is detection of brain tumors before they occur. MRIs are analyzed of people who later developed brain tumors to see if patterns can be detected in the people who developed the tumors that are absent in the people who didn’t develop tumors. This knowledge of a correlation between certain patterns and later tumor development could help specialists to further their understanding of how tumors develop as they can analyze these specific patterns.
What we see with ChatGPT and other LLMs is kind of doing the opposite by detaching the algorithm from any specific knowledge. Subsequently the algorithm can make predictions on anything and they are worth nothing.
I like the tab groups. I use them often at work to group an issue with related tabs and my attempts at solving it. Also makes it easier to pause work on one problem and work on something else because I have the tabs grouper and know exactly where to go back.
And nobody should stop you installing an extension that provides tab groups. I agree with the other commentator that some features can be left to extensions and don’t need to be part of the core web browser, though.
True, but I’m not sure that an extension would have the necessary access to manipulate the browser like that. I don’t think it should. A malicious extension could do horrible things.
I’m not sure that an extension would have the necessary access to manipulate the browser like that.
I don’t know if they still do but they used to have. That, however, is something to discuss with the genius decision makers at Mozilla who decide to break extension APIs every couple of years. Firefox on Android still hasn’t recovered from last time.
You probably look at tabs as something inherently transient. In my tab group powered workflow a lot of tabs are persistent between browser restarts and stay open at all times. To try to formalize it, there is a set of core tabs that are permanently open, and there are transient tabs are opened and closed from those core tabs. Before tab groups I used “Tree Style Tab” extension but I like tab groups more. It’s especially cool tab groups are integrated well with containers so that you can have for example I2P tab group tied to I2P container configured to use I2P proxy port to automatically browse all tabs opened within group through your I2P proxy port.
Yes, especially at work. Different tasks, different tab groups. Once the task is done, the group dies. Really useful when working on multiple tasks at “the same time”.
Pair that with multi account containers and temporary containers and it’s a godsend tool for web dev.
Now does that need AI in any capacity? Absolutely not! I’m more upset that they’re even considering such thing because ir sounds utterly useless. A browser should do the browser thing and get out of my way.
It is to some people. My approach though, when I happen to have multiple “work group” to organize, is just to use my OS ability to have multiple windows. No need for any extra bloat, the feature is already there, and it works as I’m used to.
But apparently, using the tools already available to you is not a common skill these days :(
But apparently, using the tools already available to you is not a common skill these days :(
So, are you not understanding that other people work differently, or are you just not using that skill?
Besides offering different approaches for different preferences, there are clear benefits to the extra level of organization. As an additional exercise, try to picture someone using multiple windows and tab groups.
Not everyone operates on the basic level. Hell, why even have tabs? The OS can manage multiple windows, and you can use multiple desktops to achieve the same result without that bloat.
The pathological need to find something to use LLMs for is so bizzare.
It’s like the opposite of classic ML, relatively tiny special purpose models trained for something critical, out of desperation, because it just can’t be done well conventionally.
But this:
Take out the word AI.
If this feature took, say, a gigabyte of RAM and a bunch of CPU, it would be laughed out. But somehow it ships because it has the word AI in it? That makes no sense.
I am a massive local LLM advocate. I like “generative” ML, within reason and ethics. But this is just stupid.
When I’m browsing around with multiple tabs open, the last thing I want is something to start moving them around and messing my flow up. This is a solution looking for a problem.
Yup
Auto naming functionality is neat in some cases, like the AI chat UI itself
Tab groups don’t hit those points at all
Venture capital dumped so much money into the tech without understanding the full scope of what it was capable of. Now they’re so in so deep that they desperately NEED to find something profitable it can do, otherwise they’ll lose the farm.
Firefox has little financial motivation for this, though?
Other than getting “AI” investor money, if that’s the plan… But otherwise it just feels like they’re following a meme.
It makes a lot more sense when you realize that the Mozilla corporation is a for profit run by the same techno-fascist aggrandizing bait-and-switch narcissists as the rest of SV.
I’ve been saying it for years, but I will never donate to Firefox until it is freed from the shackles of a for profit corporation that can use your donation for any profit motive it sees fit; not even related to Firefox.
IIRC, you can’t even donate to Firefox. You can only donate to Mozilla. It seems pretty clear to me why they set it up that way…
Isn’t the “for-profit” Mozilla Corporation owned by the “non-profit” Mozilla Foundation though?
I don’t care. It’s a corrosive force that causes them to pay for over priced CEO’s and integrate services that nobody cares about into Firefox (like pocket) or that runs against their principles (container VPN’s exclusive to Mozillas for-profit VPN).
90% of their cash flow comes from google to be the default search engine - they are probably trying to open up alternative routes of funding to reduce the risk, since it’s not guaranteed that the money will keep coming due to the current lawsuit.
Right, I sympathize with that.
…But also it’s ridiculous. Like why should including a feature with “AI” in it get them VC money? Even if that’s kinda reality?
TBH they should just become a contributor to llama.cpp and market that somehow.
Spoken like someone who’s never interacted with Silicon Valley VCs… just imagine someone with tons of a money, a moderately competent business background, and very little understanding of even the basics of technology that you and I take for granted. And then make them stupid and greedy.
