This is excellent - some of the best research I've seen yet regarding AI, demonstrating immediately what it can't do and what it does do - both are unarguably problematic and have nothing to do with it being self-aware, skynet or the end of all of our jobs.

Still does generate odd looking frogs too - even added a few creative flourishes, based entirely on assumptions, which is very important to note whenever interacting with AI - necessary even if one is going to use them for actual work.

AI, as is, remains one of the best tools humans have ever created but it's really not human and is always delivering to us what it thinks we want as the primary goal in every interaction.

Even if incapable of delivering what was asked of it - it will still answer, confidently, regardless of whether or not the answer is correct, heavily biased or humorously wrong. This is a valuable lesson in a simple, and hopefully self-evident, package.

Seriously, very well done - I will occasionally be sending people to your site for the foreseeable future, I appreciate your work!