There will be mistakes. Humans make them. Artificial intelligences make them, too—and how. Mistakes can be irritating or they can be disastrous.
Every Intelligent System should have a strategy for identifying mistakes; for example, by monitoring critical metrics and giving users easy ways to report problems.
Every Intelligent System should also have a strategy for dealing with mistakes. Perhaps it’s done by updating intelligence; perhaps by having humans override certain behaviors by hand; perhaps by offering a workaround or refund to affected users.
Some mistakes will be very ...
