your friend

You ask a friend for some advice about how to implement a test in your code.

Your brilliant friend will draft a great example of a test, detailing all the information so that you can actually copy their idea to use it.

Would you trust their code?

I would probably not. I’d double-check if they understood my request and see if the code might work as expected.

Or let’s imagine a different scenario: You ask a friend about some historical fact.

They start telling you countless things, but you’re uncertain if those are all entirely true. 

To which extent do you trust them?

Here’s the thing with ChatGPT: Currently, it’s like your friend.  It can be right or wrong, but we shouldn’t assume it’s always right. We should double-check when possible and trust it like we would trust anyone.