Until AI Companies Solve the Catastrophic Failure Problem, Parity With Humans Doesn’t Matter

A discussion of "parity" with humans vs. catastrophic AI errors