Hi everyone!

I've found a fun method to sharpen my skills, particularly in the challenging midgame phase. While I am able frome time to time to hold my own during openings and endgames against formidable human or AI opponents (with adequate handicap ), the midgame poses a significant challenge for me. I tend to lose about a point every move during this phase against strong players.

To address this, I've devised a new training routine. After each game, where I faced a strong opponent, I meticulously review the midgame. I analyze each minor loss or suboptimal move and try to encapsulate the essence of the mistake or the lesson learned in a single sentence. I hope this helps me to pinpoint exactly where and why I faltered.

Next comes the fun part! I use a large language model to evaluate my insights. While I'm totally not convinced about the effectiveness of this method in drastically improving my skills, I find it incredibly enjoyable and a unique way to engage with the game and AI.

I'm curious to know if anyone else has tried a similar approach or if you have other fun methods for tackling midgame weaknesses. It's always great to learn from fellow enthusiasts and perhaps add a new dimension to our training routines!

Looking forward to your thoughts and experiences!

Example result:

"Prioritize exploiting your opponent's vulnerabilities."
Rating: 9/10. Targeting weaknesses is a fundamental strategy in Baduk, allowing for effective control and influence over the board.

"Avoid over-defending peripheries without purpose."
Rating: 8/10. Over-defensive play can lead to missed opportunities.

"Evaluate the potential of crosscuts in complex engagements."
Rating: 8/10. The crosscut can create complex fights beneficial for skilled players.

"Lean towards capturing invasion stones over passive inward defense, but balance aggression with caution."
Rating: 7/10. While aggression can be advantageous, balance is key.

"Have confidence in your established connections and utilize them efficiently."
Rating: 9/10. Strong connections are a cornerstone of good play.

"Refrain from making inconsequential threats; each move should have purpose."
Rating: 8/10. Empty threats waste moves and can weaken your position.

"Avoid unnecessary defenses against cuts if your groups are resilient independently."
Rating: 8/10. Knowing when not to respond is as important as responding.

Have fun and take care.

Please explain how this works!

I have converted AI reviews into learning points myself, such as "avoid slow connections". But you seem to link the analysis itself to a language model ... How? How did these phrases come to life? What's the input? What's the machinery?

The ability to articulate AI reviews is the holy grail!

I've been reviewing and rephrasing my sentences independently (from AI) in the first place, but I find the process somewhat tedious, and I'm not completely satisfied with my English skills. So, for a bit of fun, I decided to let ChatGPT 4 assist in enhancing the style of my sentences. I thought it would be amusing to have it rate them as well, just for a chuckle.

I don't take ChatGPT's ratings too seriously; they're more for a bit of motivation. However, the improved sentence structures really aid in memorizing the content better.

I must say, ChatGPT does a surprisingly good job at evaluating my unrefined tips!



When reviewing, I concentrate primarily on moves where the optimal choices weren't initially on my radar. I then exert considerable effort to discern the purpose behind these optimal moves, focusing on both the strategic positioning and the timing involved. Afterward, I refine my initial, somewhat unpolished thoughts with the help of AI to achieve clearer and more precise articulation. Kind of a borg effort