- AI pays off where it is led, not where it is bought. Nearly nine in ten organizations in McKinsey's 2026 survey use AI regularly, but only 37% say it has added anything to their earnings.
- The manager is the strongest lever. Employees whose manager actively supports AI use are 2.1 times as likely to use it frequently, yet most employees can't say they have been given a clear plan.
- A leader's own habits are a poor guide. Leaders use AI far more than the people they lead, and the gap is widening, so the leadership team is the wrong place to judge how far adoption has spread.
- AI changes what leaders decide, not whether they decide. Faster analysis, collaboration tools, forecasts, employee data and automation each leave a decision with the leader that the tool can't make.
- In Kenya, signing off an AI output unexamined is a legal risk. The Data Protection Act and the ODPC's July 2026 guidance expect a person to review AI-driven decisions that significantly affect people, and treat AI in hiring and staff monitoring as high-risk.
- Five practices turn tools into adoption. Write the plan down, use the tools where your team can see you, redesign one workflow, decide where a person reviews, and train managers first.
The hard part of AI in leadership isn't choosing the tools. It is deciding who acts on what they produce. Most leaders now have software that drafts, summarizes, forecasts and scores, even if it is only a general assistant such as ChatGPT. Far fewer have an agreed answer to the questions that follow: which outputs can be used as they are, which need a person to check them, and who that person is.
In our work with senior leadership teams across East Africa over the past decade, we have seen technology rollouts succeed or stall for reasons that had little to do with the technology. AI is following the same pattern. The tools have arrived, and are still evolving, faster than the decisions about how to use them, and the people furthest from the leadership team are often left to work those decisions out alone.
The usual list of AI's benefits for leaders is accurate as far as it goes: better decisions, better collaboration, sharper forecasts, a better employee experience and less routine work. What it leaves out is that each one hands the leader a decision that didn't exist before.
What AI in Leadership Means
AI at work now means two kinds of tools. The first is generative assistants such as ChatGPT, Claude, Gemini and Microsoft Copilot, used on their own or built into email and office software. The second is the older machine learning systems that score, rank and forecast. AI in leadership covers two different jobs, and much of the confusion about it comes from treating them as one:
- Using AI in your own work: drafting, summarizing, analyzing data and testing options before a decision.
- Leading your team through adopting it: deciding what the tools are for, who uses them on what, where a person checks the output, and how people learn.
The first is a personal skill, and most leaders pick it up quickly. The second is a leadership task, and it is the one the evidence links to results. A leader can be a skilled user and still lead adoption badly, usually by assuming that what works for them will spread on its own.
Why Leadership Decides Whether AI Pays Off
Getting organizations to use AI is no longer the problem. In McKinsey's State of AI survey for 2026, which drew 1,719 respondents from 97 countries, nearly nine in ten said their organization uses AI regularly in at least one business function. Eight in ten said it had improved their own productivity. Yet only 37% said AI had contributed anything at all to their organization's earnings before interest and taxes, about the same share as a year earlier. The gains are real for individuals, but they are not reaching the organization.
The exceptions are a small group McKinsey calls high performers, about 6% of respondents, who attribute at least 5% of those earnings to AI. What sets them apart is how they work, not what they bought. Nearly three-quarters have fundamentally redesigned workflows because of AI, against one-quarter of everyone else. BCG put a ratio on the same idea in its 10-20-70 rule: spend 10% of your AI effort on algorithms, 20% on technology and data, and 70% on people and processes.
McKinsey's 2026 survey found its AI high performers twice as likely as other organizations to say their senior leaders demonstrate commitment to AI initiatives, and twice as likely to have defined processes for measuring the impact of those initiatives. Neither is a technology purchase. Both are things a leadership team decides to do.
Source: McKinsey, The State of AI in 2026: On the Road to ROI
Inside the organization, the lever is the manager. In a 2025 Gallup analysis of a random sample of 19,043 US employees, those in organizations investing in AI who strongly agreed that their manager actively supports their team's use of AI were 2.1 times as likely to use it a few times a week or more, and 6.5 times as likely to strongly agree that the tools they had been given were useful for their work.
