Definitely not the end of AI technology. Even if the PPO method you are using platforms there will be big improvements possible in other areas. Does the algorithm openai 5 uses have a sense of what happened in the recent past, or is it acting just on the current state?

Fantastic achievement! Are you able to provide more details re. expected shapes of OpenAI Five's learning curves over the coming months? Current thoughts on theoretical limits to the system?

Be cool if you let both teams pick and let the AI choose highest available win probability like last time :)

The results of the last two days do not justify the sheer awesomeness of what your team managed to achieve over the past year. OpenAI did not lose if you ask me. If anything it showcased the immense potential of what it can truly be in a few years. This is just the start.

Besides, considering the short timeframe you've had to make changes, adjustments and re-training, I think this is super impressive. Good luck with the future endeavours of OpenAI - hope that we will get to AGI in a safe way. :)

The AI bots are mechanically strong but they lack the element of surprise that the human teams have. the games seemed over some time before they actually ended since the humans knew that the bots were not going to do anything drastically different. Good luck for the future!

From what I felt from the last OpenAI give games is that the bot have a hard time being in unknown situations. Maybe the way they learn make their scope limited in some types of game. What about training them starting in a randomised situation.

Will you be doing a blog post analysis like the other games?