AI and LLMs · Senior · Question 45 of 75
What is the main challenge of testing an AI coding agent?
- 1Coding agents always produce identical code for a task
- 2Coding agents can't be run in automated environments
- 3Checking the produced code works and is safe, not just plausible
Show the answer
Correct answer: Checking the produced code works and is safe, not just plausible
Explanation
Run the agent's changes against real test suites in sandboxed repos, and review for security issues and scope creep.