Very cool and very interesting. I highly recommend people take a look at this thread.

I'm sorry but, "DiL-piKL?"

I may have misunderstood what he meant but Dota certainly requires a lot of cooperation to win, it's not a 2 player zero sum game.

OpenAI's dota bots were really crazy at the time and cooperated really well, I would love to see them go at it again with today's scaling and advances in algorithms

Here's the paper for this.

Were OpenAI‘s dota bots cooperating with human players or only playing against them? That might be the difference.

Correct, DOTA bots played against humans, not with them as far as I know. This shows that AI can now play cooperative games along with and against human players at expert levels.

Maybe it's the kid in me but I really wish we could actually witness it playing these games

It did include cooperation with humans https://openai.com/blog/openai-five-defeats-dota-2-world-champions/#cooperativemode

It already had some cooperation, but maybe not on the same level as this new paper https://openai.com/blog/openai-five-defeats-dota-2-world-champions/#cooperativemode

Oh, that is neat. Though this is a little bit different and of much greater importance than that cooperation.

Not just you, I want a whole eSports league where AI companies battle their fine tuned models against each other.

Does anyone know who's working on a version that can do communication too? The press and negotiation is a huge part of Diplomacy and it seems like the obvious next goal. Obviously it's much more complicated but with all the insanity text models are capable of now it doesn't seem unimaginable.

imagine a standoff between two superoptimized AI's playing Risk or something like that and they employ some ultra gigabrain strats and nobody has any clue what they AI is even planning.