Month: November 2026

As AI changes how work is done, it also changes how people develop expertise. Suzanne Coles, Partner at Curzon Consulting, explores how leaders can turn individual productivity gains into stronger judgement, collective learning and organisational capability.

As AI takes on more work traditionally performed by people, organisations face a question beyond productivity: how will they develop the judgement, experience and relationships on which future performance depends?

We tend to assess AI through the activities it can remove: research, analysis, drafting and increasingly complex professional tasks. The immediate benefits can be compelling. But these activities have also provided experiences through which people learn to assess evidence, recognise exceptions and defend conclusions.

An organisation may produce better work today without necessarily becoming better equipped to make decisions tomorrow.

In the first article in this series, I explored why improvements at the point of AI intervention can obscure their wider impact on business performance. Here, I want to examine whether individual productivity gains translate into stronger collective capability and a more adaptive organisation.

My experience of developing AI strategy, executive diagnostics and sector-specific materials has demonstrated how effectively AI can accelerate research, structure information and generate alternative approaches.

I have found it particularly valuable for challenging reasoning, exploring alternatives and identifying gaps in an argument.

But a sophisticated output does not necessarily demonstrate understanding. A well-structured recommendation tells us little about whether its author recognises its limitations or understands the consequences of acting upon it.

Removing mechanical information gathering may release valuable time. Removing opportunities to evaluate conflicting evidence, discuss findings and receive feedback may have different consequences.

The implications extend beyond individuals. AI could help teams examine competing interpretations and combine expertise. Equally, it could encourage colleagues to rely on similar AI-generated conclusions, leaving fewer assumptions challenged.

The question is what organisations do with the time and information AI makes available.

My work in life sciences and my interest in neuroscience have shaped how I think about the ways people and organisations learn, adapt and develop new capabilities.

Neuroplasticity describes the brain’s capacity to change through experience, learning and repeated activity. Neural connections strengthen or weaken, and functional networks can reorganise throughout life. Learning involves more than acquiring information: it also requires developing and reinforcing new patterns of thinking and responding.[1]

I see an interesting parallel in how organisations learn and transform. People and teams develop familiar ways of working, making decisions and interacting. Some patterns remain effective; others persist long after they have ceased to serve the organisation well.

Just as learning requires more than exposure to new information, organisational transformation requires more than introducing new technology. Both depend on experience, feedback and reinforcing new ways of thinking and behaving.

This is an analogy, not a biological equivalence. But it raises an important question: are we simply accelerating established ways of working, or using AI to develop genuinely new capabilities?

I encountered this challenge during a major acquisition and post-merger integration programme within a global industrial technology organisation.

The organisation brought together businesses with different products, technologies, customer requirements and operating practices. Each had developed expertise suited to its particular markets.

As we examined opportunities for integration and operational improvement, activities that appeared comparable at corporate level could be fundamentally different in practice.

Experienced engineers and operational leaders understood why processes had developed, which exceptions mattered and how functions needed to collaborate.

But integration required more than preserving individual expertise.

People across the businesses needed to understand one another’s operating environments, challenge assumptions and develop a shared understanding of where standardisation created value and where differences mattered.

The challenge was not simply bringing businesses under a common structure. It was helping the organisation learn together without losing the specialist knowledge that made its constituent businesses effective.

This was not an AI programme. But the lesson is relevant: making knowledge accessible and processes consistent does not automatically create the collective judgement needed to operate effectively.

Research and analytical synthesis illustrate the challenge.

AI can undertake much of the information gathering, summarisation and drafting traditionally assigned to junior professionals. There is little reason to preserve unnecessary mechanical work simply because previous generations performed it.

However, professionals also learn through evaluating evidence, recognising inconsistencies, defending conclusions and understanding why experienced colleagues interpret information differently.

There is evidence that AI can support learning as well as immediate performance. A study published in The Quarterly Journal of Economics found that customer-support agents retained productivity improvements during interruptions to AI assistance, which researchers interpreted as evidence of learning.[2]

That finding does not establish what will happen across every profession, but it supports exploring the opportunity.

Consider an early-career engineer assessing a potential equipment failure. Rather than asking AI to identify the cause, the engineer could begin with incomplete operating information, assess possible causes and recommend a response. AI could introduce conflicting data or unexpected conditions, requiring the engineer to reconsider the assessment and explain the reasoning.