The people at the top are not the ones who need persuading. Gallup's quarterly data shows how far leaders have pulled ahead of the people they lead.
Frequent AI Use at Work by Role, US Employees, 2023 vs 2025
Share of each role using AI a few times a week or more, second quarter of 2023 against fourth quarter of 2025. Source: Gallup, Frequent Use of AI in the Workplace Continued to Rise in Q4 (2026)
Leaders 17% to 44%, managers 15% to 30%, individual contributors 9% to 23%. The gap between leaders and individual contributors grew from 8 points to 21.
Each bar is the share within that role, not a share of all employees. Gallup defines leaders as managers of managers.
Read together, the figures say three things:
- Leaders use AI the most. Between mid-2023 and late 2025, the share of leaders using AI a few times a week or more rose from 17% to 44%. Among managers it rose from 15% to 30%, and among individual contributors, the people doing most of the day-to-day work, from 9% to 23%.
- The gap is widening. In 2023, leaders were 8 points ahead of individual contributors. By late 2025 they were 21 points ahead. Use is rising at every level, but leaders have gained the most ground.
- A leader's own habits are a poor guide to the organization. A leadership team that judges adoption by how it works itself is looking at the heaviest users in the building, not at the typical employee.
And the plan rarely travels down. As of May 2026, only 25% of US employees said their organization had communicated a clear plan for integrating AI into its work. Three in four could not say they had been given one.
In Gallup's 2025 data, employees who strongly agreed that their leadership had communicated a clear plan for integrating AI were three times as likely to feel very prepared to work with it, and 2.6 times as likely to feel comfortable using it in their role. A plan isn't paperwork that follows adoption; it helps drive it.
Source: Gallup, AI Use at Work Has Nearly Doubled in Two Years
Gallup's figures are for the US. We found no survey of comparable size that measures AI use in Kenyan workplaces, so the pattern is one to check for in your own organization rather than assume.
5 Ways AI Is Changing Leadership Work
The five changes most often promised to leaders are real. Each one also leaves a decision with the leader that the tool cannot make.
1. Enhanced decision-making
AI can pull together the analysis behind a decision in minutes: summarizing a long report, comparing supplier quotes, drafting the options paper. Half of McKinsey's 2026 respondents said AI helps them make better decisions. What stays with the leader is the judgment the analysis can't contain: how much risk the organization will accept, who bears the cost, and whether the question being analyzed was the right one. AI makes the options paper faster. It doesn't make the choice.
2. Improved collaboration tools
Meeting transcripts, shared assistants and automatic summaries mean fewer people need to be in the room for a record to exist. The leader's decision is what a summary is allowed to replace. A summary records what was said. It rarely records who disagreed, how strongly, or what nobody said, and those are often the things a leader most needs to know.
3. Predictive analytics for strategy
Forecasting models for demand, cash flow, staff turnover and credit risk are no longer confined to banks and telecoms. But a forecast is the past projected forward. The leader's job is deciding when the past has stopped being a good guide. After a tax change, a currency swing or a new competitor, a model will keep producing confident numbers from conditions that no longer hold.
4. Employee experience optimization
AI tools now screen job applicants, answer HR questions, recommend training and, in some workplaces, measure productivity. This is where AI touches people most directly, and in Kenya it is an area the law addresses directly. The leadership decision is what employee data is used for and who sees the conclusions. Both answers should be settled, and explained to staff, before a tool goes live.
5. Automation of routine tasks
Standard letters, invoice matching, scheduling and first-line customer queries are increasingly automated. The freed time is the point, and it is the part most often wasted. Unless a leader decides what the time is for, it fills with more of the same work, and the productivity gain individuals report never shows up in the results, which is exactly the gap in McKinsey's numbers.
