On second look:

Accusations against school monitors reported by parents across France include children being screamed at, pushed, having their hair pulled, being denied food, forced to eat until they vomited and being sexually assaulted or raped.

That’s actually believable. Untrained people lose their temper and assault kids, but there are only a few cases of rape.

Quick google says that France completely outlawed assault on children only in 2019, though schools had been instructed not to use such punishments in 1991. Systemic failures to correct the monitors seem unsurprising.

84 preschools, about 20 primary schools and about 10 daycare centres,

Indeed. That sounds unbelievable. And that’s a red flag.

https://en.wikipedia.org/wiki/Day-care_sex-abuse_hysteria

I don’t actually know what you are talking about. This isn’t about the waste heat of data centers, right?

Sheesh. You got some self-esteem issues. I didn’t call you stupid. I just observed that some people here are psychologically distressed by AI.

I posted it because it is really neat, but I think I shouldn’t have. I know many here are psychologically distressed by AI. While I still would post things that are important enough that they should be known, this is probably not it (unless you are a mathematician, but this is not a very STEM community).

everything else that surrounds this headline.

Like what?

Cool cool. How long will that stay true?

It doesn’t explain tokenizers well (or at all). There are better videos on the subject.

Anyway. Suppose you wanted to spell giraffe with the English alphabet in any arbitrary, phonetic way. You could also spell, for example; “jeeruff”, or “djirough”. You could count how many phonetically correct ways there are to spell “giraffe”.

Tokenizers break a text into sequences of characters (even individual characters), called tokens. Different tokenizers use different tokens. The one they use here has “gira” as a token, but also “g”, “i”, “r”, and “a”. So you could tokenize the same text in different ways. They have a slide where they show the possibilities.

I can see how the confusion arises. People here have been saying that LLMs are useless. Now you read about Mythos finding these exploits. But Mythos is not any better than the previous LLMs. Which means that it is useless. So any suggestion to the contrary can only be a marketing tactic. QED

Looks like we’re no longer doing the bit about the hallucination machine being useless.

Ist ja auch eine blöde Situation für die Polizei. Da stehst du irgendwo rum und weißt nicht, ob du gerade Opfer eines schweres Gewaltverbrechen wirst. Ist so eine Situation öffentlich, oder nicht? Wenn nein, dann ist eine Tonaufnahme der Gespräche nämlich digitale Gewalt (§201 StGB)! Oder stell dir vor, die veröffentlichen das unverpixelt. Da wirst du praktisch grün und blau geschlagen und weißt es noch gar nicht. Es gibt ja keinen Unterschied zwischen digitaler und echter Gewalt.

No. That’s not what these words mean.

Look… Doesn’t that feel kind of like self-pitying rot to anyone here? Why are we dependent on US technology? Because we are so ethical and pure…

Here’s the truth.

In the late 90s, a German student created a search engine in Germany. It was a little thing. It only scraped a few hundred media outlets. You signed up, defined some keywords, and when an article matching those keywords was published, you received a notification.

He immediately was sued and forced to shut down under copyright law. Google could operate in the US under Fair Use.

Eventually, years later search engines were legalized in Germany (and the EU). But by then the Internet was dominated by US companies. It makes no sense to spend billions to build a European Google that does exactly what Google already does.

The reason that there is no European Google is that we insist that information must be owned. No data processing without the explicit consent of the owner. Which means, we insist that some intellectual property owners should be allowed to extract rent from us all.

You’re so out-of-touch if you believe that the EU is against this.

It’s especially weird coming from an apparent German. You know all that rhetoric about how the internet is no “wild west”. That means locking everything down. Only the properly licensed professionals are allowed to do stuff with properly regulated tools. That’s how it goes in Germany.

This stuff is what Google is supposed to do.

I simply don’t know what you mean at all.

I don’t actually know what you mean.

Why do you think that?

*gestures at OP*

Were you thinking of this instead?

No. Why would I?

That hasn’t aged well.

Yes. I hesitated to post this because I understand that many here would prefer not to know. But, at least, people need a chance to learn the facts and make their own decisions. The amount of anti-AI disinformation is crazy.

1. The problem with trad SoMe is that it is monopolistic. That’s because these companies “own” the data and gate-keep access. If you want open social media, you must not have a gate-keeper. Which means that you can’t have someone who controls access. That’s a fundamental trade-off.

2. So what? Should posts be anonymous as long as they are short?

3. No. It’s always data+owner. It doesn’t matter if the data is only a single bit.