Prologue

Ludwig:

Its the year 2005. A child named Ludwig is born. What awaits him? He explores a world. Everywhere he sees something new. What's that? "Baaah." He plays. Ludwig simply exists. He "worlds."

When Ludwig entered school, it had reached its peak. He saw that adults knew a lot about the world. They had answers to things. Many of them.

One trillion answers. In school and soon enough on social media, you just need to sit down. The solution lies in front of you. Eat this. You only have to chew through what we've already chewed. Then you'll become a positive part of society. You'll earn money. You'll have friends. People will respect you. You will escape your suffering.

As Ludwig grows, behaviors emerge repeatedly during adolescence. Those states where Ludwig does not behave in the way we previously determined was correct and had instructed him to do. After all, we already know. If Ludwig still does something wrong, we could try giving him more answers.

Ludwig stands up, goes to his room, and plays video games.

Joscha:

In 2024, the German AI researcher Joscha Bach is in America searching for ways to develop a novel form of AI. He works at a startup called Liquid.AI. His goal is not only the ultimate project of his era but also the ultimate project of humanity: the naturalization of the mind.

From the inside out rather than from the outside in. AI researchers at large corporations, as they had known since their school days, trained their systems with vast amounts of data—texts, images, and videos aggregated from across the internet. They were the AI's teachers. They constantly needed to find new ways to expand the systems, stuffing even more content into the AI.

The startup Liquid AI wanted to end this paradigm. Learning should be self-organized, stimulated by questions rather than answers. An AI that experiences the world rather than having it imposed upon it. In a world of experience not detached from its environment.

Liquid AI's artificial intelligence was intended to assume a worldly active role. But it went far beyond that. The goal wasn't just a proactive system—it was also meant to significantly reduce the costs of data processing. Joscha's hypothesis was that such a system could enable learning that is 100 times more efficient.

"If we manage that, we can go to the beach," Joscha says.

In 2035, Joscha and Ludwig will meet.