Most enterprise generative AI investments have yet to deliver the value companies envisioned, and every day, more leaders are recognizing that people lie at the heart of the struggle. In this year’s AI & Data Leadership Executive Benchmark Survey, 93% of executives leading AI and data efforts identified human issues around culture and change management as the primary obstacle to adoption. McKinsey Global Managing Partner Bob Sternfels put it plainly on HBR’s IdeaCast: “Half if not more of the secret sauce” in getting value from AI, he said, “is organizational change, as opposed to technology implementation.”
As such, many leading companies have launched initiatives over the past several months to drive AI adoption across their workforces. These efforts run the gamut from carrot-to-stick approaches, with some rolling out hackathon programs and prizes for innovative uses. Others use weekly logins and token consumption as proxies for performance.
A PERFORMATIVE APPROACH
Leaders are right to focus on the people side of adoption. They need to be deliberate, however, about what they’re encouraging. I’ve learned something in my three decades helping some of the world’s largest companies through culture transformation. Employees prioritize what leaders model, incentivize, and reward. And initiatives built around shallow metrics can do more harm than good.
It’s understandable why many leaders today celebrate deliverables simply because they were made with AI, or reward employees for integrating it into workflows. Facing underwhelming internal adoption metrics, many have come to see any increase in AI usage as a win. At my firm, however, we call this “trophy-style” AI adoption—which is to say, a performative approach focused more on usage than results. It’s focused on participation trophies over proof of impact. Leaders need to be wary of this trap.
Because as anyone following the “workslop” problem or the emerging research on cognitive atrophy will know, not all AI use cases are created equal. Trophy-style adoption creates a dangerous illusion of progress, where activity masquerades as impact. In other words, we’re rewarding output over outcomes. A culture built around shallow adoption risks more than struggling to achieve ROI; in some cases, it might leave employees less equipped to meet business needs than prior to AI.
IMPACTFUL ADOPTION
Impactful AI adoption will look different based on the company, a person’s role, and many other factors. For some, it means deepening the quality of the same work product. For others, it means increasing output without sacrificing quality. And for still others, it means getting the same work done in less time, repurposing time and energy toward new questions and tasks. All the best adoption initiatives, however, will be reverse-engineered from the larger business strategy. They will be built around metrics that connect to it.
The process of designing an adoption initiative should start with clarity and specificity around big-picture questions. What does value look like for our organization? How can different roles change to better deliver it? Leaders cannot lose sight of these framing questions as they determine what gets modeled and encouraged.
Wise ones will drive for real business impacts that come from the usage. And when they showcase strong use cases, they will not just reward speed or deep integration. Instead, they will keep the focus on the larger picture, taking great care to explain the meaningful organizational outcomes driven by the use case. In Gagen MacDonald’s latest white paper, we dive into what it takes to do this well, and what organizations can do to bridge the separate realities that exist between leaders and employees around AI.
Because while the employees who create the most impact with AI will certainly use it frequently, it’s a mistake to think of usage as synonymous with impact.
And given how much companies have spent and plan to keep spending on this technology, it’s not a mistake many leaders can afford to make.
Maril MacDonald is founder and CEO of Gagen MacDonald.
