The author concludes that while AI can be a useful tool in certain cases, it is not yet capable of replacing human investigation and learning. The dream of auto-updating diagrams to match source code remains elusive. The author suggests that one of the best ways to learn an unfamiliar system is to try to diagram it, as this not only aids personal understanding but also helps others learn from the created diagrams.
Key takeaways:
- AI can effectively generate generic and whiteboard diagrams, but struggles with creating detailed system diagrams from source code.
- Creating system diagrams with AI is challenging due to a lack of training data, the complexity of code analysis, and the difficulty of discerning strategy from code.
- Despite these challenges, AI can be a useful tool for whiteboarding new systems, providing initial diagrams and accepting refinements.
- However, AI is not a substitute for human investigation and learning, and the process of creating diagrams can be a valuable learning experience.