An AI agent can draft pull requests, especially for mundane or boilerplate tasks, but achieving genuine human trust for complex changes remains a significant hurdle. Currently, AI excels at identifying simple bugs, refactoring small code snippets, or automating routine updates, where the impact is predictable and easily verifiable. However, trust stems from a deep understanding of project context, architectural implications, and subtle human design choices, which AI struggles to fully grasp. Human developers typically scrutinize AI-generated PRs with skepticism, requiring extensive manual review to ensure correctness, security, and alignment with overall project vision. While AI tools can act as powerful assistants, generating initial drafts or suggesting improvements, the ultimate responsibility for merging a PR still lies with a human reviewer. Therefore, truly trusting an AI to autonomously draft and merge critical pull requests for non-trivial changes is unlikely in the near future, requiring ongoing human validation and oversight. More details: https://media.webstore-internet.com/regie/www/delivery/ck.php?ct=1&oaparams=2__bannerid=365__zoneid=86__cb=1069f10c32__oadest=https://infoguide.com.ua