Do you need machine learning or was first semester graduate statistics good enough?

Artificial intelligence and, by extension, machine learning are fields with decades of research and mathematical formulations behind them. It is only now that we can use machine learning algorithms on even the most budget PCs and laptops that we have seen an explosion in their adoption. As with any tool that suddenly becomes easily accessible and dominates the titles of journal articles and special collections putting it into practice comes with challenges. What projects are suitable for machine learning? Are there traps and pitfalls to avoid? How do we continue to conduct reproducible research? What advantages and disadvantages does machine learning offer over typical frequentist statistics?