Funny, not funny

I spent some time talking with an adult child today who needed to vent about trying to use AI to do a task at work. I won't go into the why's because that's not my story, but I am fully up for listening to venting about AI, so was all in. I'm not even going to try to explain what this child was working on because I only vaguely understand it, but there were a couple of things that stood out from our conversation and I haven't been able to get them out of my head since. 

The bottom line as I see it, is that AI is really nothing more than a sassy teen who is smarter than is good for them, but hates to work harder than necessary. Why do I say this? Well, if you give Frederick or Charles or Pierre or whatever name the not-human-but-plays-one-on-the-internet set of connections has been named, if you give it instructions that are too long, Pierre (as we'll call it for simplicity's sake) will do the instructions, but the output will be wrong. When you asked Pierre why it did it wrong, you will be told that Pierre read the beginning of the instructions, but it was long, and so it didn't read the rest. 

I'm not making that up. 

And this is what we're wasting the world's water supply on? 

And do I laugh or feel horrified? 

But wait, there's more.

So, you give Pierre shorter instructions. Pierre does the task and the output is still wrong. When asked, Pierre says that while you were specific, there was a little leeway in one tiny area, so it did that instead. You then need to write the instructions yet more clearly so as to avoid any hint of a loophole, but without making it so long that Pierre will stop reading it. 

I'm not even going to touch Pierre's heavy emoji use. 

All they need to do now is make sure that Pierre won't actually do any work before noon at the absolute earliest and will require a steady infusion of Red Bull while being anointed with Axe to fully reach it's potential.