The roadmap nobody saw coming

Emmett, the chief information officer, presented a new AI roadmap to the executive team without consulting Sara, the CHRO, about workforce impact or reskilling timelines. She learned about it in the meeting.

Six months later, everything stalled

Adoption was low and employees were anxious. Emmett blamed people issues slowing his technology initiative. Sara felt handed a problem instead of a partnership. CEO Denise had to resolve their conflict.

AI created contested terrain

According to UC Berkeley professor Toby Stuart, AI has created contested terrain at the leadership level. Every function has a legitimate claim to AI decisions, and the old equilibrium no longer holds.

“contested terrain”

The real problem is behavioral

AI requires leaders to share authority and co-own resources across functions. This fundamentally different approach requires leaders to move beyond behaviors that got them to the C-suite.

Both executives had the right expertise

Emmett understood technological requirements for implementing AI. Sara understood what was feasible from a talent perspective. Their conflict arose because neither had the full picture about what implementation needed.

The CEO's real job on AI

The CEO cannot own AI, but must own the conditions under which others co-own it. This means encouraging executives to move away from wrong behaviors to right ones with specific practices.

The ownership instinct backfires

Executives rise by taking ownership and delivering results in their domains. They become territorial, believing that bringing others in can diminish their leadership and slow essential processes.

“very territorial”

Hoarding kills a thirty million dollar project

At a global bank, a chief data officer's team prototyped a thirty thousand dollar AI tool to automate legal memoranda and save hundreds of staff hours.The vendor was on the CTO's cost-cutting list.

The expertise default: not my job

Other leaders disengage from AI decisions altogether. Some refuse to build basic AI fluency, assuming understanding the technology is the CTO or CIO's responsibility, not theirs.

Abdicating authority undermines shared authority

No single executive can see the full organizational picture needed to identify opportunities, risks, and impact when disruptive technology like AI arrives. Abdication is as damaging as grabbing control.

Compliance and fear blocked exploration

At an entertainment company, the general counsel reflexively blocked AI pilots seeing only compliance risk. The CHRO resisted engaging with technology he feared could harm artists and workers. Exploratory projects never got off the ground.

Define the problem, not the owner

Begin AI initiatives by defining the business problem you are trying to solve, unlimited by organizational boundaries. Get specific about the product, process, or workflow and the outcome you want to achieve.

Early conversations reveal hidden assumptions

Sandy Carter, CEO of Equs.ai and former senior executive at IBM and AWS, said when two executives work together they discover different assumptions about risk, what good looks like, and who is accountable.

“When two executives sit down and do the work together, they almost always discover they had different assumptions about where the risk is, what good looks like, or who's accountable when something goes wrong.”

Curate the problem before the room

Shannon Kennedy, former chief compliance and privacy officer in healthcare who chaired C-suite AI adoption groups, said to curate the problem before you curate the room. Map stakeholders whose expertise the problem requires.

“curate the problem before you curate the room”

Build trust before the stakes are high

CEOs should identify cross-functional relationships their AI strategy depends on and deliberately invest in building trust before stakes are high. This means in-person social time, setting expectations of shared decision rights, and modeling open communication.

Make early plans experiments, not final

CEOs can make it easier for executives to share authority by lowering stakes of AI decisions.Make clear that early AI plans are experiments.

Tie incentives to shared outcomes

Leadership responsibilities on AI shift over time, so incentive structures must reward shared outcomes, not individual territory. Shared mandates, KPIs, bonus and equity structures, and performance reviews must be tied to jointly owned results.

Establish rules for resolving disagreement

Even with right conditions for shared authority, teams will experience conflict. Establish a friction framework: agreed-upon rules and norms for resolving disagreements, including decision rules, conversation protocols, and escalation norms.

Collaboration created unexpected value

A leading entertainment company's COO required full executive team engagement in AI decisions. The conversations surfaced emotional blockers, built consensus skills, and created greater trust, leading to stronger teamwork and faster decisions.

“I now realize that the requirement for full cross-functional investment was probably creating a much greater level of value than what I anticipated”