What If We Can Never Trust A.I.?
newyorker.com
2 threads
Never trust AI. Just because it should do the thing you ask, doesn't mean it won't go off the rails. Check and balances, trust but verify. Or don't trust at all and put restraints in place.
There’s another interesting aspect to this: the outcome is often strongly tied to the quality of the prompt. The prompt itself should include a built-in quality-control component.
Thanks to AI, you can never trust the internet too. We're going to swim in generated content like never before.