how much longer till the "pets vs livestock" attitude towards server management applies to code written by humans vs code written by ai?
and what does managing ai devs look like in practice? prompting for & reviewing functional test suites -> trusting the ai to turn them into bug free code?