Now that the likes of KataGo is superhuman and probably near AlphaGoMaster/AlphaGoZero strength, is there a way to generate 'beautiful' Go games that have it playing against itself (but with some random variation and at longer intervals than the training games for deeper play) and then once a whole bunch are generated maybe an AI of some sort to evaluate the 'elegance/beautifulness' of each game for the purpose of picking out in a more scaleable /automated fashion a handful of "perfect" Go games played at the highest level by AI etc to showcase?
Now that the likes of KataGo is superhuman and probably near AlphaGoMaster/AlphaGoZero strength, is there a way to generate 'beautiful' Go games that have it playing against itself (but with some random variation and at longer intervals than the training games for deeper play) and then once a whole bunch are generated maybe an AI of some sort to evaluate the 'elegance/beautifulness' of each game for the purpose of picking out in a more scaleable /automated fashion a handful of "perfect" Go games played at the highest level by AI etc to showcase?