“We’ve moved beyond the phase of ‘Is AI a shiny object?’ toward broader mainstream adoption and actual value creation,” says Bryce Hall, associate partner with McKinsey, noting that data from a survey of 2,300 executives shows a nearly 25% year-over-year increase in the use of AI in standard business processes.

Hall joined McKinsey colleague Michael Chui in a recent podcast Q&A, in which they note that while AI is being widely adopted, much of the work is confined to specific narrow use cases, versus more strategic enterprise-centric adoption. “We’re very early in this trend,” says Chui. “As much as we’re seeing this growth in adoption, less than a third of the companies that we surveyed have deployed AI in multiple businesses or functions.” Popular use cases in these initial stages of AI include improving logistics functions, reducing inventory, increasing inventory turns, and increasing overall equipment effectiveness, he adds.

These individual use cases are delivering positive results, Chui adds, estimating improvements — either for the top line or bottom line — of at least one percent to 10%.

At the same time, moving AI out of narrow use cases into wider deployments is “hard work,” Chui says, “not only because the technology problems are hard but also because the change management is really hard. There’s a reason why this doesn’t happen faster.”

Part of the challenge is finding business requirements that can be satisfied through the use of AI — bringing to mind the old adage of a solution in search of a problem. Many of the emerging technologies around AI are fascinating, but it’s still unclear how businesses can capitalizing on these technologies. “For example, reinforcement learning is terrifically good at being able to train machines to play all kinds of games better, but not as obvious is how much value that can create in the world,” says Chui.

Another major industry survey concurs that AI is still in exploratory stages. Just five percent of 2,200 executives surveyed by MIT Sloan Management Review and underwritten by SAS are implementing AI widely across the organization, while 18% have implemented it in a few processes, and 19% are running pilot projects.

The MIT Sloan survey authors also outlined the work ahead to move AI deeper into the enterprise. CIOs and CTOs, for example, “will need to prioritize developing foundational technology capabilities, from infrastructure and cybersecurity to data management and development processes — areas in which those with more advanced AI implementations are already taking the lead.” AI also requires “significant changes to software development and deployment processes,” along with formal data governance efforts.

In addition, the MIT Sloan report points out, AI will require an increased focus on risk management and ethics. The survey “shows a broad awareness of the risks inherent in using AI, but few practitioners have taken action to create policies and processes to manage risks, including ethical, legal, reputational, and financial risks.”

Having the right combination of skills and competency is also essential. “It’s never just a technology challenge,” says McKinsey’s Hall. When looking at successful enterprise AI implementations, they are always aligned to business goals, and these enterprises are “doubling down on investing in current employees and upskilling and training them by creating analytics and AI academies.” Another common element is “internal collaboration and having processes in place to bring business leaders, technologists, and data scientists together to identify where there are collaboration opportunities, making it far more likely that the AI technologies actually generate business value.”