AI Controls (formerly 'kill switch') are landing in today's Firefox Nightly, and will land with Firefox 148 later this month.
-
@barubary your "this is disgusting" point was in reply to me challenging someone on sneaking untrue accusations into their longer posts. That sneaky behaviour is what I was referring to when I said "I do not let people get away with that".
I did give you the benefit of the doubt with the "if". I now know you're referring to something else.
Yes, there was a tangent, but if people don't want me to pick up on deliberate false accusations in their messages, I suggest they not make them.
@jaffathecake @barubary If the "untrue accusation" you're referring to is the (within quote-marks, but not a quite, instead a restating and possible hyperbole of your position) "anti-AI crazies", then you most definitely need to take a (virtual step back) and consider that it most probably was not ever intended to be a quote, @duke_of_germany can probably confirm, but having spent a fair while reading text written by human beings, I read the quote marks as indicating not a quote, but a somewhat hyperbolic summary.
Without the quotes, I probably would have interpreted it as an actual quote.
-
> are you under a directive that explicitly told you not to do that?
No.
> Or perhaps under some implicit kind of situation (e.g. “I know $manager won’t listen”) which made you not even consider that?
No, but a poll that gives me the evidence to say "hey, you know that place that has a strong representation of people who don't like AI? They don't like AI." did not seem like a good use of my time.
@froztbyte @barubary whereas, I'd seen mocks where you couldn't block AI, whist enabling specific features like translation. I felt this was a major missing use-case, so I creating a poll somewhere that is well represented by people who'd want to use such AI controls.
It's really that straight forward.
-
@firefoxwebdevs oki good
I'd love if this stuff is either configurable via about:config or policies, since I'm able to generate those via nix but im not able access the normal settings which caused me quite some long nights of cursing and debugging in the past
@schrottkatze I guess try the setting, and see what changes in the profile.
-
@firefoxwebdevs "AI is changing the web, and people want very different things from it. We’ve heard from many who want nothing to do with AI. We’ve also heard from others who want AI tools that are genuinely useful."
Have you? Because I've literally only heard the former. See e.g. virtually every one of the 966 comments on https://connect.mozilla.org/t5/discussions/building-ai-the-firefox-way-shaping-what-s-next-together/td-p/109922
@mike @firefoxwebdevs This is precisely where I get confused.
The web is an enormously large sea of connected dynamic content.
Why does the interface to that sea of content need to provide AI? Shouldn't the user use the interface to -find- AI in the sea of content?
For example, instead of the browser providing translation services, people who want translation services can go to their content page of choice to obtain them.
Make "default translator" a setting and let the user choose where to get it.
People who want AI-generated slop can go to their AI-generating slop provider of choice.
Why is it necessary to put it in the interface instead of leaving it in the sea?
This feels like Netscape Navigator "we must have an integrated email client" all over again.
-
@davidgerard @jaffathecake @duke_of_germany @firefoxwebdevs Oh, I see: "Developer relations lead" at Mozilla (and previously "developer advocate" at Google). No wonder he's putting "developer of sorts" in all his profiles/blurbs/blogs.
-
@firefoxwebdevs@mastodon.social
Not quite
The point is that permission prompts are shown when a permission is required to fulfill a user request.
To create a tab group in #Firefox you do not need #AI, so such popup-wide, easy to mis-click, button is a gratuitous waste of user attention. It's just an #ads of a feature that you want people to use and get used to.
As for this being #privacy preserving: where are the models downloaded from? Is it a #CDN that could identify otherwise logged users by their IP and thus learning about the fact they use such model? And how often Firefox will chech for model updates?@giacomo I'll ask about the download location & update checking. I don't know off the top of my head.
But needless to say this is more privacy preserving than browsers that perform translation by sending the text off-device.
-
> are you under a directive that explicitly told you not to do that?
No.
> Or perhaps under some implicit kind of situation (e.g. “I know $manager won’t listen”) which made you not even consider that?
