Oh, I always think about earlier technological breakthroughs, and it seems as though their ethics came not from their creators, but from the people who resisted them. Strikes by assembly-line workers made factories safer and forced manufacturers to introduce standards. They would never have done it on their own initiative.
So it is always a conflict. And with AI, I suspect it will also become a conflict once adoption grows, because right now the entire discourse is still being shaped mainly by model developers and early adopters...
A checklist is such a horrible substitute for judgment. Once something is on the checklist, checking the box is all it is...no thought beyond that.
Your essay actually reminds me of a study in which Israeli daycare centers introduced a small fine for late pickup, only to see lateness increase. The fine converted a moral obligation into a priced transaction, so parents stopped feeling they owed the staff anything and started feeling they'd bought the right to be late. Runs parallel to what you're talking about with checklists, except there's an actual price attached to it. (Source if you wanna read it: rady.ucsd.edu/_files/faculty-research/uri-gneezy/fine.pdf)
Thanks so much for sharing your thoughts! I agree.
I am indeed familiar with the daycare study, and I actually used it as an example in an earlier draft of my post. I ended up cutting it to reduce the length. Great minds thinks alike! ;)
I love this! I am just now writing about how AI has amazing potential to solve big problems, but also to amplify others. We could use it to actually do good for society, like diagnose diseases faster, help teachers personalize learning, or make data analysis accessible to people who’ve never written a line of code. But we could also use it to spread misinformation, automate bias, or make decisions without understanding the consequences.
Ethical judgement cannot be reduced to rules. Every situation that requires judgement is unique. We can run endless loops of tests and feedback and prove the same simple reality that has held for twenty-three centuries. The compliance function is needed to govern our tasks efficiently. Moral decisions about what is good or bad, right or wrong in the particular situation have been taken by humans since the start of civilisation.
Oh, I always think about earlier technological breakthroughs, and it seems as though their ethics came not from their creators, but from the people who resisted them. Strikes by assembly-line workers made factories safer and forced manufacturers to introduce standards. They would never have done it on their own initiative.
So it is always a conflict. And with AI, I suspect it will also become a conflict once adoption grows, because right now the entire discourse is still being shaped mainly by model developers and early adopters...
A checklist is such a horrible substitute for judgment. Once something is on the checklist, checking the box is all it is...no thought beyond that.
Your essay actually reminds me of a study in which Israeli daycare centers introduced a small fine for late pickup, only to see lateness increase. The fine converted a moral obligation into a priced transaction, so parents stopped feeling they owed the staff anything and started feeling they'd bought the right to be late. Runs parallel to what you're talking about with checklists, except there's an actual price attached to it. (Source if you wanna read it: rady.ucsd.edu/_files/faculty-research/uri-gneezy/fine.pdf)
Great points, Lisa!
Thanks so much for sharing your thoughts! I agree.
I am indeed familiar with the daycare study, and I actually used it as an example in an earlier draft of my post. I ended up cutting it to reduce the length. Great minds thinks alike! ;)
Ah!! How wild. Maybe you can use that example in an extended argument later!
I love this! I am just now writing about how AI has amazing potential to solve big problems, but also to amplify others. We could use it to actually do good for society, like diagnose diseases faster, help teachers personalize learning, or make data analysis accessible to people who’ve never written a line of code. But we could also use it to spread misinformation, automate bias, or make decisions without understanding the consequences.
Same technology, different outcomes.
Ethical judgement cannot be reduced to rules. Every situation that requires judgement is unique. We can run endless loops of tests and feedback and prove the same simple reality that has held for twenty-three centuries. The compliance function is needed to govern our tasks efficiently. Moral decisions about what is good or bad, right or wrong in the particular situation have been taken by humans since the start of civilisation.