I’m a huge fan of Ethan Mollick, a professor of entrepreneurialism and innovation at the Wharton School of the University of Pennsylvania. His latest blog post, “Making AI Work: Leadership, Lab & Crowd,” couldn’t be more timely for aging-services providers. Backed by fresh data, he offers a clear, three-part framework that explains why we’re seeing impressive individual wins with AI (like ChatGPT and Gemini), but not much traction at the organizational level. Below, I’ve summarized his findings and offer recommendations to help aging services providers more fully embrace AI’s capabilities to achieve organizational change and even transformation.

Mollick’s blog reviews AI’s impact on workers’ adoption of AI tools (ChatGPT, Claude and CoPilot); their performance and productivity, and AI’s promise. A Danish study, he says, shows that workers shaved 50% off the time it takes to finish 41% of their daily tasks. In a U.S. survey, many workers report slashing 90-minute jobs down to 30 minutes. Adoption is growing: in December 2024, around 30% of U.S. workers were using AI on the job. By April 2025, that number had jumped to 40%, and it’s likely even higher when you factor in “off-the-books” ChatGPT sessions, especially given that ChatGPT is now the fourth most visited website in the world. AI tools are capable, able to pull off hours of research in minutes, and even edging into agent-style workflows that finish real projects end-to-end.

All of that sounds great, right? Yet, putting all of AI’s power into practice is a bit more tricky. Whenever I sit down with LeadingAge members, someone inevitably lights up about the little AI victory they’ve scored, a human resources manager streamlining job posts with ChatGPT, a marketer cranking out campaigns in Gemini, an operations lead automating a tedious report. I celebrate those wins right alongside them. But when I lean in and ask, “So how has this changed life for the whole organization?” the room often falls silent. Mollick’s research backs it up: organizations report only “small” gains. Individual improvements aren’t yet translating into broader organizational impact.

This raises a question: how do we achieve bigger successes? Mollick offers some insights in how to approach the answer. Real, enterprise-level impact demands a rethink of incentives, workflows, and even what we call “work.” His framework boils down to three pillars that make AI stick: Strong Leadership, The Crowd, and The Lab.

Leadership: AI Transformation Starts with Leadership

AI is not just a technology problem; it’s a leadership challenge. While more leaders are beginning to recognize AI’s potential, few are effectively communicating what an AI-powered future looks like for their organization.

Employees don’t typically change their behavior because of performance metrics or pressure alone. They change when they can envision a better, more meaningful future for their work. So leaders need to clearly address questions like AI’s meaning for specific job roles and how staff will be recognized or rewarded for using AI responsibly and effectively.

The Crowd — Empower Staff to Discover What Works

AI breakthroughs usually start at the ground level with the Crowd: staff who tinker with new tools to solve their own problems. Surveys suggest roughly 40% of employees already use AI—often quietly—because official guidance is fuzzy and many fear punishment or extra workload with no reward. Leaders can unlock that hidden expertise by carving out “safe zones” for experimentation, pairing short, practical training with clear guardrails, and publicly celebrating wins. When people see managers using AI in meetings, and when incentives reward those who share high-impact prompts or workflows, experimentation turns into a virtuous, organization-wide learning loop.

The Lab — Build, Test, and Scale Innovation

I’ve heard from LeadingAge members at state events about similar concepts—pop-up AI Lunch-and-Learns, cross-department hack sessions, even Friday-afternoon AI sessions—but none have formalized a dedicated Lab–a centralized place within an organization where your subject matter experts, techies, and nontechies can collaborate and experiment. A true Lab, Mollick says, has one foot in tomorrow, one in today’s workflow. Its charter is, to borrow a management term popular in the early 2000s, ambidextrous, to scout what’s next and to rapidly prototype with current models.

Recruit your best tinkerers from the Crowd, run two-week sprints, and evaluate results against benchmarks tied to real tasks (or, when metrics won’t cut it, a quick gut check). Mollick’s own experiment—an AI agent that produced a full startup analysis and pitch deck in hours—shows how fast iteration exposes both the breakthroughs AI is capable of and blind spots to protect against. Embed that build-measure-learn loop, and the Lab becomes a perpetual engine for scaling AI impact across the organization.

Bringing it Together

Leadership supplies vision and guidance/policy, the Lab turns ideas into tested workflows, and the Crowd scales what works. Miss anyone, and progress stalls. Leadership without a Lab is just talk; Lab without the Crowd becomes a silo, and Crowd without Leadership stays underground. Mollick’s framework offers a practical path to turn isolated individual successes into meaningful, organization-wide impact, leading to better care, more engaged staff, and improved outcomes.

Let’s take the next step together. I’d love to hear how it’s working in your organization: Reach out to me and share your experience!