By Greg Shewmaker
Artificial intelligence sits at the center of how leaders design teams, set strategy, and measure success. Across industries, executives are reorganizing divisions and redefining roles in a rush to hand over every task that looks automatable to AI agents. If an algorithm can draft proposals, screen resumes, or manage customer outreach, why keep people in the middle?
The logic sounds unassailable. Machines don’t sleep, complain, or ask for raises. But efficiency isn’t the same as strategy. Leaders owe their organizations, and the people who keep them running, an honest look at what’s being overlooked in this rush to automate.
The Irreplaceable Human Spark
AI can parse patterns faster than any human team. What it cannot do is imagine what doesn’t yet exist or rally a group when circumstances turn. Research published in MIT Sloan Management Review (“Does GenAI Pose a Creativity Tax?”) shows that lasting competitive advantage depends on human creativity, and creativity requires empathy and judgment that no model can reproduce.
When a company is faced with a crisis, such as a product recall, a major cyber incident, or a supply chain collapse, it isn’t algorithms that rebuild trust with employees or customers. It’s leaders who steady workforces, reassure stakeholders, and inspire confidence in uncertain conditions. AI may calculate probabilities, but it cannot project belief, purpose, or shared vision. Those are human traits, and in critical moments, they are the difference between collapse and recovery.
Reliability and Trust: Bias, Blind Spots, and Errors
AI systems also absorb the flaws of the data they are trained on. A 2025 London School of Economics and Political Science study reported by The Guardian (“AI Tools Used by English Councils Downplay Women’s Health Issues, Study Finds”) showed how tools used by English councils downplayed women’s health issues compared to men’s. The finding underscores a broader truth: When flawed data feeds these systems, flawed decisions follow.
It is easy to see how this pattern could manifest in other ways, from hiring systems that downgrade nontraditional career paths to lending models that penalize minority applicants. When managers outsource judgment to algorithms without oversight, they may miss key opportunities and risk eroding trust among both employees and customers.
Even the best models misfire. Hallucinated answers, shaky reasoning, and breakdowns under new conditions are not rare glitches. They are expected behaviors. A 2025 Live Science report (“AI Hallucinates More Frequently as It Gets More Advanced — Is There Any Way to Stop It from Happening, and Should We Even Try?”) noted that the most recent reasoning-focused models from OpenAI, o3 and o4-mini, hallucinated 33% to 48% of the time. In healthcare, aviation, or finance, one mistake can cascade into safety hazards, lost livelihoods, or market instability. Even in less critical settings, minor errors damage trust and disrupt the customer experience.
Executives often describe AI failures as “outliers,” but those outliers come with a predictable frequency. That should reframe how leaders design safeguards. When the risk of error is baked into the system, human oversight is essential.
Why So Many AI Projects Fall Short
Breakthroughs dominate the headlines, but inside most organizations, the story is frustration. A recent MIT Sloan report covered by Fortune (“MIT Report: 95% of Generative AI Pilots at Companies Are Failing”) found that 95% of generative AI pilots fail to deliver meaningful returns. The problem lies in the gap between what AI promises and what it actually delivers. Leaders are told it’s both a world-changing breakthrough and a low-cost replacement for human labor. That contradiction drives fear-based adoption: fear of missing out, of becoming obsolete, of falling behind.
Leaders often leap in without a clear plan, assuming AI can be rolled out like any other software with immediate payoff. In reality, organizational complexity makes adoption much more complicated, and many companies are instituting cost-cutting because the more challenging strategic questions remain unanswered.
Instead of pausing to understand how AI is reshaping the competitive landscape and clarifying how their organizations should respond, many leaders rush straight into deployment. By skipping the more complex strategic work, they treat AI as a quick fix rather than a transformation. That shortcut leaves initiatives without a clear foundation, so they look promising on paper but rarely scale or endure.
Automation’s Economic and Ethical Risks
Disruption and inequality. It’s easy to frame AI adoption as purely cost-saving. Replace a role here, a department there, and the numbers look good. But management cannot confuse short-term efficiency with long-term strength.