No, but a poll that gives me the evidence to say "hey, you know that place that has a strong representation of people who don't like AI? They don't like AI." did not seem like a good use of my time.
@jaffathecake @barubary okay so if the (for lack of a better term) problem is that gathering data from the fedi might be biased, what about shipping a survey and linking it in update notes?
Firefox 148, pop it into the release notes. "hey, we've done this update. also, we want to know whether you want us to work on this stuff". you'll get answers from a fairly solid bit of your userbase. it'd be quite interesting to see those numbers, don't you think?
-
@jaffathecake @barubary If the "untrue accusation" you're referring to is the (within quote-marks, but not a quite, instead a restating and possible hyperbole of your position) "anti-AI crazies", then you most definitely need to take a (virtual step back) and consider that it most probably was not ever intended to be a quote, @duke_of_germany can probably confirm, but having spent a fair while reading text written by human beings, I read the quote marks as indicating not a quote, but a somewhat hyperbolic summary.
Without the quotes, I probably would have interpreted it as an actual quote.
@vatine @barubary @duke_of_germany even ignoring the quotes…
> Stop portraying Mastodon users as the anti-AI crazies.
Is pretty clearly accusing me of portraying people as anti-AI crazies. This is untrue, and feels like a deliberate attempt to put words in my mouth.
-
@jaffathecake @barubary okay so if the (for lack of a better term) problem is that gathering data from the fedi might be biased, what about shipping a survey and linking it in update notes?
Firefox 148, pop it into the release notes. "hey, we've done this update. also, we want to know whether you want us to work on this stuff". you'll get answers from a fairly solid bit of your userbase. it'd be quite interesting to see those numbers, don't you think?
@froztbyte @barubary user research has been carried out, so repeating it doesn't seem necessary at this stage. I've been asking if there are details I can share publicly, but I haven't heard back.
-
@vatine @barubary @duke_of_germany even ignoring the quotes…
> Stop portraying Mastodon users as the anti-AI crazies.
Is pretty clearly accusing me of portraying people as anti-AI crazies. This is untrue, and feels like a deliberate attempt to put words in my mouth.
@jaffathecake @barubary @duke_of_germany
I will posit that 'anti-AI crazies' is a succinct (and slightly exaggerated) summary of (and this is a direct quote) "I'm sure you're aware that Mastodon has a high representation of folks who don't like AI, so presenting evidence that Mastodon users don't like AI is kinda… well… not really useful."
Just... not so wordy.
So, if that is indeed what you are referring to, again, please take a few virtual steps back, take a deep breath, and consider that maybe, just maybe, you overreacted.
-
@barubary it was a reply to this post https://mastodon.social/@jaffathecake/116006262879508507. It seems reasonable to assume that this is what you were referring to by "this". But if you're saying otherwise, okay.
@jaffathecake No.
"This is disgusting" were only the first three words of my reply. You can't just ignore the other 96% of the message that explain what exactly it is I found disgusting and decide it is "reasonable to assume" whatever.
Also: That post is not you "challenging someone on sneaking untrue accusations into their longer posts". Which is another thing I mentioned in my first reply. And which you ignored (again).
-
@jaffathecake @barubary @duke_of_germany
I will posit that 'anti-AI crazies' is a succinct (and slightly exaggerated) summary of (and this is a direct quote) "I'm sure you're aware that Mastodon has a high representation of folks who don't like AI, so presenting evidence that Mastodon users don't like AI is kinda… well… not really useful."
Just... not so wordy.
So, if that is indeed what you are referring to, again, please take a few virtual steps back, take a deep breath, and consider that maybe, just maybe, you overreacted.
@vatine @barubary @duke_of_germany if you feel that way, let me be clear: I do not think people who dislike AI are crazy.
If I did, I'm not sure why I'd go to this extent to gather and represent their views in order to change direction of a particular feature.
I don't think it's unreasonable to object to the accusation that I think people are "crazy" - that's a pretty strong word, and not one I want people to put in my mouth.
I hear that you feel that's an overreaction. I simply disagree.
