AlphaGo vs Lee Sedol : A Lesson in Making

Most things in life changed when DeepMind, a startup acquired by Google, designed and created a machine that can ``study´´ and ``think´´ as a human in better, faster way.

AlphaGo, DeepMind’s first-born, is the first ever artificial intelligence machine that has been able to beat the World’s champion in Go, an asian version of Chess considered one of the most difficult strategic games. Lee Sedol, the Go connoisseur, lost not only once but four times.

This competition took place on March, 19th 2016. Lee Seldon sat down against the machine and its programmer, who was in charge of moving the pieces on the board as the machine dictated (unfortunately there wasn’t enough time to include some robotic arms to make the scene even more dramatic). The event was perceived a big deal and everybody was betting on Lee. The competition was composed of 5 games. From minute 1 Lee knew that any game against AlphaGo wouldn’t be easy to win. Lee was desperate, he claimed he felt devastated and apologized to the world after losing three straight games for not living up to their expectations. The following night, Lee stayed up until 6 a.m. with other advisors trying to brainstorm different strategies in order to have as many options as possible to beat the machine in the 4th round. Exhausted but excited, he confronted AlphaGo again and as you may have guessed it, this time he won. And in that moment everyone present cheered, people had the feeling there was still hope for human brains to continue being a step ahead of technology. However, nothing could be more untrue. Lee was able to beat the machine with an extraordinary strategy called ``Hand of God´´. At that point, AlphaGo couldn’t find anymore solutions that would increase its points, which is why it lost. The fact remains that he was able to beat it because the program didn’t have that move registered, therefore it couldn’t figure  a solution for the said move. It just took one slight change in the database of AlphaGo to save that strategy and have it ``in stock´´ for future battles, in this example the 5th round, which Lee lost.

Do you not wish you were able to register anomalies at this speed and adapt at this pace? It’s only natural to be envious of this machine.

The main advantage of machines over humans is that they never sleep. They don’t get tired. They never need a beer with their robot friends to clear their mind. They never feel their brain isn’t working for the day or that they need a vacation. They can study 24/7/52 and are able to accumulate all human knowledge about, well Go for instance, in just one day. Au contraire, it took us thousands of years to study our own past.

If asked what you think is the only thing machines/robots can’t do as opposed to humans what would you say? A lot of specialists in the robotic field think that the main trait that differentiates machines from humans is curiosity, the aim of discover and create new things, the interest in improving. Until now, most of the machines that have been created were only able to perform specific tasks that were previously programmed on their system, and they couldn’t do much more further than that. Curiosity is humanity’s calling card, and it is that same curiosity which has led us to the competitive tension with machines. As long as we remain curious, we have a shot at remaining on top.