Have a glimpse at the kind of errors we will be facing with the machine-learning AI in the near future. How do we debug them and how do we know there is an err

Have a glimpse at the kind of errors we will be facing with the machine-learning AI in the near future. How do we debug them and how do we know there is an error to correct in the first place?