AI is the current hot topic across many industries. How can we leverage AI, how can we train AI, how will AI help us advance our operations? Most discussions explore the concept of what can AI do as an end goal. Nearly every application for AI makes a fundamental assumption that the already collected data is accurate and valid. This presentation will challenge that assumption and raise critical and difficult questions with regards to the validity of the data we collect, as well as presenting several suggestions for the future that can help ensure that validity.