Should QA and testing get passed off to AI agents?

Curious to hear your thoughts! How much human testing and how involved should people be in QA as this tech continues to improve?