In May this year Google’s AlphaGo AI defeated the world’s best Go player. Go is an ultra-complex strategy game invented in China 2500 years ago.
Did you catch your breath when you heard the news?
Did your heart skip a few beats?
If not, you didn’t comprehend the magnitude of the announcement.
Let me explain.
Go is a hugely complex game. Google’s DeepMind artificial intelligence team that developed the winning algorithm likes to say that there are more possible Go board positions than atoms in the known universe. It is impossible to visualize every possible move in order to determine the next best move to make. Humans learn the game by intuition, be feeling what the best move would be. We develop these skills by observation and practice.
How was AlphaGo able to defeat Go world champion Ke Jie? Observation and practice with the help of Big data and Deep learning.
‘Big data’ are two short words for something really humongous: all the information the world is generating second by second. And that’s a lot.
Every day we create 2.5 quintillion bytes of data – that would fill 10 million blue-ray discs, which if stacked, would reach the height of four Eifel Towers stacked on top of one another.
With the advent of the Internet of Things tracking increasing numbers sensors big data is growing exponentially.
What does huge sets of data have to do with an AI algorithm that learned to play the most complicated game created by man?
The AI was trained by feeding it data. In this case, the data consisted of 30 million moves from games played by human experts.
Algorithms are used for calculation and data processing and now it can be used for automated reasoning. Some algorithms are designed to allow computers to learn on their own.
It’s called deep learning.
The power of deep learning lies in using massive amounts of data to teach computers what humans know without giving the computers explicit instructions. Instead of telling a computer what a tree is, we feed it millions of images of trees already on the Internet. Take into account that a computer can sort through vast amounts of data in seconds.
Do you now see what you missed when you thought the news was just about a game won by a clever computer?
The combination of big data and deep learning has led to mind-blowing achievements in a short time. Baidu’s AI research lab in Silicon Valley developed Deep Voice, an advanced speech synthesis program that taught itself to talk in just a few hours. Google’s deep learning algorithms have learned to diagnose some medical conditions more accurately than doctors. The list goes on.
We need to understand what exponential technological advancement means for the job market. The fact that computers can learn means that that they teach themselves to do our jobs. If machines are going to do the work, we need to find some other way to afford all the wonderful ways technology is promising to improve our lives.
Think about that.