What’s the best path to success for artificial intelligence (AI) projects? There are any number of challenges – ranging from finding the right data scientists and talent to develop the system, to acquiring the right data to train the system. Fundamentally, though, the factor with the greatest impact on success or failure is who’s driving the initiative. If the project arises from a clearly defined business need, then it stands a much greater likelihood of success than if it arises from a technology-first “solution in search of a problem” approach. Unfortunately, a report from KPMG indicates that many companies are taking the latter approach. 

In a survey of 950 business decision makers across seven industries, 47% of respondents indicated that the IT department is the functional area leading the adoption of AI in their organizations – far more than any other functional area. While this may seem to make sense – after all, any AI project is going to require the assistance of the IT department − KPMG noted it is also symptomatic or an overreliance on IT for AI projects. And, when IT leads the adoption of AI, the risks of falling into the “solution in search of a problem” trap go up.

Not surprisingly, 34% of respondents indicated that IT was also the biggest beneficiary of AI – so many of those IT-led projects are for IT-specific AI systems. Data/analytics (13%) was the next most commonly cited beneficiary, followed by operations (12%). Areas one normally associates with driving business (customer service, marketing, sales) were cited less frequently.

What does this tell us? For one thing, AI is most commonly used for efficiency gains, rather than accomplishing growth objectives. And, AI still has some distance to cover on the technology maturity curve. When you think about other classes of more mature technology, business leaders – not technical leaders − typically identify the need for a solution. For example, if a company’s e-commerce platform lacks the functionality to fully engage customers and drive sales, that need will be identified by an e-commerce business leader, who then takes a primary role in defining requirements, selection criteria and timelines for procuring and deploying a new platform. IT provides technical support and a layer of due diligence for technology procurement, but does not “own” the project.

The same type of process needs to be in place for AI projects. Extending on the e-commerce example, you wouldn’t want IT taking the lead on procuring or developing AI-based chatbot technology for the e-commerce site and then handing it off to the e-commerce team. This process should be driven by the e-commerce team, which has done a needs assessment that shows how a chatbot could improve customer satisfaction and sales. As the KPMG report says, “The worst AI is developed within an IT department when there is no business involvement.” 

As AI continues to race down the road of technological progress, it’s important for organizations to put the right people in the driver seat. That can be an IT person, if the AI project is for the benefit of the IT department. But for everything else, business people need to envision the need, and then work collaboratively with IT or partners to make the vision a reality.