My method of studying is to take two go-playing AI of different strengths and pit them against each other. My goal is NOT to try and mimic the complexity of AI play. My goal is to try and train the pattern-detecting parts of my brain in the many strategies and priorities that Go embodies.
So, for the stronger AI I usually use the free desktop version of Leela, (download here: https://sjeng.org/leela.html) because
- it’s free
- it’s easy to use out of the box
- it has incredibly useful heatmaps that show the range of possible “good” moves in situations where more than one move is available
For my weaker AI I usually use COSUMI
https://www.cosumi.net/en/
Or this weaker online version of Leela (weaker because it uses fewer play-outs than the desktop version has at its disposal)
http://leela-one-playout.herokuapp.com/
So, the way I do it is, I’ll have two windows open on my computer running the two programs (i.e. desktop leela and a browser window with one of the browser-based AI). As each responds, I feed one’s moves into the other window so that both are the same.
In the desktop version of Leela, there’s no way to disable auto-respond, so what I will do is try not to look where it went, take it a move back, and turn off the heatmaps. I start with doing my own analysis of that position:
- who has sente/gote?
- which are the weakest groups - which are the strongest groups?
- is there a crucial/necessary move to make in the current fight, or is it better to tenuki?
- is it time for a cautious/solid move? is it time for a big move?
Once I’ve done my own analysis, I will force myself to pick 2 or 3 options for what I think the best move is. THEN I turn on the heatmaps, and see if I guessed correctly. If Leela suggests a completely different move, I then take the time to figure out why that particular strategy is smarter / more efficient than the moves I thought of.
If there are multiple moves on the heatmap, I then make the choice between them (or, if I’m really stumped, I force Leela to pick the next move) and keep playing.
I find this really useful because - one of the hardest things for me as a beginner was to start absorbing how a more advanced player sees the game. While I still have a long way to go in this regard, at least I have a consistent guide shining a light on those priorities, and the more I train myself to look for those things, the easier it becomes for me to recognize the correct strategy at each point, or at least be aware that there are multiple strategies available, and be able to chose between them in an informed manner.