The biggest challenge for enterprise AI is localized talent

A recent survey conducted by industry research firm, Ovum, in partnership with data science software maker, Dataiku, found that one of the biggest challenges to companies looking to create meaningful insights from AI is getting the right skills in the right places.

The whitepaper report titled ‘Profit from AI and Machine Learning’ surveyed data leaders at Global 2,000 organizations with extensive experience putting AI projects into production and did a deep dive into their insights on the industry and the future it holds.

“AI projects require the discipline of data science, but data science projects don’t always require AI,” said Tony Baer, Principal Analyst with Ovum and author of the study. “On the process side, AI projects build on the tasks involved with data science by adding numerous additional tasks, variables, and checkpoints into the process. And when it comes to people, there are critical relationships that must be maintained between the project team and the business, but also the data scientists and the data engineers, who make AI projects reality.

One of the biggest challenges to enterprise AI the report found was the ‘People Equation’ and where the actual data scientists resided in an organization versus where they are needed and how to correct this problem. All the respondents stated that data science is the foundation of AI and data scientists are needed to develop AI solutions for the organization. However, the best place to deploy these data scientists is within localized business units as ‘boots on the ground.’

The issue with staffing highly skilled data scientists in field offices outside of major metropolitan areas is that there simply aren’t enough skilled data scientists out there and the demand for these skilled workers is so great that they have their choice of locations. The solutions are to staff highly data scientists at the company headquarters and develop strong collaboration within the organization or to focus on a ‘centers of excellence’ model.

Florian Douetteau, CEO of Dataiku, added: “Time and time again, we see teams in companies around the world and across industries that aren’t able to get their data efforts off the ground because they have no way for these people in different geographies – much less different types of people with different skills – to work together. Once the enterprise is able to address the people part of the equation, only then can they can then take the final steps toward becoming data-driven through the rest of the equation – technology and processes.”

For a global organization, building a skilled data science team in one location, like the company headquarters, is a bit of a conundrum because of the need for data science teams to have a presence in local areas that the company operates, but often there isn’t a large enough talent pool of data scientists in these areas (yet).

The solution for this problem is to create a strong company culture of collaboration. If local data miners and business units have direct and easy access to collaborate with the headquarter-centric data science team, enterprise AI models are much easier to create.

Another structure that can potentially solve the ‘People Equation’ that affects enterprise AI development is to develop a ‘Center of Excellence’ (CoE) type of structure with data science within the organization. The CoE model is usually best suited towards early stage companies and it involves building a smaller, elite team that is selective in the types of projects they undertake for the organization to maintain focus and quality.

Typically, the CoE is also located near the global headquarter of the organization but is meant to be a separate entity. Similar with the headquarter-centric model, the CoE model requires fantastic collaboration with the regional business teams.

Overall, the biggest challenge that the experts presented towards AI development is solvable with collaboration, knowledge sharing, and ‘train the trainer’ objectives that are baked into the data science teams and ultimately the organization. At this stage, for any company to truly gain an advantage and profit from enterprise AI they need to take collaboration seriously.

The 15-page whitepaper report also concluded that the lifecycle of AI projects includes more complexity and change compared to traditional data science and analytics projects and numerous other insights.