01 — Core Judgment

Status: REQUIRED · the only section with a gate · everything else assumes this is done

Purpose

This is the required foundation. Everything else in this curriculum assumes you have done this work. The goal is simple: leave with decision filters that still work when the tools change.

The modules

1. What Prediction Engines Actually Do — what the machine is, and why fluency is not evidence
2. When AI Is the Wrong Tool — the five-question filter you run before reaching for anything
3. Verification Cost and Net Value — the math everyone skips, taught once, properly
4. Failure Modes and Accountability — how it breaks, how you'll know, who owns it
5. The 10 Questions Before Any AI Project — the gate you do not skip
6. Internal Assets vs Shiny Tools — what the real advantage actually is
7. Evidence Standards — completion is not competence

Who this is for

Anyone who plans to use AI for work — which is everyone still reading. You do not skip this section. You can be the most senior person in the building; this is still the floor.

What you will be able to decide after this

• What a prediction engine actually does and does not do — without the marketing layer.
• When AI is the wrong tool — and how to tell before you spend the time.
• Whether a task's verification cost outweighs its value — with numbers, not vibes.
• How AI fails, who is accountable, and where the human gate goes.
• Whether an internal capability is worth building vs. renting a shiny tool.
• What counts as evidence, and what counts as theater.

The rules of this section

1. Every module ends with a concrete artifact and a verification requirement. No reading-and-nodding.
2. 45–90 minutes of real work per module. Dense over long.
3. Nothing in this section exists to fill a slot. If a module isn't doing real work, it doesn't belong here.
4. The applied packs reference these filters. They do not re-teach them. If you skipped Core, the packs will feel like a foreign language — that's working as intended.

The checkpoint

You are ready to move into any Applied Pack when you can:

• Explain what a prediction engine is actually doing without slipping into "thinking" or "knowing" language
• Decide, with reasons, when AI is the wrong tool
• Measure net value including verification cost
• Name failure modes and ownership for a real use case
• Answer the 10 Questions for a real project
• Distinguish internal assets from shiny tools
• Produce evidence of judgment rather than evidence of attendance

If any of those are still fuzzy, stay here and strengthen the work. Moving forward without them just recreates the original problem this curriculum was built to fix.