Mozilla wants to be an AI company. This is data collection to support that. Telemetry to understand the user browsing experience doesn’t need to be content-categorized.
I want an open source AI to sort my tabs and understand them and answer my question about their content. But locally running and offline
Unless they’re going to publish their data, AI can’t be meaningfully open source. The code to build and train a ML model is mostly uninteresting. The problems come in the form of data and hyperparameter selection which either intentionally or unintentionally do most of the shaping of the resulting system. When it’s published it’ll just be a Python project with some magic numbers and “put data here” with no indications of what went into data selection or choosing those parameters.
I just want a command line interface to my browser, then I’ll tell my local mixtral 8x7B instance to “look in all my tabs and place all tabs about ‘magnetic loop antennas’ in a new window, order them with the most concrete build instructions first” 100% open source model. I’m looking into the marionette protocol to accomplish this. It would be nice if it came with that out of the box.
What does “open source” mean to you? Just free/noncorporate? Because a “100% open source model” doesn’t really make sense by the traditional definition. The “source” for a model is its data, not the code and not the model itself. Without the data you can’t build the model yourself, can’t modify it, and can’t inspect why it does what it does.
I think the model can be modified with LoRa without tge source data ? In any case, if the inference software is actually open source and all the necessary data is free of any intellectual property encumberances, it runs without internet access or non commodity hardware.
Then it’s open source enough to live in my browser.
You can technically modify any network weights however you want with whatever data you have lying around, but without the core training data you can’t verify that your modifications aren’t hurting the original capabilities. Fine-tuning (which LoRa is for) isn’t the same thing as modifying a trained network. You’re still generally stuck with their original trained capabilities you’re just reworking the final layer(s) to redirect/tune it towards your problem. You can’t add pet faces into a human face detector, and if a new technique comes out that could improve accuracy you can’t rebuild the model with it.
In any case, if the inference software is actually open source and all the necessary data is free of any intellectual property encumberances, it runs without internet access or non commodity hardware.
Then it’s open source enough to live in my browser.
So just free/noncorporate. A model is effectively a binary and the data is the source (the actual ML code is the compiler). If you don’t get the source, it’s not open source. A binary can be free and non-corporate, but it’s still not source code.
I mean, I would prefer a data set that’s properly open, “the pile” laion, open assistant and a pirate copy is every word, song, video ever written and spoken by man.
But for now I’d be happy to fully control my browser with an offline copy of mixtral or llama
Here’s the current list of categories we’re using: animals, arts, autos, business, career, education, fashion, finance, food, government, health, hobbies, home, inconclusive, news, real estate, society, sports, tech and travel.
No pr0n?
Inconclusive = pr0n is probably a pretty reliable mapping.
deleted by creator
deleted by creator
The important part that you should know (and should already be using):
Remember, you can always opt out of sending any technical or usage data to Firefox. Here’s a step-by-step guide on how to adjust your settings.
They should have put more emphasis on the possible usages for what they find out…
firefox develops an optional predictive search feature like every other search engine and browser has that actually protects user privacy that can easily be turned off so naturally the internet loses their mind over it and declares firefox dead.
don’t worry, it’s balanced out by the every other day threads of firefox shills screeching about how much more private it is and how it uses so much less ram.
people never want to admit that things aren’t black and white.
To improve Firefox based on your needs, understanding how users interact with essential functions like search is key.
Buddy, I just want to type a search term and get results. Stop spying on my search. Your only job is to transfer it to the server and then present the result. I don’t need you to suggest some bullshit to me, or think of “ways to improve search”.
This helps us take a step forward in providing a browsing experience that is more tailored to your needs, without us stepping away from the principles that make us who we are.
No. What the fuck? They are sounding more and more like Google. We need a new alternative that isn’t built from Gecko or Blink or whatever the engines are called.
Buddy, I just want to type a search term and get results.
Telemetry can help them do better at providing that. Devs aren’t magical beings, they don’t know what’s working and what’s not unless someone tells them.
That’s like saying the window pane between me and the teller has to understand the conversation and dynamically modify the light between him and I. The window pane’s only job is to let light through. Keep it at that.
No, this analogy would make more sense if it was a matter of recording a large number of interactions between customers and tellers to ensure that the window isn’t interfering with their interactions. Is the window the right size? Can the customer and teller hear each other through it? Is that little hole at the bottom large enough to let through the things they need to physically exchange? If you deploy the windows and then never gather any telemetry you have no idea whether it’s working well or if it could be improved.
