In a relatively short time, artificial intelligence has become commonplace in the workplace. Nearly half of U.S. employees report using AI at least occasionally, and close to a quarter use it several times a week, according to a December 2025 Gallup survey. Companies, meanwhile, are pouring historic sums into the technology, racing to avoid being left behind in what many executives see as a once-in-a-generation shift.
And yet, something isn’t adding up.
Despite rising adoption and ballooning investment, most AI initiatives never deliver meaningful returns. Research from MIT, for instance, suggests as many as 95% of generative AI pilots fail to produce measurable financial value. When projects stall, leaders are left wondering how so much investment can result in so little progress.
For Heather Conklin, CEO of the leadership development and coaching platform Torch, the problem isn’t the technology. It’s the way companies are trying to absorb it.
Conklin has decades of success helping companies successfully navigate transformations, which means she knows the ins, outs and intricacies of big changes and how they can work.
“You have to be able to bring other people effectively along through a lot of uncertainty and complexity, as AI is moving so quickly and disrupting everything,” Conklin says. “And all of that takes a lot of leadership skills and muscle that I think we’ve unfortunately let atrophy in most organizations.”
The Pyramid Approach
Conklin has watched AI adoption unfold across many industries, and she says these rollouts can be big on excitement but thin on planning. Leaders are eager to experiment, employees are curious and vendors promise sweeping gains. What’s missing, she argues, is a foundation sturdy enough to support the change.
A recipe for success, she says, looks a bit like a pyramid.
At the base is culture and trust: the psychological conditions that allow people to experiment, ask questions and admit uncertainty. Above that sit technical skills, specifically a proficiency with the AI tools being used. Then, at the top of the pyramid are leadership attributes such as creativity, risk tolerance and the ability to guide teams through ambiguity.
To be sure, Conklin doesn’t think every company has to master each level of the pyramid from the jump to carry out a successful AI adoption.
“But I do think if you’re not investing across all three of those, then probably that’s where things are going to fall down,” she says. “The most successful companies that I’ve seen are thinking about both the hard skills side of this equation alongside the people skills of the culture that they need in order to take advantage of that, and investing in both.”
Leaders must also practice what they preach.
For instance, if psychological safety is important to your company—and it’s certainly important for AI adoption, Conklin says—then it’s incumbent upon executives to admit they don’t have all the answers all the time. This may go against the grain of standard practice, but with AI, Conklin says it’s hardly ever the senior-most leaders who have all of the answers and technical expertise at the ready.
“I have to model learning as a leader,” she says. “I have to model saying, ‘I don’t know,' or modeling failure or things that we tried that didn’t work.”
That vulnerability, she adds, creates the perfect foundation for the pyramid approach.
“That’s what starts to create that psychological safety,” she says. “Because in any company, people are always looking to the more senior people to understand what’s acceptable or not, what’s rewarded or not.”
Demoing Leadership
Conklin has seen the most progress with AI adoption when leaders resist the urge to mandate AI usage and instead create visible examples of learning in action.
For example, she’s recently heard many stories about executives using all-hands forums to demo AI products themselves, literally showcasing their vulnerability and their learning for all to see. She calls these “culture-moving actions” that directly shape employees’ comfort level with daunting new technology.
However, even before these demos, Conklin recommends taking a step back and thinking about where AI fits into your company’s strategy.
“The companies who are going to be the most successful with AI are not going to be the ones that just take all of the work that we’re doing today and the way that we do it today and try to automate it with AI,” she says.
Rather, successful companies will ask what they should stop doing so they can free up space for people to use AI in thoughtful, exciting ways—or simply free up space, full stop.
“We gain a lot of personal value from the work that we’re doing,” she says. “So, to help people figure out how to let go of some of that so that they can go somewhere new together, you can just hear how much strong leadership is needed in that kind of a shift.”
Then, of course, there’s the data center-sized elephant in the room: fear of AI-driven layoffs.
Conklin argues that there’s no use shrinking from this tension; your employees will think about it either way, and so will a company’s top execs.
“Everybody knows that this is a game-changing technology that is going to reimagine a lot of how work gets done,” she says. “To be real and honest about that at least creates space for people to identify with it.”
Then, she says, you can work your way through the uncertainty, rather than around it.
Leadership Lessons for Your AI Project
Start with trust, not tools. Without psychological safety, AI experimentation dries up fast.
Model learning from the top. Leaders don’t need all the answers, but they do need curiosity.
Make failure visible. Talking about what didn’t work builds credibility and trust.
Invest in human skills alongside technical ones. Creativity, communication and judgment matter more in an AI-driven workplace.
Use AI to rethink work, not just automate it. The biggest gains come from redesigning processes, not speeding up old ones.
Featured image by Summit Art Creations/Shutterstock
This article was first published in the July/August 2026 issue of SUCCESS Magazine. Get your copy here.
