Leadership in the Digital and AI Era
Digital Leadership
What is digital leadership?
- It’s about judgment, rules, and culture, not technical skill
- Encouraging use while punishing mistakes drives secret use
- Define what a person must verify
In organizations where digital and AI leadership has taken hold, even the first question in a meeting is different. Instead of 'Who owns this?', they ask 'How much do we hand to the machine, and whose name is on it when it's wrong?' What leaders gain is not the power to choose tools but an authority allocation chart. Once it is written down that some decisions are open only up to the draft stage and others are closed entirely, staff no longer have to read their boss's face every time.
A useful yardstick for dividing work is “can it be undone?” Tasks you can simply redo, like drafting copy or summarizing materials, can be opened up freely. Tasks where undoing a mistake requires an apology, like notifying a customer or finalizing a unit price, stay in human hands. For the gray zone in between, use two layers: the machine drafts, a person finalizes. This concept is often confused with digital literacy, an individual’s tool proficiency, and with IT governance, which covers access rights and security, but it works at a different level. Training and policy can solve the first two, but agreeing on standards has no substitute: leaders have to sit down and set them themselves.
A learning culture is proven by the calendar, not by declarations. Fix the first 10 minutes of the monthly meeting as 'sharing wrong answers' and have an executive go first with their own failure, and that slot fills up within three months. Ask people to present only success stories, and it's empty by month two. A leader who tries the tools personally and talks about their limits sends a stronger signal than any internal announcement. Verification rules also get followed only if they are cut down to one or two lines. The moment a ten-line guideline is deployed, nobody reads it.
The most common misconception is that leaders have to learn coding or prompting. Start there, and executive training turns into a tool demo while the real question, 'What will humans verify?', never gets answered. Even riskier is an organization that tells people to use AI freely and then punishes the user when something goes wrong. The tools don't disappear. They move to personal accounts. Reported usage goes up, customer data gets pasted into outside services, and that only comes to light after an incident.
Ara LimExpert in global leadership, AI, and business communication
John KimInstructor John Kim, author of B2B Marketing Design and CEO of Nanum ConsultingThis concept took shape as research accumulated showing that, in digital transformation (DX), leadership and organizational culture matter more than technology adoption. Since generative AI went mainstream, setting decision criteria and building a learning culture have become its core elements.
Consider a hypothetical case: Company A, an industrial parts manufacturer with 320 employees. Six months after rolling out company-wide generative AI accounts, only two teams, Planning and Marketing, were actually using them, and the security team had caught two cases of quotes pasted into personal accounts. Leadership decided to run a three-month leadership program for team leaders.
- Before designing the training, the head of Management Support sent an anonymous survey to 18 team leaders. Eleven said they were already using AI on personal accounts.
- The CEO put this number on the screen at the first workshop and made it clear: 'I won't hold secret use against anyone. But let's decide together how far we go.' The tension in the room eased.
- The purchasing manager and the CS manager each ranked their 12 tasks by the cost of reversing a wrong result. For three of them (contract wording, final unit pricing, and final replies to claims), they ruled out AI drafts altogether.
- The verification rules were cut to two lines. The team lead checks any document going outside the company against the source. Internal materials need only the author’s check, with the condition that checking never takes more than 10 minutes per item.
- The first 10 minutes of the monthly meeting became a “strange answers” segment, and in the first session the head of production went first, sharing his own case of copying an inventory figure incorrectly.
- Three months later, the number of teams actually using the company-approved tools rose from 2 to 14, and the number of people reporting personal-account use fell from 11 to 2. The time to draft a quote also dropped from an average of 50 minutes to 18 minutes.