THE AI WISDOM GAP

Human judgement is your competitive advantage. Don't let AI destroy it.

By Tim Kirk and Olive Arnold

For most organisations, AI alone is unlikely to create sustainable competitive advantage. If everyone is using broadly similar AI tools and drawing on largely similar information sources, long-term competitive advantage will be hard to find.

What becomes scarce is the ability to interpret information within a wider human context, where it is surrounded by the ambiguity of specific people, groups, events or moments.

The leaders who become most valuable to their organisation will be those who ask better questions, exercise sound judgement, balance competing priorities and make decisions that reflect not only organisational goals but ethical, cultural and human considerations.

Most organisations acknowledge the critical role of human skills, aiming for a form of human-AI collaboration: "human in the loop".

Unfortunately, evidence shows that as organisations roll out AI, its impact can be to diminish the very human skills that organisations acknowledge are required to succeed.

THE WISDOM GAP

There are some uncomfortable questions within the AI shift that few organisations are asking: if judgement is what the best leadership now depends on, what is judgement and where does it come from? And are firms deskilling themselves by removing the conditions that build it?

Judgement is the ability to form a sound conclusion when information is incomplete, ambiguous or competing.

Our research identifies nine attributes that leaders combine to form good judgement:

• Relevant knowledge and experience.
• The ability to synthesise and analyse information from a range of sources.
• Awareness of the bigger picture.
• Steadfastness in the face of peer pressure or groupthink.
• Sensitivity to their own and others' feelings and biases.
• Confidence and flexibility to change direction or let go of previous decisions.
• Knowledge of what is acceptable and ethical.
• The ability to consider wider implications and "what if?" from different courses of action.
• The ability to make choices to determine what is appropriate and feasible in the given situation.

Unfortunately, the rollout of AI risks reducing opportunities to apply human judgement where it adds value, as well as reducing the ability of leaders to make good judgements when called upon.

We call this the AI wisdom gap: AI reliance reduces people's judgement and decision-making skills at the same time as increasing their confidence in potentially weaker or incorrect AI-driven decisions.

THE SCIENCE: USE IT OR LOSE IT

A key part of judgement has always been built through practice, repetition, feedback and learning at low stakes. This gave junior employees somewhere safe to hone their skills before they moved on to decisions that mattered, and provided a track record of experience to draw upon.

The hippocampus plays a direct role in decision-making by drawing on this library of experience and working with the prefrontal cortex to test out possible options and make choices. Every previous, low-stakes decision is a deposit into that library.

But AI is now very good at exactly the tasks junior employees used to build their judgement, taking away opportunities to learn at low stakes. We believe this one factor lies behind BCG's latest leadership survey, which found 53% of leaders say their junior talent is developing more slowly than before.

MIT Media Lab's 2025 study showed that people who used AI to write showed weaker neural connectivity and recalled less of their own reasoning than those who wrote unaided, what researchers call "cognitive debt".

Learning science helps explain why it matters. Desirable difficulty, a principle from UCLA psychologist Robert Bjork, shows that the discomfort of attempting something before being shown the answer is what makes it stick. Skill is built through supervised struggle, not through reviewing an answer generated by AI.

The risk of a wisdom gap is gravest for today's junior staff if the tasks that once built judgement are automated out of early careers. But AI overuse creates the same risk at all levels: junior staff, professional staff and senior executives. The International Journal of Medical Informatics reported research in the medical profession indicating that AI can lead to an erosion of practitioners' diagnostic skills, and let misplaced confidence in incorrect AI answers override the physician's own medical judgement.

Without care and conscious attention, organisations may end up with a management and professional layer that has no track record of having made a difficult call and learned from it, or has lost the confidence to use its own judgement. By the time that gap is visible, it is too late to fix quickly.

BUILDING JUDGEMENT BY DESIGN

The most important leadership question is not "how do we get people to use AI?" It's "how do we combine the unique skills of humans and AI for maximum positive impact?"

That's not a training question. It's a design question. It means clarifying where and how AI should automate, where human judgement remains essential, and deliberately creating opportunities to build and maintain the cognitive processes that form good judgement.

Applying this through our Organisational Change Model, leaders need to take coordinated action in Strategy, Organisation and Deployment.

Strategy: harness and protect human brain power as a strategic asset

• Develop an AI strategy centred on the "best use" of AI, not the "widest use" of AI.
• Protect against de-skilling by codifying what should be left to humans, and how and when AI should be used.
• Build the development of human skills, judgement and decision-making into HR processes: talent development, performance assessment, management objectives.

Organisation: build an environment based on "human-led, AI-assisted"

• Maintain human accountability for all outputs, reports, recommendations and decisions. No one can hide behind an AI tool, even if the tool was wrong.
• Build a "question asking" culture, not an "answer giving" culture. For example, leaders asking "what do you think?", "what options did you consider?" and "what trade-offs matter here?" before offering their own view. It forces the cognitive work that builds judgement, rather than shortcutting round it.
• Hard-wire ethics and the organisation's values by regularly bringing these into discussions and the appraisal of options.
• Help people be aware of the bigger picture that's going on around them, such as the team's, department's and organisation's goals, and wider industry and customer trends.
• Institutionalise feedback loops by building short after-action reviews into team rhythms so people, and the organisation, can see how decisions played out and learn and adjust.
• Provide opportunities for human mentoring to give junior staff access to experienced colleagues to form and test their thinking with.

Deployment: design in and protect spaces for human learning and input

• Create first-pass thinking moments before AI enters the process, and decide how to use AI as a prompt or comparator for human judgement, not a replacement: a way to augment, challenge or pressure-test thinking.
• Train people in "intelligent interrogation" of AI: how different phrasing, structuring and ordering of questions can glean more insight, how to spot signs of error or hallucination, and how to create challenge and counter-challenge between human and AI thinking.
• Build reflective practice into the workflow. What did I notice? What did I miss? What surprised me? What would I do differently? Reflection is the mechanism that turns experience into capability. Without it, repetition alone doesn't reliably build judgement.
• Help people develop emotional regulation skills to manage themselves and others in ways that AI cannot: stress management, empathy, and awareness of biases and strengths.
• Help people develop their analytical skills to avoid reliance on, or default acceptance of, AI: problem-solving, decision-making techniques, scenario building, risk management.

None of this happens by accident. AI will keep evolving, but the organisations that thrive won't be the ones with the newest tools. They'll be the ones that recognise the human brain is a critical asset, treat judgement as a skill to be practised, measured and protected, and maintain a pipeline of talent ready to add human value in an AI world.

About the authors: Olive Arnold is an Organisational Psychology and Wellbeing Specialist at Step Inside Group (olivia.arnold@stepinsidegroup.com). Tim Kirk is Managing Partner at Step Inside Group (tim.kirk@stepinsidegroup.com).

More information: If you'd like to explore how Step Inside Group can help your organisation build the human capabilities needed to lead effectively in an AI-driven world, we'd love to hear from you. Contact Olive or Tim, or email us at info@stepinsidegroup.com.