Today, AI is helping us change lives in almost every sphere. Chess is just one of them.

On the occasion of 20th July, which is the International Chess Day, we would like to enlighten you on the role of AI in chess.

There will be many things you’ll learn here, starting from how AI originated to how it’s being used at the top level today. Also a lot of people don’t know the difference between an AI engine and a traditional chess engine. We’ve covered it here

Before starting, would you like to take a guess how much time it took for an AI engine to learn chess and beat the best in the field?

Keep reading to find out.

Let’s start with the origins first.

It can be traced to the time when computers started playing chess.

The legendary Garry Kasprov believed computers could never beat strong grandmasters. In 1996, Kasparov proved this by beating IBM’s chess playing computer Deep Blue 4-2 in their match. However, Kasparov was beaten by the same computer(2.5-3.5) in 1997, a year later.Since then, computers have only become stronger.

Today, the strongest computer would crush the best player in the world, even by starting the game with a pawn down.

But in recent times, the strongest computer was beaten by the strongest AI machine – Alphazero. Wait, so is Computer chess and Artificial Intelligence different from each other?

Yes!

Note – To keep things clear, computer chess, computer engine, normal engine and traditional chess engines are the same and used interchangeably throughout the article.

A traditional engine is coded and ‘taught’ how to play the game, on what parameters to evaluate the position etc. Based on the parameters, it generates millions of options and chooses the best out of that. It’s believed that the traditional engine Stockfish can evaluate over 60 million positions(!) in a second and make the best decision out of that.

The quality of the decision depends on how well it’s programmed. That’s where the skill of the team behind the engine lies.

Examples of traditional engines include – Stockfish, Komodo

On the other hand, AI is coded such that it learns chess on it’s own. It’s designed to think like a human. For this purpose, AI classifies information using Neural Networks. These are networks which store information like a human would. While it cannot evaluate as many moves ahead as Stockfish, AI has the intelligence of a human. Based on that, it’s capable of making better decisions.

Basically a normal engine is like the old Nokia phone. It cannot set up a meeting if you speak to it. On the other hand, the AI engine is like an Android smartphone. It has learned how to operate the mobile, and you set a meeting by just telling it.

Example – Alphazero is an AI engine which runs on Machine Learning algorithms.

AlphaZero is a product of Google’s subsidiary DeepMind, which develops AI engines for video games. It learned chess by playing random games against its own and improving from there onwards. In early 2018, Alpha Zero beat Stockfish, the strongest chess engine at the time by 72-28 in a 100 game match.

Can you guess how much time it needed to learn chess and beat the strongest engine?

4 months?

4 days?

Nope!

It took only 4 hours!

To make things clear, it started learning from scratch and within 4 hours was capable of beating the strongest player or engine in the world.

No, it won’t.

AI is only a machine. We have to come to terms with the fact that a machine is going to be better than us, just like calculators are better at mathematics. Human competitions and tournaments will continue to happen.

AI is being used by top players for 2 major reasons –

• To find powerful opening ideas.
• To analyze the game from the AI’s perspective and find flaws in their own thinking process.

Even chess programmers are using AI and integrating it within their traditional engines. Stockfish, Komodo and Leela Chess are some of the top engines that are adapting to this model.

Remember –

We aren’t competing with the machines. They help us get better at the game, not rob us from enjoying it.