With machine learning what if the model predicted outcome contradicts with human expertise?

Question

When it comes to machine learning what happens when a model predicts a certain outcome and human experienced and expertise suggest something different? 

Answer 1

Machine learning means acquiring all information from the human case studies and from the human feedback. For example, let’s say machine learning for a particular instance was based on a certain set of data supplied by an operator based on his or her experience. Tomorrow if another set of data comes from another operator, that will also be fed for machine learning and this new set of data which may have additional or even contradicting information will be taken into consideration before a recommendation is made. Any extra feedback or information will only help curate the process to deliver better results next time. In any case, or in most instances, it is always recommended to have a human interface too. Machine learning or artificial intelligence should only support and complement the decision making rather than to reach a fully automated approach without human interface.

Answer 2

Machine Learning requires data to identify data patterns with which to develop predictions. If the input data is biased (for example it is referencing design data that has consistent errors) then this bias will be apparent in to predictions/solutions it provides. Of course there also needs to be enough data for the Machine Learning algorithm to identify patterns with sufficient confidence. If this is lacking then the outputs may be unreliable.