Is IT a Top Cause of ML and AI Failures?

Its a well-known mantra that culture is a big part of your ML and AI success or failure. Data-driven is the term that is used — overused if I may say — to indicate a company or organization that uses data in all of its business without in a natural way instead of being forced to do it by a top-down mandate or making a conscious effort to do so. Business literature is replete with articles like [1] that provide a high level overview or a list of do’s and don’ts. Those who are in the trenches need more specifics.

Photo by imgix on Unsplash

In this post I share my thoughts on one of the key organizational and cultural element that, in my opinion, is a key issue and has not been openly discussed: Information Technology. The reality for over 90% of the organizations is that your IT department in the way of your ML success and its a leadership, culture issue that must not be ignored. There is the obvious technology skills gap issue but actually this is more of a culture problem with traditions carried over the history. One concrete example is that for most companies, data has not been considered a carefully controlled asset but exists in a scattered, inconsistent fashion.

Here are common dialogues from ML and IT interactions — heard across the globe. Ring a bell?

• ML team: We need data <insert a long list here>and oh, where’s the data dictionary? why is the…