Seriously, we need to create some kind of usage agreement for AIs.

'AI may provide incorrect information and you must always cross-reference or compare it with information you found yourself to determine its veracity. Any negligence resulting from failure to comply with this is entirely the user's responsibility.'

Every time I ask the AI something, they should make it so you can't use it unless you read that out loud three times. This is a direct shot at a popular post, but it's something I've thought about often anyway. People are increasingly arguing that 1+1 equals 1 if the AI says so, and then they pull the 'Are you smarter than the AI?' card... I'm glad I'm dumber than the AI, but at least I'm smarter than the people who blindly trust it.