概念
结论
综合整个课程，理解我们从理论到高级治理的旅程，并部署你完整、基于证据的人工智能安全论点。
| Old Role Focus | New Role Focus |
|---|---|
| Code Execution (Does the model work?) | Risk Elicitation (How can I break it?) |
| Debugging (Fixing a mistake) | Forensics and Audit (Finding hidden bias/deception) |
| RLHF Implementation (Code) | Alignment Architecture (Steering intent at scale) |
| System Testing (Compliance checklist) | Assurance and Governance (Proving safety with a formal Safety Case) |