AI Is the Crane. Who's in the Cab?
Lire en français: IA, prenez les commandes !
Source: AI with Canva
In a study of 244 BCG consultants working with generative AI on a realistic business task[1], researchers found three ways of working with the machine. Most stayed actively involved: 60% worked hand in hand with AI step by step, and 14% directed it while keeping the key decisions for themselves.
But 27% did something else. They handed the task over and accepted the output without modification, or with only small edits.
The researchers call them “self-automators”. They got the job done. But unlike their colleagues, they developed neither domain expertise nor AI skills along the way. They had the same tool as everyone else. What they didn't have was a hand on the controls.
The technology is not going away
There is no shortage of reasons to be cautious about AI. Investors debate whether valuations have run ahead of reality. Regulators, researchers and employees raise serious questions about ethics, jobs and power. These debates matter.
But they won't send the technology back. Whatever the markets do next, AI is now built into the tools millions of people use every day. For leaders, the question is no longer whether to use it. It's how.
AI is the crane
Source: Canva
A crane lifts loads that no crowd of people could carry. It's an extraordinary machine. But a crane doesn't decide what's worth lifting, where it should go, or what happens if it drops something it shouldn't have. Someone has to be in the cab, the crane cabin.
Most organizations today are focused on the crane: which tools, which use cases, how fast, which jobs disappear or, on the contrary, become necessary. Far fewer ask who sits in the cab, and how the machine is operated. Yet that is what decides whether AI enhances what makes an organization special (its culture, its internal networks, its people's competence) or quietly erases it.
Not designing is also a design. When nobody chooses how AI enters the way people work, the defaults choose for you.
Three levers in the cab
In my work, I keep coming back to three levers. Each can be pushed up or down. Pushed down, the risks compound. Pushed up, so do the opportunities.
Agency: each person's power to act. AI can sharpen human judgment, or replace it. The self-automators above show what pushing this lever down looks like: the work gets done, and the people doing it stop growing. In another study, ideas produced with ChatGPT were more similar to one another than ideas produced without it[2]. Each idea was fine on its own; together, they were more uniform. Push the lever up, and AI becomes a sparring partner that builds competence and the confidence to take initiative.
Networks: how people connect and share knowledge. A recent study tracked Microsoft 365 activity in 11 large companies after they introduced Microsoft Copilot[3]. Among employees who used it intensively, activity in documents rose by 21%, while activity in email and chat rose by only 7%. The authors see possible benefits, such as less information overload. But they warn that a relative decline in communication may reduce the diffusion of diverse information, and ultimately hinder innovation. Organizations innovate through their connections. Push this lever down, and people become more productive alone and less connected together. Push it up, and AI frees time for the conversations that matter.
Community: why people commit to act together. In June 2026, 52% of American adults said they were more concerned than excited about AI in daily life, up from 37% in 2021[4]. For the first time, a majority of adults under 30 said the same: 55%, up from 31%. Worried people don't commit to a transformation because it was announced. They commit to what they help create. Push this lever down, and AI becomes something done to people. Push it up, and it becomes a shared project, connected to why the organization exists.
The operator
There's a fourth principle, and it isn't a lever. It's the operator, staying alert in the cab month after month. I call it Coherence.
Most transformations don't fail on day one. They fade. Attention moves on, sponsors change, and people learn to wait it out. With AI, the cost of that fade is higher, because the technology keeps moving even when leadership attention doesn't.
I’ve seen the opposite in several organizations I’ve worked with. When volunteer coalitions work in partnership with senior leaders, and those leaders stay engaged over time (launching a second coalition, then a third), they deliver the most profound, most sustainable, most value-creating changes.
Four questions for your leadership team
Where is AI quietly replacing our people's judgment, instead of sharpening it?
Are our people connecting more because of AI, or less?
Do they see this transformation as theirs, or as something done to them?
Who will still be in the cab – with a hand on the transformation – in 2028 and beyond?
From ideas to practice
Last week, I designed and delivered an executive workshop on these questions for a group of senior HR leaders. We didn't just discuss the principles: we applied them to ourselves, with AI in the room.
Participants left with a clearer view of where AI strengthens or weakens judgment, connection and commitment; new insights worked out with their peers; first-hand experience of using AI together; and concrete actions of their own.
This workshop is one of the first results of a deep dive into the latest research on AI and organizations. More will follow in the coming weeks.
Would you like to bring it to your leadership team? Get in touch. The workshop is available in English and French.
As Myron Rogers puts it (I paraphrase slightly): the way you get to the future is the future you get.
AI is the crane. The question is who's in the cab, paying attention, fully present.
Sources:
[1] Randazzo, Lifshitz-Assaf, Kellogg, Dell'Acqua, Mollick, Candelon & Lakhani, "Cyborgs, Centaurs and Self-Automators," Wharton/HBS working paper (SSRN); summarized by the authors in Fortune, Jan 2026 – Full working paper via Harvard Business School Publishing.
[2] Anderson, Shah & Kreminski, "Homogenization Effects of Large Language Models on Human Creative Ideation", Creativity and Cognition 2024
[3] Yu, Chen, Hu, Suri & Counts, Adoption of Generative AI in the Workplace, arXiv preprint, Aug 2026.
[4] Pew Research Center, "Young adults in the U.S. are increasingly wary of AI", Aug 18, 2026. Survey June 22-28, 2026, 3,488 adults