What Kenyan Law Expects of Leaders Using AI
For Kenyan organizations, who makes the decision is a legal question as well as a management one. Section 35 of the Data Protection Act, 2019 gives every person "a right not to be subject to a decision based solely on automated processing, including profiling, which produces legal effects concerning or significantly affects the data subject." The Act allows three exceptions:
- Contract: the decision is needed to enter into or perform a contract with the person.
- Law: a law authorizes the decision and sets out safeguards for the person's rights.
- Consent: the person has consented to it.
Where such a decision is made solely by automated processing, the organization must tell the person in writing, and the person can ask for it to be reconsidered or taken again by a human.
In July 2026 the Office of the Data Protection Commissioner (ODPC) published a Guidance Note on Artificial Intelligence that spells out what this means in practice. Its background section notes that Kenyan employers already use algorithmic tools to screen job candidates. Two points matter most for leaders:
- A rubber stamp is not human involvement. The guidance defines automated decision-making as deciding "solely by automated means, without any meaningful human involvement." A manager who approves an AI recommendation without examining it would struggle to show that their involvement was meaningful.
- Hiring and staff monitoring are high-risk. The guidance lists AI in employment screening and monitoring as a high-risk use. That requires a data protection impact assessment, disclosure of AI screening tools to candidates and employees, human review of AI-recommended hiring or termination decisions, and a right to contest them. Its screening table marks AI-based employee productivity monitoring as needing an assessment too.
The ODPC's guidance also closes off an obvious shortcut. It says consent "is not an appropriate lawful basis" where there is a significant power imbalance, and gives "employer-employee AI monitoring" as its example. A signed consent form doesn't settle it, because an employee is rarely free to refuse their employer.
Source: Office of the Data Protection Commissioner, Guidance Note on Artificial Intelligence (July 2026)
This puts pressure on a familiar habit. In many organizations across East Africa, deference to seniority means a recommendation that arrives with a senior manager's approval is rarely questioned further down. A recommendation from a system the senior manager chose can carry the same authority. The law now asks for the opposite: a person who reviews the output, can explain the decision, and is willing to overrule the tool.
Leading a team through AI adoption?
The Leadership in the Age of AI and Digital Transformation module of our Leadership and Organizational Change Management program covers AI fundamentals for leaders, human-centered AI leadership and leading teams through digital change. Request a proposal.
How to Lead Your Team's AI Adoption
1. Write the plan down
In most organizations the AI plan lives in the leadership team's heads. Put it on one page and share it with everyone: what AI is for here, what it isn't for, which tools are approved, what information must never go into them, and who to ask. A plan people can read is what lets a manager answer the questions their team will ask.
2. Use the tools where your team can see you
Telling a team that AI use is encouraged achieves little if they never see their manager use it. Show the work instead: share a first draft and say what the tool produced and what you changed. That one habit gives permission and sets the standard at the same time. It also makes the manager the person people ask, which is the support Gallup's figures point to.
3. Redesign one workflow
Adding a tool to an unchanged process speeds up the same steps. Redesigning the process is what most of McKinsey's high performers have done. Pick one workflow that matters, such as the monthly management report or recruitment shortlisting, and rebuild it from the start with AI in mind: which steps disappear, which change, and where the checks sit.
4. Decide where a person reviews
For each use, decide in advance whether the output goes out as it is, is checked first, or only informs a person who decides. Write that down next to the plan. In Kenya, hiring and termination decisions belong in the last group, since the ODPC guidance expects a person to review them.
5. Train managers first
Managers are the ones who connect a plan to the work their teams actually do, so they need to understand it before anyone else. The gap below them is wide. In BCG's 2025 AI at Work survey of more than 10,600 employees in 11 countries and regions, more than three-quarters of leaders and managers said they used generative AI several times a week, but regular use among frontline employees had stalled at 51%.
BCG's 2025 AI at Work survey found that the share of employees who feel positive about generative AI rises from 15% to 55% when leaders show strong support for it. Only about a quarter of frontline employees said they received that support, which leaves most frontline staff without it.
