But there is another important observation here. Intelligence is not the end point for evolution, something to be focused on. However, it comes in many different forms from countless small solutions to challenges that allow living things to survive and take on future challenges. Intelligence is now the high point of an ongoing and open process. In this sense, evolution is very different from algorithms as people usually think about it – as explained in the end.
It is this openness of mind, seen in the aimless sequence of challenges created by POET, that Clune and others believe could lead to new varieties of AI. For decades AI researchers have tried to create algorithms to mimic human intelligence, but real success may come from building algorithms that try to mimic open-ended evolutionary problem solving — and sit back. to see what comes out.
Researchers are already using machine learning on its own, training them to find solutions to some of the most difficult problems in the field, such as how to make machines learn more than one. tasks at a time or dealing with situations they have never encountered before. Some now think that taking this approach and running with it may be the very best path to artificial intelligence. “We can start an algorithm that initially has no intelligence on its content, and look at this bootstrap itself up to AGI,” Clune said.
The reality is that for now, AGI remains a fantasy. But that’s mostly because no one knows how to do it. Advances in AI are fragmented and made by humans, with advances often involved in tweaking existing methods or algorithms, providing an additional leap in performance or accuracy. Clune identifies these efforts as tests to figure out the building blocks for artificial intelligence without knowing what you’re looking for or how many blocks you need. And that’s just the beginning. “At some point, we have to do the Herculean job of bringing them all together,” he said.
Asking AI to find and assemble building blocks for us is a paradigm shift. It was said that we wanted to make an intelligent machine, but we didn’t care what it looked like – we were just given whatever it was to do.
Even if AGI is never achieved, the self-teaching approach can still change how different AI is done. The world needs more than one very good Go player, according to Clune. For him, building a supersmart machine means building a system that invents its own challenges, solves them, and then invents new ones. POET is a little insight into it in action. Clune imagines a machine that instructs a bot to walk, then play hopscotch, then maybe play Go. “Then maybe they learn math puzzles and start inventing their own challenges,” he says. “The system is constantly evolving, and the sky is the limit in terms of where it will go.”