But just to understand, what does small mean? What can you do on small or what could you do on

small? So back to the first models, you know, I was super happy if we could get it to generate a coherent sentence.

That's the level to begin with, right? Then as the models became a little bigger and our skills in training the models on data also became better, we could, of course, get it to have a conversation where it was kind of like...

To the point, some of the time, and also going a little round on about half, the other half of the time, right? But still at a level where we could create the model or the AI in a way where it could engage with human beings in the artworks.

But then what happened was that... that basically the rest of the world moved into cloud.

And in most of our projects, we stayed locally for different reasons.

First of all, because we really, you know, we're super deep into the vectors of the neural network.

good thing? I mean, just because it's so damaging towards the planet, isn't that a good thing?