If you had to explain a complex ML concept to a non-technical stakeholder, how would you approach it?

Situational