-
@davidgerard @jaffathecake @duke_of_germany @firefoxwebdevs Oh, I see: "Developer relations lead" at Mozilla (and previously "developer advocate" at Google). No wonder he's putting "developer of sorts" in all his profiles/blurbs/blogs.
@barubary @davidgerard @jaffathecake @duke_of_germany @firefoxwebdevs You could always report the user as a persistent troll misrepresenting who they are.
-
@mike @firefoxwebdevs Preference falsification, it's seen as socially condemnable to be ok with AI, especially in the circles that Mozilla frequents, so I'm not surprised public comments trend negative. Given they've sent an anonymous survey out, I'm pretty sure the balance is split between the no-AI and AI everything camps.
I don't get the vitriol towards Mozilla, at least not for this change, they literally show you how to turn off their AI features
@budududuroiu @mike @firefoxwebdevs It's not about being able to turn off the features. It's about having to download and store them and give implicit permission to have them on your computer in the first place.
It's like someone saying "Hey, I mailed you that stack of books you want, but people told us they wanted more pictures so we made illegal copies of artworks and bound them in. The art pages come with preset double-sided tape on them so it's easy for you to stick the art pages together and not see them if you don't want. No harm, no foul, right?"
The harm isn't in the viewing (or not viewing), it's in the implied consent to the unethical behavior that made the viewing possible, and in asking customers to fix it after the fact.
-
@froztbyte @barubary user research has been carried out, so repeating it doesn't seem necessary at this stage. I've been asking if there are details I can share publicly, but I haven't heard back.
@jaffathecake @barubary just to clarify, with "user research" do you mean the polls on bsky/fedi/etc, or were there other surveys conducted in places?
(asking because I don't think I saw any such surveys anywhere, and I'm _moderately_ on top of seeing this stuff go around)
-
@jaffathecake @barubary just to clarify, with "user research" do you mean the polls on bsky/fedi/etc, or were there other surveys conducted in places?
(asking because I don't think I saw any such surveys anywhere, and I'm _moderately_ on top of seeing this stuff go around)
@froztbyte I don't believe they were social media surveys, as the goal was to get a representative sample among locations and types of user. But it isn't information I currently have, and maybe won't be able to share.
-
@budududuroiu @mike @firefoxwebdevs It's not about being able to turn off the features. It's about having to download and store them and give implicit permission to have them on your computer in the first place.
It's like someone saying "Hey, I mailed you that stack of books you want, but people told us they wanted more pictures so we made illegal copies of artworks and bound them in. The art pages come with preset double-sided tape on them so it's easy for you to stick the art pages together and not see them if you don't want. No harm, no foul, right?"
The harm isn't in the viewing (or not viewing), it's in the implied consent to the unethical behavior that made the viewing possible, and in asking customers to fix it after the fact.
-
@froztbyte I don't believe they were social media surveys, as the goal was to get a representative sample among locations and types of user. But it isn't information I currently have, and maybe won't be able to share.
@jaffathecake if it's not information you currently have, doesn't that leave the possibility that they might in fact reflect a negative or inconclusive outcome in polling?
I mean, I get that you don't have eyes on this yourself and that you can't speak to it, I'm not putting this on *you*. but do you see how it could be possible that, without these results being open, someone could be going full steam ahead _in spite_ of the findings?
-
@firefoxwebdevs @robotistry @budududuroiu TBF, client-side translation is pretty cool, and generally better than sending the text to Google or similar.
-
@alextecplayz here's the help page, so you can judge for yourself https://support.mozilla.org/en-US/kb/firefox-ai-controls. We mostly refrain from using the term "opt-in" because people have different definitions of opt-in.
Models don't download until you engage with the feature, but some folks have said it's only opt-in if even the entry points are in a separate binary.
I asked for UI that shows downloaded models, but there wasn't time for that in 148. I'll keep asking for it
@firefoxwebdevs @alextecplayz For the time being, you can see those models in about:addons, under the “On-device AI” page, and there will be a “Firefox uses this to sugest tabs” description, like this.
