In total, the company’s workforce has gathered over 200 ideas thus far, and several of them have received a first round of internal funding, says Ken Washington, Medtronic’s senior vice president and chief technology and innovation officer. Casting a wide net for ideas is also critical to create a culture where generative AI becomes a key part of everyone’s job. It won’t be as successful if a small pocket of AI experts were empowered to build solutions for the company, he says.

“We touch more than 74 million patients every year,” says Washington, who spoke during a Fortune Brainstorm AI virtual discussion held in partnership with Accenture. “If you do the math, that’s two people every second of every day getting touched by technology that’s coming.”

Among the generative AI tools Medtronic has already made available to its staff include Microsoft 365 Copilot and an internal version of ChatGPT, a chatbot known as MedtronicGPT. Beyond those employee-focused productivity tools, the company will prioritize AI investments that can improve patient outcomes, including using AI to improve the ability to detect polyps during a colonoscopy or reduce false positives from cardiac monitors.

“These are outcomes that come from applying AI to a medical procedure and a medical device that changes the patient outcomes and improves the lives of the patients and clinicians,” says Washington. “And we’re just getting started.”

Accenture estimates that 90% of companies are exploring generative AI or AI capabilities, but less than a third of those firms are building the proper capabilities to set themselves up for success, like creating an AI center of excellence or developing use cases in a structured way like at Medtronic. And fewer than 20% of companies are getting close to the targeted value they hope to achieve from their AI investments, according to Muqsit Ashraf, Accenture’s group chief executive of strategy, who also spoke at the virtual event.

“The prevalent approach has been more a hammer looking for a nail,” says Ashraf. “Which is, ‘What can AI do for me?’ And then picking a set of disparate use cases.”

In California, the state government is using AI both to improve workflows for its employees and to make life better for the state’s nearly 40 million residents. Some examples include an AI-powered chatbot to sort through requests at the Department of Motor Vehicles and a program developed with the University of California, San Diego, to train AI to detect smoke and help prevent forest fires.

“Any technology we’re looking at, it should not only be looked at financially, it needs to look at the human [return on investment] and also what we’re doing on policy,” says Liana Bailey-Crimmins, California’s state chief information officer and director of the California Department of Technology, who also took part in the discussion.

Bailey-Crimmins says California has also embraced the broad array of AI vendors in the market, asking companies to pitch solutions to the state, which include addressing traffic bottlenecks, confronting climate change, or making better decisions about when to time construction projects.

Bailey-Crimmins says she’s open to any technologies that help solve problems, and said more traditional forms of AI and other emerging technologies could be a better fit for some of the state’s issues. “Maybe gen AI is not the solution,” she says.

Brazilian-based Natura & Co, which makes personal care and beauty products, has sorted AI use cases into two separate buckets. There’s everyday AI, which relates to making employees more productive. And then there is “change-the-game AI,” which is about bigger-picture advances that impact production, customers, and the environment, says Chief Information Officer Renata Marques. 

Marques likens the excitement around generative AI to an iceberg. At the tip is the technology, but under the water, companies need to sort out the proper data usage, engineering, governance, and strategy. 

“It’s our job to invite people that don’t understand all the complexity and dimensions and work together,” says Marques.

Natura is also very focused on measuring the investments made in AI technologies, the objectives the company hopes to achieve, and the results it wants to obtain. “Without the business results, it is just a lab,” says Marques. “And we are not a lab.”

As AI technologies continue to develop, Accenture says companies shouldn’t lose sight of the fact that technology costs only account for about 30% of the spending on AI. The remaining 70% is for training, hiring, and the change management necessary to support the new uses for AI. 

Medtronic, for example, has focused a lot of energy on an education campaign internally to teach employees about what is possible with AI technologies and what isn’t. Washington says with all the hype around AI, it is critical that there’s “clarity around what this technology really is and what it means to your business.”

“It’s not just about the tech, it is about the transformation,” says Accenture’s Ashraf. “The value is unlocked when you reimagine or reinvent functions, processes, or ways of working.”