Well argued piece here on the dangers of machine learning and AI when data is still the weak link

The Economist Intelligence Unit (EIU) found that although 70% of business executives rated analytics as “very” or “extremely important”, just 2% are ready to say they have achieved “broad positive results”

See "Damning analysis of Analytics?"

Data is usually the reason- incomplete, inaccessible, out-of-date, corrupted- often all four!

Enterprises often cannot access 60%, 70% even 80% of its data hidden in silos and/or unstructured formats. That's why BI and Analytics so often disappoint and why AI could just supercharge the disappointment.

BI and Analytics needs the whole organisation from frontline to CEO to be team players. See how in "Analytics is a Team Sport". which links to good McKinsey Advice.