Recently, I've been attending a number of conferences on AI, listening to CIO's, CTO's through to People and Operations leaders talking about the challenges and opportunities of integrating AI in their organisations. The repeated theme from speakers was that human judgement is the answer to identifying the best AI integrated opportunities and preventing AI risks. I ended up on stage sharing our perspective on why human judgement is much more than critical thinking. This article shares some of our thinking.

A new working relationship with AI is evolving. AI now gathers information, sees the patterns and provides insight. We, humans, then review, interpret and apply the judgement that ensures the outputs (and our uses of AI) are safe, ethical, and high value.

AI’s insights are generated from the data it is trained on and the information we give it, rather than from lived experience of the world. Human judgement is where we connect those outputs to our reality: deciding what they mean in context, whether to act on them and what consequences we are prepared to accept. That makes judgement a critically important capability for us as individuals and as leaders of organisations.

At times, it seems we assume human judgement is something we naturally possess; that, faced with an AI-generated answer, we will know what is accurate, what matters, and what to do next.

However, human judgement is not an inherent trait. It is a learned capability. It is our capacity to make responsible sense of a complex situation: to frame the right question, test the information in front of us, recognise our own assumptions and emotions, bring together different perspectives, weigh competing consequences and act within uncertainty.

I’m incredibly positive about the potential of AI when paired with effective human judgement; however, I’m also relatively critical of how good our collective human judgement is at this point. The following three lessons for leaders share some of our foundational principles in building effective human judgement.

Lesson 1: Apply human judgement to the higher order of ethics and purpose ahead of being AI’s quality assurance

This might seem obvious, but it’s harder than it sounds. Our human judgement is deeply entangled with how we use AI. Already in many industries, there is no single place in a workflow or decision where AI’s input ends and human judgement begins; like it or not, AI technology is part of our human thinking ecosystem, and we cannot separate from it to interrogate it.

And yet we humans are the decision makers, ethical guardians, and vision shapers of how we use AI most brilliantly in service to the highest potential of our humanity (and, on a more day-to-day level, our business outcomes).

Human judgement has a deep history in philosophy, ethics, critical thinking, systems thinking and decision science. But AI changes the context in which we practise it. We are now presented with immediate, plausible and confidently expressed answers before we have necessarily done the work of forming our own view.

Many feel the brain drain of being AI’s error checker; there is little engagement, and the machine and system have framed the judgement context. So, our first task in establishing effective human judgement is to define when and where we use it. Are we using our human judgement at a higher-order level, or are we reduced to evaluators on a conveyor belt of decisions, offering quality assurance?

Lesson 2: Human judgement is a capability and mindset we learn, not a given trait

Empathy was once widely treated as a personal quality. You were either naturally empathetic, or you weren’t. The research and development of Emotional Intelligence by Richard Boyatzis and others gave us richer language and a discipline to develop and measure EI individually and collectively. It helped us recognise that, while people may begin with different aptitudes, capabilities such as emotional awareness, empathy and relationship management can be observed, practised and developed.

I think human judgement is at a similar stage.

This leaves judgement hidden inside individuals and makes it difficult for an organisation to strengthen it deliberately.

We have a whole host of research we can use to measure and develop nuanced human judgement, and we need to be mindful that we're applying it in a context of rapid, unpredictable leaps of change.

This means our human judgement on AI needs both depth and breadth: close scrutiny of the evidence for a particular AI application’s effectiveness and accuracy, alongside strategic foresight that accounts for the social, technological, regulatory, environmental, economic, ethical, and political factors involved.

It means weighing ‘good’ values that may compete in AI ethics, and communicating opportunities and risks clearly. It means moving quickly to adapt to new information, while also slowing down to think deeply together. We are going to get things wrong, often, so the importance of accountability as part of human judgement cannot be understated.

Sound human judgement develops through maturity, expertise and self-awareness, combined with the humility to actively seek and welcome challenge so we can improve.

Lesson 3: The human judgement we need for AI is a collective capability

Even with exceptional individual human judgement skills, there is likely no one ‘expert’ in your organisation that can see across AI’s impact on people, technology, regulatory, security and ethical risk. This brings us to our next critical lesson: great human judgement is collective, not individual.

When deciding how and where to use AI, people in technology, data, operations, marketing, HR, legal and risk functions will each see different parts of the puzzle. That does not mean putting a broadly representative group into a room and hoping that collective wisdom will emerge. Collective judgement needs structure and social mechanisms to work.

Research on collective intelligence suggests that group collaboration is more important in predicting collective intelligence than the skills of individual members. Additionally, the average social perceptiveness positively predicted collective intelligence.[2]

In other words, attending to how a group connects and communicates matters more than individual intelligence in developing collective judgement. That might look like ensuring diverse perspectives are heard, and people can share their views without the most confident voice dominating. It needs someone to ask what assumptions are being made, what evidence would change the conclusion, whose perspective is absent and what might happen if the organisation chooses not to act. It needs competing options to be considered before the group converges on a preferred answer.

What does this mean? Psychological safety, self-awareness, self-efficacy, inclusion, listening capability, and constructive challenge are all foundational skill sets needed for sound human judgement.

The risk to collective judgement is clear: important perspectives and concerns may never be voiced. We also need to interrogate how shared human judgement relates to accountability and decision-making structures in organisations that are still largely siloed by function and geography.

Building human judgement as an organisational capability

Excellent human judgement is founded on the behaviours and culture of an organisational system.

Human judgement depends on both: building the nuanced capabilities and behaviours needed not only for effective critical thought, but also to influence and enable the best critical thinking to surface and be heard in an organisation.

Our research on leading human-AI teams highlights the importance of responding to AI’s emotional impact. Anxiety, excitement, shame and fear shape whether people share their judgements. Good judgement happens in a culture of trust, lived purpose, freedom to experiment, ambition for excellence and genuine human care.

Employees take their cues from leaders, who need to make their judgement visible: that includes sharing publicly their evidence, assumptions, uncertainties and the perspectives informing them. Changing their minds when better evidence emerges must be part of the examples as they share how their human judgement is formed.

Organisations need enough AI knowledge, psychological safety and time to think together. They must slow down where decisions warrant care and accelerate proven, lower-risk applications, supported by structures, rewards and decision-making that enable collaboration and manage risk.

These lessons inform our work with executive leaders. Developing judgement goes beyond a critical thinking model: it means deepening leadership capability, drawing on diverse ways of knowing and building the conditions for richer, braver conversations that reshape decisions on AI and build the behaviours and cultures that create thriving.

To explore how our executive programmes and team coaching can strengthen human judgement in your organisation, get in touch.