glass half full
approach in this
"AI can be used to determine subsets of test cases that are predicted to run for an incremental commit on the source code. He says this should reduce the overall test cycle time and the speed of deployment"
Nearly anyone who has done software dev will be aware of changes in areas A bizarrely breaking area C that nobody expected.
It's always safest to run all your tests (even on supposedly minor changes that are seemingly very limited in scope) - yes, it takes longer, but if you don't run your tests every time may as well just do hack and hope coding.
Seen classic screw ups with JS jockeys, someone blithely uses a new version of a framework as its got some new "must have" feature (hint - you really do not need it) but that same framework has deprecated some other functions / changed method signatures and so lots of other JS in other areas now breaks.