A new analysis reveals that hybrid PRs—containing 25-50% AI-generated code—create unique review challenges. While 84% of developers use AI tools, only 3% highly trust them. This mismatch creates an uncanny valley effect, where code blends human and AI patterns, making it harder to review than either pure human or pure AI code.

The core problem is cognitive switching. Reviewing pure human code involves understanding intent, while pure AI code requires pattern verification. Hybrid PRs force constant mental model shifts, increasing cognitive load and error rates. Studies show AI code has more critical issues, but hybrid PRs are uniquely disruptive because they prevent reviewers from applying a consistent evaluation strategy.

Velocity is outpacing review capacity. PR sizes are growing 154% with AI, but only 10% of developers use AI for reviews. Teams must develop new skills for verifying AI code and structure PRs to minimize context switching. The solution isn't hiding AI contributions but making them clearly identifiable, allowing reviewers to apply the correct verification strategy.