“Building Machine Learning infrastructure and frameworks almost always benefit the business.

On the ROI of AI’s pioneering efforts.

Scenario #1

Imagine a company that makes vacuums. Now imagine this vacuum company deciding it wants to make coffee tables. If this experiment goes wonderfully and the coffee tables sell like hotcakes, then it’s fair to say this effort was successful! Hooray! The vacuum company is now a vacuum and coffee table company. Now imagine this experiment goes woefully wrong and the coffee table experiment fails. The company is now a vacuum company sitting on a technology that is broadly inapplicable to their core business. They can make vacuum cleaners and have table-making equipment sitting in a warehouse.

Scenario #2

Imagine a company that makes vacuums. Now imagine that company decides it wants to invest in a handful of engineers to develop a new technology that uses Artificial Intelligence to scan for particles on the floor using AI. Now imagine they succeed and create the vacuum cleaner of the future. Hooray! What a successful endeavor! Now imagine if that effort fails. Imagine they ship their new vacuum cleaner and the market isn’t interested. Darn. The company is now a vacuum company sitting on a useless technology and most likely tooling, infrastructure, and talent that can be displaced and used in many different parts of the business. This is good! Even if our ROI is much lower, at least we’re well equipped for our next effort.”

The article: AI/ML Projects Are Allowed to Fail