Jyunmi

And their efficiency in generating new models and all of that.

And I was going to say, when you said, you know, not as good as, you know, 3.0 thinking, it's like...

For now.

So next week, you know, there's rumors floating around.

OpenAI is going to release something and then something in January and all this kind of stuff.

So we'll see what happens.

Probably while we're talking, they're going to... A new release.

That's right.

Right, right.

Exactly.

I find it interesting that they're able to come up with this sparse system because that seems to be on the...

That seems to be after ingestion and part of this process.

Because I recently heard about something called Tune.

I don't know enough about it yet, but apparently it's a process of cleaning up the input tokens.

So I wonder if working with various efficiency technologies or features, we can even see a more efficient output and things like that.

OK, here's the story out of South Korea that that dropped today.

And I thought it was interesting because it's a on a national effort.

So South Korea require advertisers to clearly label any ads created with artificial intelligence starting next year in response to a surge of misleading promotions that use AI manipulated images and videos.

Officials say many of these ads feature deepfakes, which are AI-generated images or video that makes real people appear to say or do things they never did.

The problem ranges from fake celebrities endorsing weight loss products to fabricated medical experts pushing unimproved treatments, often targeting older people who may struggle to spot AI-made content.