“AI? Yes please! Here’s some money, I’ve heard of Firefox so I know you’re good for it.” It’s not really any more complicated than that, I don’t think.
Well, exactly. Then why the pretense?
They could contribute to some existing local inference effort, do actually useful dev work, and slap their brand on it. It would both be cheaper and “look” better to VCs.
Basically do what ollama’s doing but less shady.
Yeah. There would be a way to do it that I feel like might potentially be useful. The described method (doing clustering instead of just having a similarity threshold to group tabs together, vectorizing the entire tab title through a whole fucking network instead of just tokenizing it and calling two tabs similar if they have uncommon tokens that are within a certain similarity level) really sounds to me like people who have no real idea what they’re doing, just being “ML experts” all over the codebase and fucking things up, and probably walking away very proud of themselves while helping themselves to bunches and bunches of the Mozilla Foundation’s Google-money.
I agree with you on almost everything.
Here i disagree. ML is using high dimensional statistics. There exist many problems, which are by their nature problems of high dimensional statistics.
If you have for an example an engineering problem, it can make sense to use an ML approach, to find patterns in the relationship between input conditions and output results. Based on this patterns you have an idea, where you need to focus in the physical theory for understanding and optimizing it.
Another example for “generative AI” i have seen is creating models of hearts. So by feeding it the MRI scans of hundreds of real hearts, millions of models for probable heart shapes can be created and the interaction with medical equipment can be studied on them. This isn’t a “desperate” approach. It is a smart approach.
Fair point. Not on the semantics per se, but on taking the best approach, 100%.
How do you tell what the patterns are, or how to interpret them?
The recognition of the pattern is done by the machine learning. That is the core concept of machine learning.
For the interpretation you need to use your domain knowledge. Machine learning together with knowledge in the domain analyzed can be a very powerful combination.
Another example in research i have heard about recently, is detection of brain tumors before they occur. MRIs are analyzed of people who later developed brain tumors to see if patterns can be detected in the people who developed the tumors that are absent in the people who didn’t develop tumors. This knowledge of a correlation between certain patterns and later tumor development could help specialists to further their understanding of how tumors develop as they can analyze these specific patterns.
What we see with ChatGPT and other LLMs is kind of doing the opposite by detaching the algorithm from any specific knowledge. Subsequently the algorithm can make predictions on anything and they are worth nothing.
even without AI, to me tab groups are already feature creep bloat in browsers. do people really put that much effort into organizing tabs?
I like the tab groups. I use them often at work to group an issue with related tabs and my attempts at solving it. Also makes it easier to pause work on one problem and work on something else because I have the tabs grouper and know exactly where to go back.
And nobody should stop you installing an extension that provides tab groups. I agree with the other commentator that some features can be left to extensions and don’t need to be part of the core web browser, though.
True, but I’m not sure that an extension would have the necessary access to manipulate the browser like that. I don’t think it should. A malicious extension could do horrible things.
I don’t know if they still do but they used to have. That, however, is something to discuss with the genius decision makers at Mozilla who decide to break extension APIs every couple of years. Firefox on Android still hasn’t recovered from last time.
No, but I think the idea of a second layer of organization to tabs is a wonderful idea. Maybe not a gig of RAM to sort them, sure.
FF already has tab groups. Right click on one.
Yeah, they were added somewhat recently, I know.
You probably look at tabs as something inherently transient. In my tab group powered workflow a lot of tabs are persistent between browser restarts and stay open at all times. To try to formalize it, there is a set of core tabs that are permanently open, and there are transient tabs are opened and closed from those core tabs. Before tab groups I used “Tree Style Tab” extension but I like tab groups more. It’s especially cool tab groups are integrated well with containers so that you can have for example I2P tab group tied to I2P container configured to use I2P proxy port to automatically browse all tabs opened within group through your I2P proxy port.
Yes, especially at work. Different tasks, different tab groups. Once the task is done, the group dies. Really useful when working on multiple tasks at “the same time”.
Pair that with multi account containers and temporary containers and it’s a godsend tool for web dev.
Now does that need AI in any capacity? Absolutely not! I’m more upset that they’re even considering such thing because ir sounds utterly useless. A browser should do the browser thing and get out of my way.
For work at any given point I have 17-20 tabs open. It’s totally useful for me to sort them into tabs to cut out the “noise” when I’m doing research.
I started using tab groups when they released vertical tabs.
It is to some people. My approach though, when I happen to have multiple “work group” to organize, is just to use my OS ability to have multiple windows. No need for any extra bloat, the feature is already there, and it works as I’m used to.
But apparently, using the tools already available to you is not a common skill these days :(
So, are you not understanding that other people work differently, or are you just not using that skill?
Besides offering different approaches for different preferences, there are clear benefits to the extra level of organization. As an additional exercise, try to picture someone using multiple windows and tab groups.
Not everyone operates on the basic level. Hell, why even have tabs? The OS can manage multiple windows, and you can use multiple desktops to achieve the same result without that bloat.
The very first five words of my message was that this was useful to some people.
yes, that’s exactly what i was getting at.
Another local LLM guy here, i fully agree with you - this is probably just a move to acquire capital in the case that the google-cashflow stops.