An International Labour Organization 2025 report (“Generative AI and Jobs: A Refined Global Index of Occupational Exposure”) estimated that one in four jobs worldwide could be exposed to generative AI, rising to one in three in high-income countries. Middle-skill roles, once the foundation of stable employment, are the most vulnerable. When those jobs disappear without alternatives, communities suffer and companies lose seasoned talent. The ripple effects hit consumer demand, turnover costs, and institutional knowledge.
Consider regional economies built on steady, middle-income work, such as insurance adjusters, customer service specialists, and paralegals. When those roles vanish, the immediate impact is unemployment. The long-term effect is a weaker customer base and lower demand for the very products AI was supposed to deliver more efficiently.
The ethical minefield. Delegating decisions to machines raises an uncomfortable question: Who is accountable when AI makes a mistake? Regulators are already pressing for answers. The European Union’s AI Act (“Regulation (EU) 2024/1689 laying down harmonized rules on artificial intelligence”) mandates stringent standards for transparency and accountability, with obligations increasing in proportion to the risk level.
Companies that stumble face reputational damage, legal exposure, and regulatory fines. The speed of consequences often outpaces the speed of correction. When managers treat AI outputs as final answers rather than inputs to human judgment, they risk making mistakes and abdicating the trust placed in them as leaders.
Management in the Age of AI: A Human-Centric Approach
The agent-only vision misses the essence of management. Real leaders don’t just allocate tasks. They create environments where people thrive, innovate, and commit to a purpose greater than themselves.
The real opportunity is not to replace people but to make them more capable. That shift requires a new playbook:
Coach, don’t boss. Employees don’t need reminders that AI works faster. They need leaders who help them adapt. As Harvard Business Impact observed (“AI-First Leadership: Embracing the Future of Work”), management is moving from directing tasks to enabling human and AI collaboration. Equipping managers with AI literacy ensures that they can guide teams without ceding control.
Build trust through transparency. Explaining how AI works, where it falls short, and how it complements human judgment helps reduce fear and build engagement. Some organizations now publish clear guidelines for AI use and hold open forums for employees to raise concerns. These actions aren’t cosmetic. They set cultural expectations for openness.
Integrate ethics into daily work. Efficiency and fairness collide every day. Ethical choices cannot be relegated to annual training. They must be practiced continuously. When a manager decides how to allocate AI-generated leads, for example, they are making both a business and an ethical choice.
Lead across generations. In 2025, Pew Research (“How the U.S. Public and AI Experts View Artificial Intelligence”) found optimism among AI experts and deep skepticism among the public. Managers can bridge that divide by fostering shared learning and dialogue. In practice, pair younger employees who are enthusiastic about AI with veterans who know the organizational context and customer history.
Stay adaptable. McKinsey’s 2023 resilience survey (“The State of Organizations 2023: Ten Shifts Transforming Organizations”) found adaptability to be a defining trait of successful organizations. The pace of AI change makes this skill non-negotiable. Leaders must normalize experimentation, learn from failures, and treat adaptability as a core competence, not an afterthought.
Looking Ahead
The hype of AI’s potential will continue to grow. New models will launch, investment will surge, and headlines will predict the end of management. I don’t buy it. Leading in the AI era is not about handing control to agents. It is about guiding teams through uncertainty, balancing human and digital contributions, and keeping purpose at the center.
The leaders who succeed won’t be those who chase automation for its own sake. They will be the ones who recognize the enduring value of human judgment and the responsibility that comes with it.
We don’t face a choice between people and machines. We face a choice between leaders who abdicate responsibility and leaders who rise to it. A practical starting point is to audit how AI is used inside the organization, convene cross-functional teams to weigh risks and opportunities, and establish regular ethics reviews. When we build these habits, we create organizations that prepare managers to lead in an AI-enabled future.
If there’s one lesson to carry forward, it’s that technology rarely rewrites the fundamentals of leadership. Every era brings tools that promise to change the game. What separates the durable organizations from the fragile ones is not the toolset, but rather the mindset of their leaders. The AI era will be no different. Managers who embrace that truth will not only keep their organizations resilient, but they will also keep them human.
Greg Shewmaker is CEO of r.Potential, which is building an enterprise OS that deploys digital workers that learn to think, collaborate, and operate as true co-workers: grounded in context, transparent in reasoning, accountable and secure by design, and built to amplify the things that make humans extraordinary.