GitHub · GH-300
Responsible AI with GitHub Copilot
GH-300 · Lesson 1 of 11 · 3:34
YouTube loads only after you press play. Watch on YouTube ↗
In this lesson
Prepare for the GH-300 exam (GitHub Copilot certification). In this lesson: the risks of generative AI (confidently wrong answers, insecure patterns, out-of-date knowledge, limited context, bias, code close to licensed public code), the six principles of responsible AI followed by Microsoft and GitHub, harms and mitigations (governance, content exclusions and the public code filter, transparency, human oversight, training), and validating Copilot's output with review, tests, linters and security scans. Includes three original practice questions.
Preview the topics
- A team insists that a person, not the AI, answers for every change that Copilot helps produce. Which principle is this?
- A company worries that Copilot suggestions could match licensed public code. Which mitigation fits?
- A developer accepts a long Copilot suggestion that compiles, and ships it without tests. What’s the main mistake?