Guide
How to read the labels
1 · Level — how important?
The most prominent label. The core path is built from must-know material.
Must know
Understand this properly before starting graduate study in AI/ML. The core path is built from these.
Should know
Not strictly foundational, but it makes papers and discussions much easier to follow.
Frontier
Modern or fast-moving. Know that it exists and what problem it addresses; mastery can wait.
2 · Depth — how well?
Only a small subset asks you to implement anything.
Understand
Know what it is and why it exists — conceptually.
Know well
Understand the mechanism and its trade-offs well enough to explain them.
Implement
Be able to code a small version yourself.
3 · Evidence — how settled?
Attached to specific claims, mostly in the chapters on LLMs, reasoning and current research.
Established
Widely replicated and accepted.
Interpretation
The evidence is solid, but this explanation of it is one reading among others.
Active research
Supported by some results, still being tested and debated.
Speculative
Plausible, emerging or contested — treat with caution.
4 · Paper tier — read it when?
Essential
A landmark you should eventually read yourself.
Important
Worth reading once you're comfortable with the essentials.
Optional
For going deeper on a specific thread.