Back to blog

5 Common Mistakes in Go and How AI Helps You Spot Them

Published on March 30, 2026 by StoneBase Team

Every Go player, from beginner to dan level, has blind spots. The tricky part is that your blind spots are, by definition, things you don’t notice. That’s where AI analysis becomes invaluable: it sees what you miss and quantifies exactly how much each mistake costs you.

Here are five of the most common mistake patterns we see when players review their games with KataGo, and how to use AI feedback to correct them.

1. Playing Too Many Small Moves in the Opening

One of the most frequent patterns AI reveals is players making low-value moves early in the game. Moves that are worth 5-10 points when there are still 30-point moves available on the board.

What it looks like in AI analysis

You’ll see KataGo consistently suggesting moves in a different area of the board than where you played. The score drop per move might be small (just 2-4 points), but these small losses add up across 10-15 opening moves, creating a deficit of 20-40 points before the middle game even begins.

How to fix it

When reviewing, pay attention to moves where the AI suggests a completely different area. Ask yourself: “Why is that area bigger?” Usually, it comes down to the framework being more open, the potential territory being larger, or an urgent move being needed to prevent your opponent from securing a large area.

The opening is about efficiency. If the AI says your move was worth 8 points and the best move was worth 25 points, that’s a 17-point lesson in prioritization.

2. Not Responding to Opponent’s Threats

This mistake shows up as a sharp win rate drop right after your opponent makes a threatening move. You play elsewhere, thinking it can wait, and the AI shows that your opponent’s next move in that area is devastating.

What it looks like in AI analysis

The win rate graph shows a pattern: your opponent plays a move, you tenuki (play elsewhere), and then there’s a sudden drop on your opponent’s next move in the original area. The AI’s suggestion for your move was to respond locally.

How to fix it

When you see this pattern, study the opponent’s follow-up move carefully. What did it threaten? What did it accomplish? Understanding the severity of the threat helps you judge similar situations in future games.

Not every threat requires an immediate response; sometimes tenuki is correct. The AI helps you calibrate which threats are urgent and which can wait. Over time, this calibration becomes part of your intuition.

3. Fighting in the Wrong Place

Go is a game of choosing your battles. A common mistake is getting drawn into a fight in an area where the result doesn’t matter much, while ignoring a more important area of the board.

What it looks like in AI analysis

You’ll notice that during a local fight, the AI consistently suggests moves elsewhere, perhaps a simple extension or a key point in another area. The fight you’re engaged in might be evaluated as close to equal regardless of the outcome, while the area the AI suggests is decisive.

How to fix it

Before engaging in a fight, ask yourself: “What do I gain if I win this fight? What do I lose if I lose it?” If the answer is “not much either way,” it’s probably not worth fighting over.

AI analysis trains you to zoom out and see the whole board. When you review positions where you fought locally but the AI suggested a global move, you’re building the skill of identifying what truly matters in a position.

4. Misreading Life and Death

Life and death mistakes are the most dramatic on the win rate graph. One moment you’re ahead, the next you’ve lost 30+ points because a group died. These can happen at any level, from misreading a basic eye shape to missing a complex capturing race.

What it looks like in AI analysis

A massive, sudden drop in win rate, often 30-50% or more. When you look at the position, you’ll see that a group you assumed was alive was actually dead, or a group you were trying to kill was actually alive.

How to fix it

For each life and death mistake, pause and work through the reading:

  • What was the vital point?
  • What was the correct sequence?
  • At what point did you go wrong?

Then try to solve the position again without looking at the AI’s answer. If you can solve it on the second attempt, your reading is close and you just need more practice. If you can’t solve it even knowing there’s a problem, it reveals a gap in your life and death fundamentals.

StoneBase lets you explore variations freely: place stones, try different sequences, and see the AI’s evaluation at each step. This is much more effective than just looking at the answer.

5. Ignoring Endgame Technique

Many players focus their study on openings and middle game fighting, but the endgame is where games are decided. A 10-point endgame mistake in a close game is the difference between winning and losing.

What it looks like in AI analysis

In the last 50-80 moves of the game, you’ll see a gradual decline in your score estimate. Not from dramatic mistakes, but from a steady trickle of 1-3 point losses: playing endgame moves in the wrong order, missing sente moves, or failing to count properly.

How to fix it

Endgame review is different from middle game review. Instead of looking for dramatic mistakes, look for:

  • Sente moves you missed: Moves your opponent has to respond to, letting you keep the initiative. These should be played before gote (non-forcing) moves.
  • Move order errors: Playing a 5-point gote move before a 10-point gote move costs you 5 points.
  • Boundary plays: Moves at the border between territories that you played from the wrong direction or not at all.

The endgame is the most calculation-heavy phase of Go, and AI tools excel at showing you the correct order and value of each move.

Using AI Effectively for Improvement

The common thread in all five mistake types is that AI doesn’t just tell you that a move was wrong. It tells you how much it cost and what was better. This quantification is what makes AI-assisted review so powerful.

Here’s a productive review workflow:

  1. Sort mistakes by size. Focus on the biggest point losses first.
  2. Categorize them. Which of the five patterns above does each mistake fall into?
  3. Look for repetition. If you’re making the same type of mistake across multiple games, that’s your priority study area.
  4. Practice the correction. Don’t just note the AI’s suggestion. Try to understand the principle and apply it to similar positions.

Improvement in Go isn’t about eliminating all mistakes at once. It’s about identifying your most expensive mistake pattern and working on it until it becomes less frequent. Then moving on to the next one.