Something that hasn't been made clear: Firefox will have an option to completely disable all AI features.
We've been calling it the AI kill switch internally. I'm sure it'll ship with a less murderous name, but that's how seriously and absolutely we're taking this.
…
All AI features will also be opt-in. I think there are some grey areas in what 'opt-in' means to different people (e.g. is a new toolbar button opt-in?), but the kill switch will absolutely remove all that stuff, and never show it in future. That's unambiguous.
…
I'm not asking for faith in our direction - the thing I love about the Firefox community is how open, honest, and technical it is.
But I do ask that you don't have the opposite of faith. Like, try not to be determined that we're going to do the wrong thing here.
…
I hope we can (re)gain your trust here.
I don't personally work on this stuff, but I'll try hard to answer any questions you have.
And other than that, I'll get back in my lane, and stick to web platform stuff.
- Jake (@jaffathecake)
Just be glad this thread wasn't a long-ass video. It almost was.
@firefoxwebdevs Yeah I think most people mainly deplore the hype and the resources spent on technological trends whose benefits are not always obvious. Before that, Mozilla advertised about FirefoxOS, before killing it to focus on IoT, before moving on to blockchain, then crypto, then NFT's and now IA. In more that 10 years, none of this projects produced anything useful for the users.
@firefoxwebdevs Right now, Mozilla would probably be the first company to be diagnosed with ADHD. It really can't seem to focus and do something productive. The question was never "should Firefox have IA?". The question is "to do what?". Mozilla is communicating that IA is coming. Not announcing a new feature. TBH, it's worrying. IA should be an implementation detail, not the central point.
@firefoxwebdevs It's like you're a car company and you're advertising a new car with leather in it. Ok, cool but what is it? A berline, a pickup, a SUV? Will I recharge with electricity or fuel? And your answer is: "it has leather in it!"
It's… not great.
@christophehenry @firefoxwebdevs look, you have a point about communication. It's hard and Mozilla isn't top notch at it, to be polite. But also, Mozilla never worked on IOT, nor blockchain nor crypto stuff. There were vague talks of transitioning some of the Firefox OS resources into IOT exploration for a very brief time, which didn't end up happening so I'll give you that one. But where the hell hell is the rest coming from?
There have been some hysterical responses to the Mozilla AI announcements, with a number of people instantly swearing off Firefox forever.
Frankly, I'll leave it and see what actually happens. Firefox is too important for the things I do.
They can play around with so-called "AI" a bit, so long as it's truly private, free software and I can completely remove it if I wish (which I probably do).
@ecadre agreed. Let's put the pitchfork away until something bad actually happens. Right now most of the the AI in Firefox is things like tiny models that do local translation (rather than send the whole text to Google who would use their own neural networks to do it), automatic captioning of images that lack alt text, text-to-speech, speech-to-text, and other small neural nets that take less energy to train than a run of our test suite. Not all machine learning is chat-freakin-gpt.
@nical @ecadre Why isnt there already a kills switch or opt in for these "tiny models?"
Having a dozen about:config settings be the only effective configuration option to disable Mozilla's current forays into AI doesnt exactly inspire trust about its future plans when the new CEOs first post is all about Firefox's AI future.
We've seen the direction the browser is already moving in with AI and user consent, so it should be no surprise that people are even more skeptical now.
@liquidlamp @ecadre that's not my area so all I can offer is the following perspective: would you like a kill witch for all features that begin with the letter "t" or that are implemented with functional programming? That's about the same level of rationality. These huge LLMs that gobble up energy like there is no tomorrow are not making the world a better place (and don't get me started on training on the work of artists without their consent) but a neural isn't necessarily power hungry
...
@nical @liquidlamp @ecadre well, things like "suggest tab group name" shouldn't go anywhere near LLM anyway!
