Yes but if ai raises your kids for you, you can work longer hours in the mines!

Stonks go up!

Technology, solving problems nobody had.

Because LLMs are really bad at telling stories.

...Look, I'm an advocate for LLMs here. I was using 6B models to mess around with roleplaying and storywriting in like 2022, with GPT-J finetunes. But that's with human guidance, using it as a tool or "sounding board."

They're fascinating writing aids.

But by themself, unattended, off-the-shelf LLMs have a horrendous architecture for writing long form stories, basically. They go off the rails at the drop of a hat, and cannot self correct. They lose track of the longer context, and write things that don't make sense. They have weird, very inhuman patterns of using the same character names and ideas, or mimicking prose from earlier in the context where it's not applicable at all. Their prose is terrible in an eerie, repetitive way.

So for children, there's a risk of:

• Not-safe-for-family content popping up literally randomly, in the middle of the story, which a parent might miss skimming.

• Weird, nonsensical storylines I would classify as "brainrot"

• Very repetitive themes and ideas, basically a lack of variety.

• Instilling strange linguistic habits.

• Emotions and tone that don't line up with the content, especially when TTS is used (as most TTS models aren't tightly integrated into the main LLM; they're tiny, dumb models reading a script blind, with no context).

...Have you seen any horror movies where something is supposed to be human, or acting like it, but it isn't quite right? Because it clearly doesn't understand, its just mimicking?

That's the risk here, one you wouldn't get even with TikTok doomscrolling. And that's the best case scenario, without even getting into influence ops, engagement maximization techniques, or enshittification Meta is notorious for.

Researchers have built finetunes with frameworks to try and improve autonomous LLM writing, but as someone who's followed their research, I can tell you there's a zero percent chance Meta implemented it.

Obviously don’t trust Meta to deliver content to a child, but how is this different than other forms of media that kids are already ingesting on TV, or phones or tablets?

Okay so like this was obviously a false flag attack to try to justify removing guardrails so AI companies can be even more unregulated and out of control, yeah?

Would be interesting, being an island and all, if data centers were required to use sea water as intake.

Salt bad for cooling? Sounds like you should desalinate it first then. Possibly adding water to British supplies.

Granted the output wouldn't be "clean" but would be fresh and use existing facilities to clean it now that the hard parts are done.

It's crazy how its robust to boxes being taken offline during training. That's honestly magic.

I went to the Google reviews, and man... It looks like shit. Whoever looked at these and said eh good enough is a fucking moron. I low key wonder if they did it as ragebait to get people talking about the establishment. No such thing as bad publicity and all that. There's now a bunch of people 5 starring them just because of the negative press who likely didn't actually go there.

“take pictures of your food or drinks”

Did I say they were? I didn't make that claim.

How do you know they were even taking pics of real food before? It has always been fake.

You know what, fuck them. This is exactly the kind of low-effort shit that shows you how much an owner cares about their product.

If you can't find the time to take pictures of your food or drinks, you deserve the backlash.

if there's never an out of stock message on a service, then that service is overselling

Reality usually catches up with hype eventually.

Understanding model behavior seems more valuable than chasing bigger benchmarks.

That's exactly the kind of AI application I'd like to see more of.

More competition is healthy, especially if it drives openness.

Interesting idea. Efficiency matters as much as capability.

Running locally with tiny hardware requirements is genuinely exciting.

simply performing the underlying mathematical operation

does this not assume that a) there is always an analytical mathematical operation, and b) all humans do is maths, complicated maths

neither of these are the case, btw

machines are not like humans, nor humans, machines

One of the definitions of insanity.

This is amazing if it's any good!

Clickbait title but cool data science

LGTM

Hidden or hallucinated?