CIRCLE WITH A DOT

@MicroSFF And it doesn't even work.

@elkaki Oh, that's neat!

@firefoxwebdevs @robotistry @budududuroiu TBF, client-side translation is pretty cool, and generally better than sending the text to Google or similar.

@budududuroiu @firefoxwebdevs I wonder what your motivation can be for so strongly wanting a 99% consensus to be misrepresentative.

@Mastokarl @alahmnat @firefoxwebdevs First, apologies for ruining your sleep!

Second, it is encouraging to think Kagi can do this. If it has an effective AI filter that might be what distinguishes it enough from Google that a critical mass of people switch.

(I can't imagine how you could program such a thing, though, when the whole point LLM output is to mimic human output as closely as possible.)

@budududuroiu @firefoxwebdevs "Preference falsification" is a neat rhetorical trick to ignore results one doesn't like. But the simple fact is that Mozilla made a post asking for feedback, and LITERALLY 99% of the replies were strongly negative.

So, who are Mozilla building this for? Not the users they asked about it, that's for sure.

@Mastokarl @alahmnat @firefoxwebdevs I don't think anyone has illusions that the web was free of bullshit before LLMs.

But LLMs have made it vastly vastly easier to generate bullshit, and as an absolutely inevitable result, LLM-generated bullshit is now drowning out everything else.

This seems pretty uncontroversial to me.

@alahmnat @firefoxwebdevs Precisely.

@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