Thought leadership
Human Judgement in the AI Era
Ewen Fleming & Jack Merriman
AI is making intelligence abundant.
Human judgement is becoming the defining leadership skill.
As large and small language models (LLMs and SLMs) become embedded across business functions, an important shift is taking place. Increasingly, organisations are relying on machines not just to automate tasks, but to shape thinking itself. Activities that once demanded extensive research, interpretation and experience — thought leadership, market analysis, note taking, strategic planning and even performance reviews — can now be produced in seconds.
AI can dramatically accelerate analysis, but it cannot replicate human judgement. That distinction is becoming increasingly important as leadership teams navigate geopolitical instability, regulatory fragmentation, sovereign AI agendas, supply chain disruption and declining trust in government, major organisations and technology ecosystems.
In this environment, leadership is becoming less about having the answer and more about understanding which signals matter, which risks are emerging, and which decisions should remain fundamentally human. The challenge leaders face is no longer gaining access to information but instead when to trust the data, when to challenge it, and when to slow down despite pressure to move faster.
“Research from Microsoft and Carnegie Mellon University in 2025 found that increased confidence in generative AI tools often correlated with reduced critical thinking effort among knowledge workers.”
The future leader will therefore need to master three critical capabilities.
1. Discernment: separating insight from noise
LLMs and SLMs can generate impressive outputs at scale, but volume does not equal value. Future leaders must develop the ability to distinguish credible insight from plausible-sounding fusion of data from often multiple and varied sources.
This leadership capability needs to be built through exposure to diverse situations, active curiosity and deliberate challenge not just past experience when many aspects are changing rapidly. Leaders should regularly test assumptions, seek opposing perspectives and compare AI-generated conclusions with real-world outcomes. Confidence comes not from always being correct, but from learning how to evaluate and handle ambiguity with discipline.
2. Human judgement: understanding consequences beyond the data
AI is highly effective at statistical pattern recognition, but weaker at understanding real-world context, politics, culture and unintended consequences. Human judgement remains essential when decisions affect people, trust, reputation or long-term strategic positioning.
Leaders need to build this capability through experience and reflection. Cross-functional exposure, international environments, crisis situations and difficult stakeholder decisions all sharpen contextual awareness. Mentorship and honest post-decision reviews are equally important because judgement is refined through examining both successes and mistakes.
3. Adaptive thinking: knowing when to challenge the machine
As AI becomes more embedded in workflows, there is a growing risk of passive acceptance. The strongest leaders will not simply become better users of AI, they will know when to question its framing entirely.
This requires intellectual independence and confidence, and leaders can strengthen this by practising scenario thinking, debating alternative futures and intentionally exploring “what might we be missing?” and not just being focused on ‘speed to output’. The ability to introduce new angles, unconventional thinking and human nuance will increasingly differentiate leadership quality.
AI can inform decisions, but accountability cannot be delegated to a model. Leaders remain responsible for the consequences of decisions, regardless of how sophisticated the system supporting them becomes.
The organisations that benefit most from AI will not necessarily be those with the largest models or fastest deployment cycles. They will be those that build cultures where AI outputs are routinely challenged, assumptions are debated openly and human accountability remains clear.
Conclusion
The dangers of hallucinations and unverified AI outputs are already well discussed. Less discussed is the growing importance of the human contribution around the model: the art of prompting, critical review, interpretation and reframing. The value no longer sits solely in generating answers, but in shaping better questions and applying experienced judgement to the output.
Ultimately, the competitive advantage of future leaders is unlikely to be their ability to use AI faster than others. Instead, it will be their ability to combine AI capability with human wisdom, discernment and accountability. This is a new age and leaders can’t rely on just recognising patterns because they have seen them before, instead, they will learn as they negotiate complexity and uncertainty.
Judgement becomes the premium skill. The paradox of the AI era is becoming clear: as machine intelligence becomes more abundant, human judgement becomes more valuable.