Chasing Business Value with AI and ML

Something you’ve heard before:

“If a tree falls in the woods with nobody around, does it make a noise?”

My version of this saying:

“If a predictive algorithm makes a prediction and nobody acted on it, does it add value?”

The answer to the latter is of course, no.

Many businesses are desperately trying to invest in their own artificial intelligence (AI) and machine learning (ML) projects. What starts out as a high-potential project can result in a wasted effort if it doesn’t start with a strong connection to the end-goal. Perhaps there is a misconception around what that end goal should be.

It’s not “can we predict this metric?”, but rather something like, “can this team take action if they were to receive this prediction?”. This shifts the focus from the analytical capability back to the business. It also points to questions that need to be answered before there should be any confidence in the outcome. Maybe they can act, but would they?

Boiling a project idea down to the true value-add is more of a business question than a research one. Which brings us to the major disconnect that many businesses are likely facing. The typical data scientists within a company aren’t all that connected to the rest of the business, and the typical business isn’t all that connected to the capabilities of the data scientists.

There isn’t anyone to blame for this disconnect. The field is relatively new to most everyone, and almost everyone who has the capabilities of a data scientist learned from outside of the business. Sure, there are a handful of companies leading the charge, but I’m speaking to the rest of the businesses that are only just now trying to discover the value of AI and ML for themselves.

To give a single solution would do injustice to the complexity of the problem, but there is at least one thing to always remember before your AI/ML project moves too far along. If nobody is going to act on this prediction then we just shouldn’t do it.