Remember when The Trolley Problem took the internet by storm? That particular manifestation was around the moral dilemma of autonomous vehicles: If an autonomous car is going to kill 3 people, and you the human driver have the ability to serve and only kill 1 person, would you do it? Should you do it?
For me this dilemma came down to fault: if you do nothing, 3 people die but you can absolve yourself of blame, “it was the car that killed them.” If you choose to swerve and kill only 1 person, you saved a ‘net 2’ people – but you are now at fault for that one death as you made a conscious choice.
By saying “ChatGPT said…”, you are attempting to absolve yourself of fault. “If this is wrong, blame the AI!”
Here’s the thing: Right now AI is not perpetually writting applications autonomously with zero oversight. We prompt it. We ask it to write code and provide answers. By accepting the output, you now own it. You prompted it, you read the output, you had an opportunity to correct it, and now it’s yours.
Before AI, engineers routinely copied code from StackOverflow et.al. But rarely did anyone say “I asked StackOverflow and…”. We passed that code off as our own (perhaps with a knowing wink and a nod).
You copied it
You had an opportunity to modify it
You had an opportunity to cross-check other sources
And ultimately you pushed that code to Master under your name
If that code broke Production – that merge had your name on it. And management came to you asking why you broke Production.
You owned that code. You were at fault for merging something you did not fully understand, regardless of where you got it from.
The fact that we feel the need to preface anything we get out of AI with “ChatGPT said…” tells me we do not fully trust the output, nor our skills at evaluting it. We’re distancing ourselves from the answer it gives, absolving ourselves of fault.
In my humble opinion, it is fine to ask AI questions. It is a source of information. But if you accept what it says, you now own those words. You own that code. I’m going to ignore the part that says “ChatGPT said…” No it did not, you did.
So please think critically about the output of your prompt. Re-read it. Research it further. Or don’t – up to you.
It’s your name on the merge.