There are plenty of little things that you can add to make even an automated system more humane. For example, assign a unique ID to every automated decision an

There are plenty of little things that you can add to make even an automated system more humane. For example, assign a unique ID to every automated decision and display it in the automated replay email. That way, affected humans can at least sue to get the basis of their decision disclosed. Or why should noreply-support@company.com actually be a no-reply? Let's forward it to the manager who is responsible so that if people are genuinely upset, their "feedback" reaches the correct person. Or attach a footer to those emails, which contains an email address for legal matters. Make sure to say "Please only contact this address for urgent legal matters" so that everyone correctly infers that it'll reach an actual human. As for those gig apps, it's probably good enough to just include the actual reason in the JSON reply that the server sends, without ever displaying it in the app. Of course, that's for internal debugging only, but nobody's going to get the production deployment flags set correctly. Or you can make the decision really obvious, by instantly loading it as an AJAX. Customer fills out a form, the spinner starts, and the result is displayed. "We're terribly sorry but your claim has been denied." But keep the inputs visible, so the customer can tweak the values and submit again. They'll surely figure out what makes the decision flip. And I wouldn't worry too much about colleagues who'd do anything for money. They will probably go to FAANG for better pay anyway. And if all non-FAANG companies have significantly better service, the overall public will notice. Bonus tip: If you ever want to reach a human yourself, there are services like hunter.io that'll procure the email address of a human for you. And then you just circumvent all the automation and directly CC the CTO to say "f* - oops - thank you for your great customer service!"