@froztbyte I suppose that's literally true. But, I also think a mountain is made out of the molehill that is AI in Firefox. The vast majority of the dev time is on other things.

To be clear, I'm not someone who has personally found AI generally useful in browsers (aside from a couple of one-off automations), but my feelings aren't strong enough to deny those features to others.

@jaffathecake @duke_of_germany @firefoxwebdevs @davidgerard forget AI, at this point Jake is putting me off Firefox

@sotolf @firefoxwebdevs the "don't care" market means people love AI, of course!

And if that forms part of Mozilla's measurements, they're absolute idiots.

Better option: release a "No AI" build - https://notnull.space/@paul/statuses/01KGHVQ05QJFHPMBDJT16KCESZ

@duke_of_germany @jonny @jaffathecake @firefoxwebdevs @davidgerard since 20 years ago I've been switching my friends, family, and coworkers and classmates to firefox. Its depressing, but I no longer recommend firefox. Let's ditch the AI please.

@paul well, if we created binaries for all combinations of the current 5 features, that would be 32x-ing the number of binaries per build. And I think people would still be unhappy depending on which was seen as the default.

The AI Controls give an easy way to have that granular control, and you don't need to switch binary just to try a feature.

@sotolf
haaaa. well it is true.

@firefoxwebdevs

@jaffathecake @froztbyte
To be clear. Making commercial chatbots available as first class citizen of the browser knowing their baggage in terms of ecological and social destruction is OK with you ?
Making non-authored, non-reviewed translations/summaries available as a first-class citizen of the browser doesn't even light an ethical warning ?

These could be available to users *who want them* without being promoted on the level of a standard experience of the web browser, who's denied anything?

@ddelemeny @froztbyte I don't personally use the chatbot feature.

I do use translation, with full awareness that it's a machine translation, and I consider being able to read parts of the web that aren't in my native language a wonderful thing, and I'm glad it's done in a privacy-preserving way.

The models were downloaded when I asked for the translation to happen. They weren't there beforehand.

@jaffathecake @froztbyte you aren't reading a part of the web. The translation was never on the web. Nobody had the opportunity to make sure it's right and nobody ever will. Ethics go farther than privacy.

I don't care if you personally don't use the chatbot feature. The existence of it in the default build will actively normalize and promote it to a userbase larger than the population of Brazil. Is social irresponsibility part of the manifesto here ?

@jaffathecake @duke_of_germany @firefoxwebdevs @davidgerard You're really not helping yourself or Mozilla with this stuff, Jake. We're never going to see you as someone's victim when you're so set on avoiding the *real* issue that literally everyone is calling out to you. You don't get a medal for seeking feedback if you then ignore that feedback and openly disdain the users who provided it. Bogus shit.

@duke_of_germany @jonny @jaffathecake @firefoxwebdevs @davidgerard

It is impossible to read this post without thinking of this recent article about Wikipedia’s internal strife and how communities become ossified and refuse to adapt to new conditions and then promptly die.

The organization version of progress advancing one funeral at a time

Is Wikipedia's Volunteer Model Facing a Generational Crisis?

The clash over AI-generated summaries reveals Wikipedia's challenge in adapting to younger audiences' media habits.

IEEE Spectrum (spectrum.ieee.org)

@ddelemeny @froztbyte how do you feel about a11y tooling that analyses images to describe them? Therefore providing people with visibility into things they otherwise wouldn't have.

That also is generating content in a format that was never on the web. Nobody had the opportunity to make sure it's right and nobody ever will.

@duke_of_germany @jonny @jaffathecake @firefoxwebdevs @davidgerard hi duke can you please give some context,i am new here in mastodon so i have no clue what are you talking about and beside that what do you recommend instead of firefox cause as far as i know google opera and brave all of them worse in many aspects
Edit:sorry duke i just saw his interesting comment about users here but i am still asking you if you have better recommendation

@jaffathecake @duke_of_germany @firefoxwebdevs @davidgerard In case my feedback isn't clear, here's how your responses are (correctly) received here:

Chaff (countermeasure) - Wikipedia

(en.wikipedia.org)

@jonathankoren @duke_of_germany @jonny @jaffathecake @firefoxwebdevs

> the volunteer community that built this encyclopedia has lately rejected a key innovation designed to serve readers

this article is AI shilling, nonsense and that should have been obvious at a glance

@jaffathecake @froztbyte exactly the same. You don't go around the ethical problem of the absence of authorship by invoking pathos (and borrowing legitimacy from disabled folks is morally sketchy to put it mildly).

Even more so as available technologies are far remote from being able to convey intent, context and nuance. This is misplaced half-baked solutionism.

@ddelemeny @froztbyte ok, then we simply disagree.

@sotolf @barubary @jaffathecake @duke_of_germany @firefoxwebdevs it is literally Jake's job to be like this

@davidgerard @duke_of_germany @jonny @jaffathecake @firefoxwebdevs Is “shilling” what the kids call “an opinion that makes me uncomfortable because it doesn’t comport to my preëxisting beliefs”?

This type of stimulus-response behavior is really beneath you.

You could have stopped to think about how organizations actually adapt without buying into the billionaire hype, but instead it’s just banging rocks together and grunting

@jaffathecake @ddelemeny I’ve seen a number of accessibility-using people critique this stuff, fwiw. A number of them echo the “let me choose my own”

For my own part: given the error rate problem combined with the (presumably) reduced capacity for the affected user to verify veracity/accuracy, I think it’s *even more* dangerous when this stuff hallucinates or miscategorizes

I agree the *idea* is nice, but do not believe that this implementation/generation has solved the problem