Comment by danieltk76

My goal with this blog post is to help others keep their AIs in scope. It seems like recently people are happy to just trust that models will follow your prompts, and that just isn't reliable for production systems.