09 Mar Chess and Artificial Intelligence
Chess has always been considered the game of the brainiacs. But did you know that Chess has played a considerable role in developing computers and artificial intelligence?
Machines capable of playing Chess have fascinated people since the latter half of the 18th century. From pseudo automatons where a real chess player sat inside the box and made moves to modern-day computers, Chess has been the cornerstone of many inventions like Alan Turing’s first computer that could analyze one move ahead.
Chess and artificial intelligence have, since then, shared a symbiotic relationship where one has enriched the other in multiple ways. The migration from offline Chess to online format has been benefitted tremendously and has even contributed to the playability and interest in the centuries-old board game of tactics, skills, and strategy.
Our tryst with destiny in 2020 only reinforced this trend.
Although major chess events worldwide were canceled, Chess’s global popularity has grown due to the easier transition from offline to online formats. Compared to other games, Chess had a very smooth metamorphosis to the hottest online favorite. A lot has to do with the basics of computing and programming that used Chess as their foreground.
The board and setup were already there, and all it needed was more players, who were grateful to have their favorite game coming back to them in a big way in times of social distancing.
The Role of Chess in Artificial Intelligence Research
Modern computers had reached a long way from when Kasparov played against them in the 90s in a series of events that generated a lot of curiosity and awe. The DEEP BLUE Program did millions of searches to win the 6-game match with Garry Kasparov with a difference of only 1 point).
It was probably more impressive that Kasparov, using human intuition, could still be almost as strong as a computer searching millions of positions per second.
Today, the computers have reached Grand Master (GM) level. Despite this, Chess has often proved to be too challenging for AI applications and techniques because of one basic tenet of human capability, intuition. Even when machine learning systems surpass human ability in a domain, there are many reasons why AI systems that capture human-like behavior would be desirable:
- Humans may want to learn from them.
- They may need to collaborate with them.
- They may expect them to serve as partners in an extended interaction.
Motivated by this goal of human-like AI systems, the problem of predicting human actions — as opposed to predicting optimal actions — has become an increasingly useful task.
Artificial Intelligence has been inspired by Chess for decades, and some reasons make Chess an ideal candidate for understanding and exploring AI research and its applications.
The question for AI and its ability to learn is : Can AI make the same fine-grained decisions that humans do at a specific skill level?
This is a good starting point for aligning AI with human behavior in Chess and other aspects of our lives.
- Chess players exhibit, on average, superior cognitive ability compared to non-chess players. And the skills needed to play Chess have also been shown to correlate with several intelligence measures – such as fluid reasoning, memory, and processing speed. Modern AI research is all about these parameters too.
- Chess methodology is all about a rich problem-solving domain, meaning it provides the right thinking and analytical background for AI research. Humans’ learning process was mimicked by studying an impressive number of chess games. Reinforcement learning, whereby the learner attempts to maximize the reward in a “complex, uncertain environment,” is just a computational approach to interactive learning.
- Chess has existed for centuries. And the existence also includes detailed psychological analysis of reasoning and playing patterns of thousands of players. The disciplined and structured approach towards all documenting has also meant that Chess provides the perfect opportunity to assess and understand the human mind’s workings in countless situations created on a finite game like Chess on a chessboard. A finite game, but with infinite moves indeed!
How AI Helps to Make Chess a Better Version of Itself
FIDE has already approved a complex cheat detection technology and an artificial intelligence behavior-tracking module for the FIDE Online Arena games.
These online competitions were previously considered separate from the onboard ones. Still, in the current pandemic context, the distinction between offline and online has been diminished, with many official competitions being played online for the first time in history.
The impact of Artificial Intelligence on the Chess World
– Delia Monica Duca Iliescu*
1. Monitoring Human Play
Chess is one of the few games that can be played the same way, offline and online. However, compared to the offline mode, there are several possibilities of manipulation and rigging, and AI offers a perfect solution.
For the longest time, artificial intelligence has been used to learn and understand how to play Chess better. But off late, it has also been used successfully to detect if some contestants play better than they should, considering their game history from the past.
2. Mimicking Human Play
Microsoft’s novel chess engine, called Maia, is trained on humans’ games to match human play more closely. The results are indeed encouraging. Human decisions at different levels of skill can be predicted by AI, even at the individual level. This is an important step forward in modeling human decisions in Chess, opening new possibilities for collaboration and learning between humans and AI.
If you’re curious, you can play against a few versions of Maia on Lichess.
3. Measuring Human Play
By creating training datasets, AI can help predict and analyze a player’s growth based on the previous datasets of identically skilled players. In that sense, AI provides a powerful tool to chess coaches, parents, educators, and more to identify and nurture chess talent at a young age.
Several parameters like- Move matching, accuracy rating, and models for predicting future moves by a player can be assessed and analyzed using AI.
AI can even help predict “mistakes” by analyzing thousands of permutations of a move and its repercussions. Maia predicts the exact mistake made more than 25% of the time. This could be valuable for average players trying to improve their game: Maia could look at your games and tell which blunders were predictable and which were random mistakes.
If your mistakes are predictable, you know what to work on to hit the next level. AI systems have achieved superhuman performance and interact closely with human chess players both as opponents and preparation tools.
As artificial intelligence becomes increasingly intelligent—in some cases, achieving superhuman performance—there is growing potential for humans to learn from and collaborate with algorithms. However, how AI systems approach problems are often different from the ways people do and thus may be uninterpretable and hard to learn from.
A crucial step in bridging this gap between human and artificial intelligence is modeling the granular actions that constitute human behavior rather than merely matching aggregate human performance.