Source: BCG, AI at Work 2025: Momentum Builds, but Gaps Remain
Three Common AI Leadership Mistakes
Underestimating how much your people already use AI
In a McKinsey survey of US executives and employees in late 2024, employees were three times as likely to be using generative AI for at least 30% of their daily work as their executives expected. Policies written for adoption that hasn't happened yet arrive after it has, and unplanned use can end up on tools nobody approved.
Buying tools before deciding what they're for
When Gallup asked US employees for the biggest barrier to AI adoption at their workplace, the most common answer was an unclear use case or value. Among employees who don't use AI, 44% said the main reason was that they didn't believe it could help with their work. A license nobody can connect to their job is a cost, not an adoption.
Treating the recommendation as the decision
An AI recommendation is an input. When it becomes the decision by default, nobody can explain the outcome to the person it affects, and "the system said so" satisfies neither an employee asking for a review nor the ODPC. The fix is the fourth practice above: decide in advance where a person reviews, and make sure that person actually does.
Frequently Asked Questions
What does AI in leadership mean?
It covers two different jobs. The first is using AI tools such as ChatGPT, Claude, Gemini or Microsoft Copilot in your own work: drafting, summarizing, analyzing data and testing options before a decision. The second is leading your team through adopting them: deciding what the tools are for, who uses them on what, where a person checks the output, and how people learn. The first is a personal skill most leaders pick up quickly. The second is a leadership task, and it is the one the evidence links to results.
Does AI actually improve business results?
For individuals, usually; for organizations, much less often. In McKinsey's 2026 State of AI survey of 1,719 respondents in 97 countries, nearly nine in ten said their organization uses AI regularly and eight in ten said it had improved their own productivity, but only 37% said it had contributed anything to earnings. The roughly 6% that McKinsey calls high performers are far more likely to have redesigned their workflows around AI rather than adding it to existing ones.
How is AI changing the work of leaders?
In five main ways, and each one leaves a decision with the leader. AI speeds up the analysis behind decisions, but the leader still makes the choice. Collaboration tools record what was said, but not who disagreed. Forecasts project the past forward, so the leader has to judge when the past has stopped being a guide. Employee-facing tools raise the question of what staff data is used for. And automation frees time that someone has to decide how to use.
Can a Kenyan organization let AI make decisions about people?
Only within limits. Section 35 of Kenya's Data Protection Act gives people a right not to be subject to decisions based solely on automated processing that significantly affect them, with exceptions for contracts, decisions authorized by law and consent. The ODPC's July 2026 guidance defines automated decision-making as deciding without any meaningful human involvement, and expects human review of AI-recommended hiring and termination decisions. Approving a recommendation without examining it is unlikely to count as meaningful involvement.
Do Kenyan employers need staff consent to monitor them with AI?
Consent is not the answer. The ODPC's 2026 Guidance Note on Artificial Intelligence says consent is not an appropriate lawful basis where there is a significant power imbalance, and gives employer-employee AI monitoring as its example. The guidance treats AI in employment screening and monitoring as high-risk, which means a data protection impact assessment, disclosure of the tools to candidates and employees, human review of hiring and termination decisions, and a right to contest.
How should a leader lead a team's AI adoption?
Support it visibly, and decide what the tools are for before buying more. In a 2025 Gallup analysis of 19,043 US employees, those in organizations investing in AI who strongly agreed their manager actively supports their team's use of AI were 2.1 times as likely to use it a few times a week or more. In practice that means five things: write a one-page plan, use the tools where your team can see you, redesign one workflow that matters, decide in advance where a person reviews the output, and train managers before everyone else.
Conclusion
AI gives leaders more analysis, more forecasts and more time than they have had before. It doesn't decide what any of that is for. That choice stays with the leader, and it is what determines whether AI pays off.
Which decisions in your organization would you let an AI tool make on its own, and has anyone written that list down?
If your leadership team is working out how to lead AI adoption, the AI module of our Leadership and Organizational Change Management program is built for it. Request a proposal and we will start from how your teams use AI today.
