Changes Brought by AI
The emergence of AlphaGo not only defeated top human players but more importantly changed our understanding of Go. Many concepts previously considered "common knowledge" need to be re-examined under AI analysis.
Major Style Changes
The 3-3 Point Became Common
Previous thinking: Invading the opponent's star point with a 3-3 early in the game gives them powerful outside influence - a "losing" move.
AI's view: The 3-3 invasion is not a loss. While it gives the opponent outside influence, the territory gained is substantial, and the influence isn't as valuable as previously thought.
Current trend:
- 3-3 invasions early in the opening have become very common
- Professional players often play 3-3 as early as moves 5-10
- Even games where both sides directly occupy 3-3 have appeared
Greater Emphasis on Territory
Previous thinking: "Outside influence" is very valuable, can be used to attack opponents and create more territory.
AI analysis: The actual value of influence is often overestimated. Territory is definite profit, while whether influence can be converted to territory involves much uncertainty.
Insight for amateur players:
- Don't over-believe in "thickness"
- Territory's value is greater than imagined
- Learn to judge whether influence can actually be effective
Changes in Opening Point Choices
| Position | Before AI Era | After AI Era |
|---|---|---|
| Star point (4,4) | Most common opening | Still common but proportion decreased |
| Komoku (3,4) | Very common | Still a mainstream choice |
| San-san (3,3) | Rarely used directly | Became very popular |
| Takamoku (4,5) | Occasionally used | Frequency increased |
Challenges to "Common Knowledge"
AI challenged many Go concepts we took for granted:
Overturned "Common Knowledge"
| Previous Common Knowledge | AI's View |
|---|---|
| Don't invade 3-3 in the opening | 3-3 invasion is completely viable |
| Outside influence is very valuable | Influence is easily overestimated |
| Certain joseki are "best play" | Many new variations discovered |
| Second line is "losing line" | Sometimes second line is the best point |
Principles That Still Hold
While many concepts were overturned, some basic principles remain correct:
- Basic theory: Concepts like liberties, eyes, connections remain unchanged
- Efficiency principle: "Corner, side, center" mostly holds
- Shape principles: Bad shapes are still bad shapes
- Life and death: This is objective, unchanged by AI
What AI changed is mainly "value judgment" (what's better), not "rule-based knowledge" (what's correct).
How Amateur Players Can Adapt
1. Don't Blindly Accept AI Moves
AI's moves are built on superhuman calculation. Many AI moves:
- Require precise follow-up calculation
- May be hard to execute at human strength
- Are not suitable for direct imitation
Suggestion: Understand the "concept" behind AI moves rather than memorizing specific sequences.
2. Re-examine Your Own Concepts
The greatest gift AI gives us is: learning to question.
- Is this move really bad? Or is it just "common knowledge" says so?
- Is this influence really that valuable?
- Do my judgments have blind spots?
3. Maintain an Open Mind
| Closed Attitude | Open Attitude |
|---|---|
| "This move must be wrong" | "Let me think about why AI plays this" |
| "The teacher must be right" | "Teacher's experience is valuable, but we can discuss" |
| "AI moves don't suit humans" | "Let me see what concepts I can learn" |
4. Focus on Improvement Methods Suitable for You
For amateur players, the most important thing may not be learning AI's new moves, but:
- Solidify fundamentals
- Reduce obvious mistakes
- Develop feeling for positions
- Enjoy the playing process
Game Study in the AI Era
How to View AI Recommended Moves
When using AI to analyze your games:
- Don't just look at "best move": Understand why that move is better
- Note win rate changes: See which moves were turning points
- Focus on general direction: Don't obsess over minor differences
Understanding "Win Rate" Meaning
The win rate AI shows (like Black 55%) represents:
- At AI's playing strength, Black's winning probability
- Based on assumption both sides play optimally
- For human players, it's just a reference
Don't over-focus on small win rate changes. For amateur players, 1-2% win rate differences usually have no practical meaning.
Summary
The most important change from the AI era may not be specific moves, but a shift in mindset:
- From "accepting authority" to "independent thinking"
- From "memorizing joseki" to "understanding principles"
- From "seeking standard answers" to "accepting multiple possibilities"
This mindset shift has positive implications for learning Go and even facing life in general.