For me, AI helps most when predictive capabilities are defined well from the beginning.
Judgment to outcome are all based upon defining the problem best at the very beginning. AI, machine learning and any other thing you want to put in front must be defined well or it’s just more junk in the system.
For example, if you’re using cookie-based or session based web data inside of an AI engine, you could be informing the machine poorly. It’s something that it can’t recover from at all. Just because you have data does not mean AI can self correct and discover something great.
By understanding the question and having good predictive theories that are causal in nature, you help create great business decisions that have a solid foundation, resulting in value.