You’re describing telemetry to improve the overall performance of the window. That’s very different from what Mozilla: listening in to what is sent between the teller and I. They even gave an example of a trip to Spain and recording it as travel. That’s going way beyond the performance of a window. The teller is probably already doing that. The window operator has no business listening in on that discussion nor recording even a summary of details of the discussion.
The analogy isn’t perfect, no analogy ever is.
In this case the content of the search is all that really matters for the quality of the search. What else would you suggest be recorded, the words-per-minute typing speed, the font size? If they want to improve the search system they need to know how it’s working, and that involves recording the searches.
It’s anonymized and you can opt out. Go ahead and opt out. There’ll still be enough telemetry for them to do their work.
Telemetry doesn’t need topic categorization. This is building a dataset for AI.
That would be a terrible AI.
The example of the “search optimization” they want to improve is Firefox Suggest, which has sponsored results which could be promoted (and cost more) based on predictions of interest based on recent trends of topics in your country. “Users in Belgium search for vacations more during X time of day” is exactly the sort of stuff you’d use to make ads more valuable. “Users in France follow a similar pattern, but two weeks later” is even better. Similarly predicting waves of infection based on the rise and fall of “health” searches is useful for public health, but also for pushing or tabling ad campaigns.
lol use a fork - I’m sure they’ll have it turned off. Writing a browser engine is non-trivial.
Its exactly this kind of bullshit that firefox should not do…
As much as I hate to say it, Firefox is a privacy mess.
Pocket and Fakespot have very bad privacy policies. The Windows version has a unique Mozilla tracker if you download the installer from the website, and the android version has Google Analytics built in. The existing and new telemetry is a but heavy, but it’s anonymised so it’s really the lesser of the various evils.
My recommendation is LibreWolf & Fennec as alternatives.
sigh
All we want is 1990s Google, guys. That’s really all we want. None of this AI BS that kind find a country in Africa that starts with a K, just Google without the evil enshitification layer on top.
I think people forget how awful google pre ~2008 was. Not in terms of the bullshit they do nowadays, just in quality of results really.
Huh. I used it pretty much since the start and I certainly don’t recall it being that bad? Like you got a lot of relevant content up front usually.
If you had the right query, yes. But getting there if you didn’t know the exact words in the website used to take a number of attempts and google-fu. By early 2010s this was vastly improved.
I feel like you had to learn how to use it, operators and phrasing etc. They dumbed it down with search suggestions and even further by changing search terms to synonyms, and now outright ignoring terms. Height of Internet search was definitely pre 2008. More like 2005.
I switched from Alta Vista at Google in the early 2000s because the Alta Vista index was stale and full of spam. Google search tools were comparatively primitive (av let you do things like word stem search) but the results were really good.
deleted by creator
I support anonymous telemetry collected by a small non-profit that helps protect our freedom. Not big tech.
This isn’t even telemetry, it’s data collection for AI. That they refused to say that let’s you know that they think what they’re doing needs to be obfuscated.
If they refused to say it how do you know its the case? Also how would the data described in the article be useful to an ai, genuine question.
In life, people will frequently say things to you that won’t be the whole truth, but you can figure out what’s actually going on by looking at the context of the situation. This is commonly referred to as “being deceptive” or sometimes just “lying”. Corporate PR and salespeople, the ones who put out this press release, do it regularly.
You don’t need to record content categories of searches to make a good tool for displaying websites, you need it to perform predictions about what users will search for. They’ve already said they wanted to focus on AI and linked to an example of the system they want to improve, it’s their site recommender, complete with sponsored recommendations that could be sold for a higher price if the Mozilla AI could predict that “people in country X will soon be looking for vacations”.
I am a dev and I do not support telemetry
deleted by creator
Same. If It’s to exist at all, it should be opt-in and explicit about what it’s doing.
No one supports telemetry. People support Mozilla, because they are the maintainers of the last standard respecting, open source and independent browse engine.
That’s pretty important as Microsoft and Google etc are trying to take possession of the internet for themselves .
Will this affect libre wolf?
Nope, they cut all the Mozilla stuff out
To disable it in about:config
browser.search.serpEventTelemetry.enabled = falsebrowser.search.serpEventTelemetryCategorization.enabled = falselike and subscribe!
