This article is from Quanta Magazine and it is written by Mathew Hutson and he writes about Kenneth Stanley, a pioneer in nueroevolution which uses the basics of biological evolution to create smarter computer algorithms.
Stanley and his students back in 2007 created a game called Picbreeder, in which “each image was the output of a computational system similar to a neural network. When an image spawned, its underlying network mutated into 15 slightly different variations, each of which contributed a new image.” What is surprising is that, “Stanley didn’t intend for Picbreeder to generate anything in particular. He merely had a hunch that he, or the public, might learn something about evolution, or about artificial intelligence.”
One day Stanley saw a pictured that looked like an alien face and went on to evolve it into something resembling a car by moving the eyes closer together. He realized that if the picture at first resembled nothing in particular then he would never have changed into something else like a car and that when he made the connection to how we solve everyday problems. “It had a huge impact on my whole life,” he said. He looked at other interesting images that had emerged on Picbreeder, traced their lineages, and realized that nearly all of them had evolved by way of something that looked completely different. “Once I saw the evidence for that, I was just blown away.”
Stanley’s realization led to what he calls the “steppingstone principle” — and, with it, a way of designing algorithms that more fully embraces the endlessly creative potential of biological evolution. In other words, this approach tries to mimic the outcome of evolution using an evolutionary algorithm to output better results over time. A parallel can be seen here with Darwinian evolution and the way it allows for descent with modification through organisms.
The steppingstone principle goes beyond traditional “evolutionary” approaches. Instead of optimizing for a specific goal, it embraces “creative exploration” of all possible solutions. Earlier this year, one system based on the steppingstone principle mastered two video games that had stumped popular machine learning methods. And in a paper published last week in Nature, DeepMind — the artificial intelligence company that pioneered the use of deep learning for problems such as the game of Go — reported success in combining” deep learning” with the evolution of a diverse population of solutions.
The steppingstone’s potential can be seen by analogy with biological evolution. In nature, the tree of life has no endgame, in essence the goal of evolution a sort of trial and error when it comes to the species that have lived on this planet. This groundbreaking study of different algorithms has the chance to do something as incredible, but again such computing needs a lot of power and much more time to develop compared to the four billion year history of evolution.
“Biological evolution is also the only system to produce human intelligence, which is the ultimate dream of many AI researchers. Because of biology’s track record, Stanley and others have come to believe that if we want algorithms that can navigate the physical and social world as easily as we can — or better! — we need to imitate nature’s tactics. Instead of hard-coding the rules of reasoning, or having computers learn to score highly on specific performance metrics, they argue, we must let a population of solutions blossom. Make them prioritize novelty or interestingness instead of the ability to walk or talk. They may discover an indirect path, a set of steppingstones, and wind up walking and talking better than if they’d sought those skills directly.“Mathew Hutson
Evolution “invented sight, it invented photosynthesis, it invented human-level intelligence, it invented all of it, and all of it in one run of an algorithm,” Stanley said. “To capture one tiny iota of that process I think could be incredibly powerful.”
I recommend the article to anyone interested in the way computers are shaping our understanding and found the analogy with evolution to be very interesting. I am a biology major and currently taking a class in biological evolution, but this idea of computers converging to a human like intelligence is fascinating, but also terrifying to think about.