AI is learning how to create itself

But there is another crucial observation here. The mind was never a goal for evolution, something it had to be a goal. Instead, it was created in different forms of small and numerous solutions to challenges that allowed living things to survive and take on future challenges. Intelligence is the culmination of an ongoing and open process. In that sense, evolution is quite different from algorithms, as people usually think about them – as a goal.

It’s an open idea, seen in a seemingly unobtrusive sequence of challenges posed by POETs, that Clun and others believe could create new types of AI. For decades, AI researchers have tried to build algorithms to mimic the human mind, but the real breakthroughs will come from building algorithms that try to mimic an open solution to the problem of evolution and sit back and watch what arises.

Researchers are already using machine learning in itself, training them to find solutions to some of the most difficult problems in the field, such as how to make machines to deal with situations they may learn more than one task at a time or have not encountered before. Some believe that taking this approach and running with it may be the best path to general artificial intelligence. “We could initially launch an algorithm that doesn’t have a lot of intelligence, and see it potentially launch up to AGI,” Clun says.

In fact, for now, AGI remains a fantasy. But that’s largely because no one knows how to do it. The advancement of AI is done in installments and is carried out by humans, with advances usually involving changes in existing techniques or algorithms, giving incremental leaps in performance or accuracy. Clune tries to find these building blocks of artificial intelligence without knowing what you are looking for or how many blocks you will need. And that’s just the beginning. “At some point, we have to take on the Herculean task of uniting them all,” he says.

Asking AI for us to look for and assemble these constructions is a paradigm shift. He’s saying we want to create a smart machine, but no matter what it might be, give us anything.

Even if AGI is never achieved, the self-teaching approach can change what kind of AI is generated. Clun says the world needs more than a very good Go player. For him, creating a super-intelligent machine means building a system that invents its own challenges, solves them and then invents new ones. The POET is a small view of this action. Clun imagines a machine that teaches him how to walk a bot, then play with the grasshopper, and maybe play Go. “Then maybe he learns math puzzles and starts guessing his challenges,” he says. “The system is constantly innovating, and the sky is the limit as to where it can go.”