AI coding assistants like GitHub Copilot have changed how quickly junior engineers ship code, but speed isn't the same as growth. In this talk, I share real, specific incidents from my own team, including a two-line change that turned into a full file rewrite because of an under-specified prompt, and a production bug caused by incomplete testing after an AI-assisted change. I also share where these tools genuinely accelerated learning, like helping engineers pick up React hooks and patterns far faster during a major framework migration.

This talk is not a generic "AI is good" or "AI is bad" take. It's a practical, engineering-manager's view of what's actually changing in code review, mentorship, and onboarding, and concrete habits that help junior engineers build real skill instead of just shipping fast. Attendees will leave with specific review practices, mentorship adjustments, and advice they can use with their own junior engineers starting the next day.

Target audience: engineering managers, tech leads, and senior engineers who mentor or review code for junior developers. Preferred duration: 25-30 minutes plus Q&A. No special technical requirements. This will be my first public conference talk on this topic.