This is a horrible article written by someone clearly upset about the state of things to process it objectively.

Please refute this argument: how can you expect 100% accuracy when your input is natural language which is fuzzy by definition?