Overall Performance

With 3 code edition tasks constructed manually aimed at different  scenes (bug fix, refactoring and feature enhancement), we find that CoEdPilot gained advantage in 2 of them, based on users' average completion time.

P1 on Task 3 screen recording as support of our observation:

some participants modifies the edit content with their own interpretation, which leads to buggy code.

P2 on Task 3 screen recording as support of our observation:

some participants finished the tasks quickly with suggestions.