Those artificial intelligence agents learned everything by themselves, exploring and experimenting on the complex Dota playing field at a learning rate of 180 years per day
Source: The OpenAI Dota 2 bots just defeated a team of former pros – The Verge
Yet another important idea that’s often lost in the noise of reporting on AI; machine learning takes place at a rate that we can’t match. And as computation gets more powerful at marginal increases in cost, machine learning will continue accelerating. In 4 months the AI bots developed to the point where they were beating human players who had many years of experience at the professional level.
The developers of OpenAI noted that the OpenAI Five were losing to amateur players within their team back in May. By June, the AI had matured enough to defeat casual players, and today, it’s shown that it’s capable of overwhelming people who’ve been playing Dota 2 literally since its inception.
Another thing to take note of; Dota 2 is a complex strategy game that requires skill in planning, navigation, and decision-making. Even if algorithms do nothing but continue advancing in terms of efficiency and brute force computation, it’s obvious that they’re going to outpace human beings in every task that requires what we call “intelligence”.
Update (28-08-18): When OpenAI put their bots up against a team of professional Dota2 players, it became clear that humans still have the upper hand. See also this post on the more technical aspects of how the OpenAI bots play.