What Leadership Requires When Control Is Impossible

Responsible Stewardship of AI | 1 of 15

Image by Russell E. Willis

AI leadership is usually framed as a problem of control. Leaders are told to control the data, access, risks, outputs, vendors, and employee use. Much of that work is necessary.

None of it gives us control.

No leader, organization, or governance system can fully control AI’s development, use, interactions, or consequences. Models behave differently as data, incentives, workflows, users, and circumstances change. Tools acquire new purposes as they spread. Effects emerge far from the people who authorized the system—and sometimes long after the decision that set them in motion.

The task is not to abandon governance, but to replace the aspiration to control with the practice of stewardship.

Stewardship is not passivity. It does not excuse leaders from setting limits, assigning authority, demanding evidence, monitoring consequences, or stopping a system. It makes those obligations more demanding by refusing to confuse governance with mastery.

I call this Responsible Stewardship of AI: the sustained practice of remaining answerable for power whose consequences no one can fully predict or control.

That begins with a different leadership question:

How do we practice responsibility when control is impossible?

This series develops two parts of the answer.

The ASSUME Model—Anticipation, Scope, Systems Awareness, Uncertainty, Moderation (Self-limitation), and Expertise—cultivates the judgment leaders need when power exceeds prediction and control.

But responsible judgment is not enough. Leaders and organizations also need structures that allow what they see and judge to shape what the organization actually does. The Five Structures of Responsible Stewardship give responsibility that institutional form.

The two movements belong together. ASSUME helps leaders see and judge responsibly. The Five Structures help organizations act on that judgment. Neither promises control. Together, they make responsibility practicable.

Over the next fourteen posts, I will develop the framework one dimension at a time. Each will end with a Monday Morning Question—not a slogan or compliance box, but a question leaders can bring to an actual AI decision.

The aim is not to solve responsible AI in a LinkedIn series. It is to make responsibility more concrete wherever AI is shaping what organizations do and what they are becoming.

The Monday Morning Question

Where are we relying on the language of control to avoid naming the responsibility we must continue to practice when control fails?

Responsible Stewardship begins when leaders accept that uncertainty will remain—and build the judgment and structures required to remain answerable anyway.

                            *************************************

This series draws upon my book, AI and the Crisis of Control.

#ResponsibleStewardship #ResponsibleAI #AILeadership