Poisoning what? Intentional poisoning doesn’t work, and self-consumption only works when it’s the exact same model feeding its own next version.

Women using sycophantic chatbots that they 100% control, is stil about power and control. One is about bodies, the other is about minds, that’s the only difference.

Is it a majority stake? In the video, he makes it sound like they’re co-owners… which says nothing, and the “our brands” wording is also common as a hyperbole.

If they earn money from pushing quality content, adding similar channels, adding Patreon financing, backer-only specials, merchandising, placing content on other platforms… “the other shoe” could be simply making a profit.

Whether Electrify will eventually sell out in turn, is a separate question.

When one has 20 million subscribers who get notified of the videos, those few cents per Ad, add up quickly. He’s also worked his ass off on the quality side of YouTube, which has earned him a lot of good will and supporters.

Goes to show that quality content, can also be profitable. It’s not all about the bottom dwellers.

Depending on how much is “too much power”, people might still want to purchase them at a discount for self-hosting purposes. The future is most likely to go through a decentralization of AI services, with spme higher efficiency large providers, combined with lower efficiency edge nodes for less demanding usage… at least, until the next order of magnitude technological shift.

Counter proposal: dynamic tactile buttons.

The tech already existed in 2013, just develop it further.

https://www.robaid.com/gadgets/tactus-technology-haptic-feedback-on-dynamic-user-interface.htm

Check the history of ReCaptcha: it started by helping digitize booksxfir the Gutenberg Project, then once it got acquired by Google, it switched to house numbers, street signs (auto driving?), and is now helping with object identification.

Strictly speaking, math gets proven from scratch by every math student. Software is slightly different, since most of it never gets a formal proof at all.

Using a complex GPT-4o prompt, they sought to pull out tweets that focused on “superficial topics”

Wait a moment… They asked an LLM, to tell them what was “junk”, and another LLM, trained on what an LLM marked as junk, turned out to be a junk LLM?

It talks about model collapse, but this smells like research collapse.

deleted by creator

Yes. The issue is the accepted belief (among professionals) that “IQ doesn’t change through life”… which is nice as a goal to develop a less biased “ideal IQ test”, but also a really bad preconception when evaluating actual IQ test results.

There are many preconceptions like that in psychology, they need a periodic kick in the butt from actual data.

The problems found on an IQ test are of a very specific nature. “Complete the following number sequence”, “Which of these shapes doesn’t belong”, etc.

A full IQ test has more kinds of problems than that, some of them more open ended and with multiple possible answers. It’s still kind of a mess, the tests need to be kept secret to have any value, and the interpretations of what is a valid answer, are sometimes dubious.

The problem comes from people responding to them, in communities that are text-first. You can flag a person as a troll… but unless you ban them, then eventually the discussions spill over.

Overwatch doesn’t have that same issue, because text is not at its core; you can mute people, and keep playing the game. Game-disrupting players though, still get banned.

That would work, if it wasn’t individuals who don’t see others as people. The moment someone dehumanizes others, there are dedicated professional who can babysit them back to sanity.

My bad, Kiwix is on IzzyOnDroid. You’re right, F-Droid itself might have a tougher time.

F-Droid does automated patching already, they often remove “anti-features”. The migration is a PITA, but not the end of the world, Kiwix did that recently (split into Play Store and F-Droid versions).

No matter how much epoxy you put on a broken condom, it’s not going to work. The moment ads get sent via the same encrypted “tunnel”/connection as the service, you can no longer separate them at the network level.

I downloaded Sora.

Not Sora, Sora’s app. Sora is a cloud service.

And if the business model of open AI is advertising

It’s not, it’s to sell AI services to other businesses.

Cuz the business model was supposed to be 40% of people lose their jobs

Was not, it was always to sell AI services to other businesses. Marketing hyperbole… that’s marketing for you.

in that world, maybe maybe it’s worth half a trillion dollars. In that world, maybe Nvidia is worth the entire pharmaceutical industry combined.

The largest breakthrough in AI so far, is precisely its use in the pharmaceutical industry.

But Nvidia is definitely not worth the entire pharmaceutical industry combined

No, it’s worth more.

if the business model is making slop talk worse by putting slop advertisements into it.

It’s not. Not to say there won’t be advertisements, and propaganda, and manipulation… that’s just an extra, not the business model.

I’m scrolling Martin Luther King Jr. going 67. Like, that’s got to be illegal somehow. Like, I’m pretty sure that you can’t just do that.

It was not, is not, and public domain is public domain for a reason.
Asking for something else, is censorship, like what Trump is asking for right now.

The friction matters. How easy it is to do this stuff matters.

Censorship talk again.

And then they put like a little watermark here that’s like extraordinarily easy to crop out. It’s never in the middle. It’s always at the top or the bottom.

Plenty of videos out there with the watermark right in the middle.

They shouldn’t be in charge of the kind of thing that could literally by their own admission destroy humanity.

Hook, line, and sinker. More marketing hype.

Update: I see the issue now.

I don’t see the issue:

1. F-Droid compiles the apps
2. F-Droid publishes the apps
3. F-Droid gets registered
4. F-Droid subclasses app identifiers under its own

…and things go back to normal.

You won’t be able to update an F-Droid app with an Obtainium one, but Android is already warning about those, not allowing automatic updates from mixed sources.

$8.76/kWh is a terrible fucking investment

Not kWh, it says watts not watt-hours.

Still a silly way to refer to computing power, though.