Not that long ago people would have said chess was not going to be solved anytime soon even given the exponential rise of machine intelligence. I guess those people figured machines would always continue down the bruteforce path that programs like Stockfish (that calculates 70 million moves a second) use instead of thinking like humans and discriminating between logical moves. I think with AlphaZero it's safe to say that chess is a solved game. Thoughts?

they forgot ibm's deep blue which beat chess worldchampion.

A sufficiently powerful computer will beat any human at any full-knowledge game. Something where they struggle is a game like Starcraft, where you face a lot of unknowns when it comes to your enemy. To this day, nobody has been able to even come close to the best players in Starcraft, and they have been trying for decades at this point.

chess etc are easy to teach to computer but rts-games are different thing.

DeepMind is about to take on Starcraft. My money is on DeepMind by this time next year.

OpenAI has beat the best DOTA players in single lanes, they're building on it for a soon to come all lane matchup.

Marshal,
DeepBlue lost to Kasparov. He won 4-2. That was obviously before AI got really good and now players would get obliterated by even the weakest AI systems in place as they've evolved since then, but what makes AlphaZero different is that it wasn't playing against players... it was playing against the best computer out there, and completely, utterly destroyed it despite having only a few hours of training against itself and analysing 1/1000th of the moves that StockFish did.

I would put more money against it. Having played DOTA a little and Starcraft a lot, Starcraft has more advanced micro and macro. It is a far more complex game. There are 50-100 types of units, 50-70 types of buildings, and about as many different things to research among the tech tree. I don't think DOTA comes close to this complexity. On top of this, I find the micro of Starcraft far more difficult, and also the part that I expect computers to take more time to figure out. Moving units properly for attack in exactly the right way is the most important thing in winning in battle.

To balance this all out in order for a computer or AI to beat a good human, I think, will take at least several years, and a lot of spoon feeding the computers to achieve.

Then he lost in the rematch a year later. 3.5-2.5