Hobbies & Leisure
23 h

Did Pluto AI prove decades of StarCraft pros used the wrong builds?

It has basically proven that you should use marines and wraiths and sometiems Ghosts in Terran vs Terran and Terran vs Protoss, and it has absolutely proven that you should NEVER fast expand before teching up, but it also proved you do not need 3 workers on gas to tech up fast enough to BEAT the best possible 2 base fast expand build. You don't even need but 1 worker on gas to build a factory, starport, and vultures and wraiths and first dropship fast enough to harass the 2 base player to death.

I've been telling Korean pro gamers for decades that you are not supposed to be massing 50 siege tanks per game, and that absolutely CANNOT be the correct build for either game. Tehy kept right on doing the wrong thing anyway.

The AI spends less than half it's income on tanks, and spends most of its income on Marines, vultures and Wraiths (aircraft).

Then the Korean Grand Master/Code S player rage quits because he does not understand that 10 Siege tanks are not supposed to beat air units anyway. He didn't understand the Bot literally OUT THOUGHT HIM in addition to the fact the Bot can now play at around 8000 actions per minute.

I've been saying it for 20+ years for SC1. The only thing that made Koreans better at the game than me is they could hit the buttons faster than me. As far as "what beats what and when to do something," I was always better than them, and now the bot has proven my strategies work and the Korean Pro gamer's strategies never worked.

Updates
23 h
The thing that scares me is if the Bot is smart enough to beat Grand Masters at StarCraft, then it is also smart enough to beat real Generals at real warfare. StarCraft Grand Masters usually have IQ scores in excess of 145, and some of them are pushing 200. In contrast, real-world generals are actually politicians and often don't even understand how their own technology works.
Did Pluto AI prove decades of StarCraft pros used the wrong builds?
Post Opinion