IT vs. business: Who owns the machine learning capability?

You can't have machine learning without help from IT, but the discipline is bigger than IT. Practitioners weigh in on who owns the machine learning capability.

Enterprise machine learning capabilities are still in the early stages, but some IT pros already have an idea as to where in an organization a machine learning team should live -- and it might come as a surprise to CIOs.

The first step in figuring out how to deploy your machine learning capability is assessing the maturity of your analytics practice. If you already have a data science or analytics team and a good use case, then your transition to machine learning might not be too difficult. That's according to Norbert Monfort, vice president of IT transformation and innovation at Assurant, who took his data science team and started training them up on machine learning by hiring just one machine learning specialist.

"If you combine the statistical knowledge of a traditional data scientist and bring in someone who's done machine learning, that's a very powerful combination," Monfort said.

Now that you've got your new machine learning team, who in the organization should own it -- IT or the business side? Panelists at AI World -- Monfort; Anju Gupta, head of sustainability campaign at Syngenta; and Robert Bogucki, CTO at deepsense.ai -- had varied thoughts on the question.

Assuming there's a centralized machine learning capability within the organization, where should it be and to whom should it report?

Anju Gupta: I don't know if there's a right answer because we're still in the early stages of [machine learning implementation]. There has been debate on whether the whole thing sits within the CIO role or outside the CIO role. I've seen different models. I'm a big proponent of it staying outside the office of IT because machine learning-based analytics serves way beyond just IT. It is an integral component of IT, but it's more than that. For us, it's more outcome-driven. Analytics is not anything unless we're able to launch a new product because of how we implemented machine learning algorithms. It's a very business-focused, analytical solution.

Norbert Monfort: I agree, and I'm in IT. We have our machine learning and data analytics team as a peer of our CIO. They work very closely together, but it's not in IT. We do have data engineering that's in IT, however. But I agree [with Gupta] that the scope is so much broader than just IT. Having machine learning and data analytics outside of IT -- and having that independent influence -- is a good idea.

Robert Bogucki: I've seen it both ways. It's sensible to have [the machine learning capability] outside of IT, but it's always necessary to collaborate with IT. We often see that there's some kind of innovation or new solution that is actually sculpted outside IT, but at the end of the day it has to be IT that takes over and maintains the project. You have to be sure that if [a machine learning project] is not in IT, whoever owns it will still heavily collaborate with IT.

Monfort: You need that synergy! That's one thing I would definitely recommend -- you always want a cross-functional team that includes IT and others so that everyone is on the same page.

Gupta: Data engineering often doesn't get talked about, so I'm glad you [Monfort] brought it up. Data engineering is a critical component that absolutely must sit within IT.

Dig Deeper on Digital transformation

Cloud Computing
Mobile Computing
Data Center
Sustainability
and ESG
Close