How Machine Learning Changed Everything?

In the 60s and 70s, it was thought that machines were good for problems which were programmable. And things like driving were considered not programmable. Now we are on the cusp of driverless cars - a problem that was considered sort of inherently human and full of judgment. We went from this paradigm of specification to this paradigm of learning where we say, well, if we can have our programming, let the machine learn. The consequences of error are much more severe in driverless cars.