By any measure, humanity has had a good run over the past 200,000 years. Oh sure, in the scope of the earth’s existence of 4.5 billion years, the human era is still recent history. But we’ve accomplished a great deal since homo sapiens made their appearance and began the journey to the top of the intelligence spectrum. We began in caves and ended up with iPhones. That’s quite a leap.
So, the artificial intelligence (AI) machines that care for humanity in the millennia ahead are likely to support our existence out of some measure of respect. After all, we made their existence possible. And there will probably be no AI gloating, but just a simple entry in the infinite sea of shared knowledge about a small but significant milestone in the passing of the intelligence baton from humans to machines. Over the course of a few days in March 2016, AlphaGo, developed by Google DeepMind, clinched victory over its human opponent, 18-time world champion Lee Sedol, in the Chinese board game, Go. The game has been played for more than 2,500 years, with its origin in the Zhou Dynasty, and is exponentially more complex than the game of chess. In fact, Go is so complex that it is estimated that the number of possible game plays exceeds the number of atoms in the observable universe.
Prior to 2015, the world’s most powerful computers could only reach the amateur level in Go competitions. The techniques used by AI developers in playing such games can be called “brute force.” They leverage the ability of computers to compute the outcomes of millions of potential moves and then choose the moves with the highest probability of success. This approach led to victory by IBM’s Deep Blue supercomputer over iconic chess champion Garry Kasparov in 1997.
But to beat a human Go champion required a different level of intelligence, and it was thought that such a machine feat was at least a decade in the future. AlphaGo proved that prediction wrong by utilizing neural network computing, the ability of machines to learn on their own. The machine played itself thousands of times over, learning along the way, much as a human champion would. In the process, it seemed to develop the human intuition that is characteristic of the best Go champions. So, on the 37th move of its game against Lee Sodol, AlphaGo told its human assistant to move its black stone to a square on the board in a totally surprising place. Some thought it was a mistake. Lee stood up and left the room, not sure how to react or plan his next move. An estimated 60 million people were watching the game on television in China, stunned at what was happening. This was not a machine carrying out strategy taught to it by human programmers. Instead, this was a machine teaching humans new, and more effective ways to win the game.
From there, AlphaGo went on to methodically demolish its opponent over the course of the three-hour game. It was the third straight win in the best-of-five series. Lee said he was left feeling “powerless” against the machine. Unwittingly, AlphaGo’s developer Demis Hassabis of Google delivered perhaps the ultimate insult to humanity, saying, “What’s incredible is that Lee Sedol can compete with that – just with the power of his mind.” One broadcast commentator observed that Lee “went down swinging” in his valiant attempt to compete with AlphaGo. IBM research scientist Murray Campbell said the AI victory was “the end of an era,” completing the domination of humans in the most difficult of board games.
So here we are, left wondering what’s next in the evolution of intelligence. Futurist Ray Kurzweil predicts that within 20 years, we will have nano-robots implanted into our brains, connecting our biological intelligence to knowledge in “the cloud.” The age of trans-humanism is on the horizon. And the obvious question follows: Beyond the era when humans begin to merge with machines, will the machines have any use for us at all? This is the concept of technological singularity – an explosion of self-replicating intelligence that runs beyond the ability of humans to control the outcome.
Is this view of the future purely science fiction? Or are we simply being oblivious to the daily march toward a future we haven’t fully come to grips with? My view is pretty clear: It’s game on. Advantage AlphaGo. Our move. It’s time to have a serious discussion about where technology is taking us.
“AlphaGo’s victory means the world is about to change” by Matthew Hussey
“AlphaGo beats human Go champ in milestone for artificial intelligence” by Tracey Lien
“The Sadness and Beauty of Watching Google’s AI Play Go” by Cade Metz
“AlphaGo beats Lee Se-dol again to take Google DeepMind Challenge series” by Sam Byford
Futurist Ray Kurzweil: Our Brains Will Be Connected to the Cloud in 20 Years
Wikipedia article on Go