Many corporate leaders are embracing artificial intelligence in theory but falling short when it comes to execution, according to Dan Priest, who was named PricewaterhouseCoopers’ first chief AI officer a year ago.

With generative AI radically reshaping how everything from accounting and human resources to sales and marketing gets done, CEOs’ leadership skills are being put to the test. How they go about their AI strategy today, warned Priest, will likely mean the difference between achieving greater cost savings and faster growth in the next few years versus falling behind the curve.

“It is a disruptive journey that needs to be managed,” he told Business Insider.

Consider, for example, a PwC survey of approximately 4,700 CEOs last year found that four out of 10 expect their business models to no longer be viable in the next decade if AI continues to develop at its current rate. Priest said this suggests companies will need to come up with new — and likely AI-powered ways — of generating revenue, which can be difficult.

Given how fast generative AI has been evolving, Priest stressed the importance of CEOs investing in AI tools and strategic planning around them now, if they haven’t already, to set their businesses up for success. But he conceded that the task is challenging.

For one, leaders need to find ways to distinguish their companies from others using AI if they want to stand out from competitors, said Priest. Most use cases today are merely setting a new standard for table stakes.

“If AI is ubiquitous and everybody’s got it, it can’t be your differentiator alone,” he said.

Leaders also need to figure out which job functions will be aided by AI and to what extent, and which ones will become obsolete, said Priest. Further, they should determine where new skills are needed, invest in helping employees develop them, and assess where talent may need to shift to other areas of the business.

“If you believe that people are an important part of your success in the future, you should invest in their reskilling,” he said.

Workers aren’t all using AI tools in the same way, added Priest.

“Early-career-stage team members are more likely to turn over the thinking too much to AI,” he said. “Late-stage-career team members are probably too reticent to use it consistently.”

One tip Priest has for anyone using AI to write memos or other text is to only rely on the technology for a second draft. People should produce a first and last version on their own, he said.

“You want the thinking to be yours. That’s why the first draft is so important,” Priest said. “You want the benefit of the edit and you want the final draft to be in your voice.”

This is also an example of why he believes humans should be at the center of companies’ AI-related initiatives.

“The shiny new object is AI, but I don’t know a single AI agent that is changing a business,” he said. “It’s the humans combined with those AI agents that change the business.”