Artificial Intelligence is a hot topic in today’s day and age. This technology is quickly taking over entire industries, from transportation to education. And esports is one of the best testing grounds for big AI companies such as OpenAI and DeepMind.
Some of AI’s past achievements
AI first scared the world in 1997, when Garry Kasparov, the world chess champion at the time, was defeated by Deep Blue in a Bo6 match. As you might imagine, over the last 24 years, AI and the computer hardware on which it runs have been improved to such a degree that it is no longer a question of “if AI will conquer many of the human activities” but rather a question of “when?”.
Around 2014 – 2015, DeepMind developed AlphaGo, a computer program that can play the board game known as Go. But unlike its predecessors, it used a revolutionary method of mastering the game. Instead of being spoon-fed by its creators, it was instructed to play against itself for the equivalent of thousands of lifetimes in a short period of time.
The result was absolutely scary. In 2016, AlphaGo demolished the most accomplished Go player in the world, Lee Sedol, and only lost one game out of 5. But again, as you can imagine, 5 years is a long time in AI research and today’s systems are even stronger.
The same type of computer program was then developed for chess. It’s called AlphaZero and it destroys everyone with ease.
AI’s esports achievements
In 2019, Deep Mind introduced another one of their creations, this time to the world of esports and in particular, StarCraft 2. It was called AlphaStar and it attained Grandmaster level in less than 8 months after it started competing. A little later, AlphaStar was tested against Serral, the best SC2 player in the world. The 5-game series ended with a score of 4 – 1 in favor of the AI program, which is a scary thought if you know just how sophisticated the game is.
The implication here is that Deep Mind could create AI software that can master highly complex game-like areas and end up being better than any human being. Imagine what the implications would be for numerous industries that are currently hiring professionals to do those jobs.
Another AI project, called OpenAI Five, was developed by a company co-founded by Elon Musk in 2015. It consisted of a team of video game bots that mastered Dota 2, a popular esport that many people play, watch, and bet on, to such an extent that in 2017, at the game’s world championship (called The International), the program was tested in 1v1 matches against the best players in the world. And it crushed them.
A few years later, in 2019, OpenAI Five played a match against the team that had won The International 2018 and was going to win The International 2019 as well. The games were so one-sided that people were absolutely shocked. And keep in mind, Dota 2 is far more complex than anything we’ve seen, not just in esports but games in general.
To get an idea of how sophisticated this esport really is, just consider the fact that there are over 120 heroes that you can pick from at the start of the game and over 200 items that can be bought during its duration to enhance their stats and abilities. Each hero has at least 4 unique abilities, which means that in every Dota 2 match, you have a composition of 20 abilities competing against a composition of 20 other abilities. And that’s without taking into account all of the other aspects of the game.
An AI that can master this complex arena provided by MOBA games will be able to master a lot of other arenas of similar complexity in very little time. And the implications are very serious for everyone. But let’s stick to gaming and esports. Can Artificial Intelligence conquer this industry? Well, in many ways, it already has.
Of course, there are still many games out there that have not yet been touched by AI. But given the fact that their complexity is lower than that of a MOBA game like Dota 2, it’s quite hard to believe that the result will be different.