Colleagues from maintenance, operations and customer-facing functions could contribute different perspectives. An experienced engineer would validate the technical content, challenge assumptions and provide feedback.

The learner could then tackle an unfamiliar case, with appropriate limits on AI assistance, to establish whether the learning could be applied independently. Insights could subsequently inform training, operating procedures and shared organisational knowledge.

This is a proposed approach, not an implemented programme. Its effectiveness would need testing. But it illustrates how AI could develop individual judgement, strengthen collective problem-solving and create knowledge that the organisation retains.

Crucially, experienced colleagues need time to supervise and challenge. If every productivity gain becomes a higher output target, the opportunity to develop expertise may never materialise.

How do we know whether AI is making an organisation more capable rather than simply making individuals more productive?

I would examine three interconnected levels.

First, are individuals becoming more capable? Are they developing independent judgement, recognising exceptions and applying their knowledge appropriately, including when AI is unavailable or produces an unreliable recommendation?

Second, are teams becoming more effective? Are they sharing knowledge, challenging assumptions, resolving problems across functions and making better collective decisions?

Third, is the organisation learning and adapting? Are improvements being embedded in processes, systems, operating models and decision-making practices, or do they remain dependent on particular individuals and isolated teams?

These are related but different outcomes.

An individual may become more productive without improving collective decisions. A team may develop effective practices without those improvements becoming embedded across the business. Sustainable AI transformation requires understanding how learning progresses through all three levels.

I would add one question to every significant AI business case:

What capability are we removing, and how will we replace the value it provides?

This belongs alongside financial and operational assessments, not in a separate bureaucratic exercise. Five questions should guide leadership teams.

1. What does the work contribute beyond the formal process? Identify the knowledge, judgement, informal controls, relationships and learning opportunities embedded within activities being automated. Consider both individual expertise and collaboration.
2. Which capabilities will remain important? Not everything needs preserving. Some knowledge can become accessible through shared systems; other capabilities require human judgement and accountability. Decide what to retain, redesign or replace.
3. How will future expertise be developed and shared? If AI changes early-career work, learning pathways must change deliberately. Simulation, structured feedback, mentoring and collaborative problem-solving all have potential roles.
4. How will we know whether decision-making is improving? Complement productivity measures with decision quality, errors, exceptions, escalation, rework and responses to unfamiliar situations. Assess individuals and teams.
5. Is the organisation becoming more adaptive? Is AI improving knowledge sharing, collaboration and learning from experience? Are those improvements embedded in the operating model, or dependent on particular individuals?

The accountable business leader should own this assessment, drawing on operations, technology, HR, learning and development, finance and risk.

It should also be revisited after implementation. If productivity improves while judgement, collaboration or the ability to manage exceptions deteriorates, leaders need to establish whether they have improved performance or merely transferred costs and risks into the future.

These questions are not about preserving existing structures. They are about ensuring AI transformation develops the capabilities needed for sustainable performance.

AI will remove considerable work, and much of it should disappear. But the organisations that benefit most may not be those automating the greatest number of activities or reporting the largest immediate productivity gains.

They will be those that understand how expertise develops, how knowledge is shared and how learning becomes embedded in the business.

AI offers opportunities to make expertise more accessible, challenge established assumptions and create better ways of learning together. Realising that potential requires leaders to consider how individual performance becomes collective capability, and how that capability changes organisational decisions and ways of working.

Ultimately, the greatest opportunity may not be to automate what organisations already do, but to enable people, teams and organisations to think, learn and operate differently.

What is changing alongside your AI investment?

About the author

Suzanne Coles is a Partner at Curzon Consulting, where she leads the firm’s AI proposition, helping organisations turn AI investment into business value.

A version of this article was first published on Consultancy.uk.

References

[1] Zatorre, R. J., Fields, R. D. and Johansen-Berg, H. (2012). Plasticity in gray and white: neuroimaging changes in brain structure during learning. Nature Neuroscience, 15, 528–536. https://doi.org/10.1038/nn.3045

[2] Brynjolfsson, E., Li, D. and Raymond, L. R. (2025). Generative AI at Work. The Quarterly Journal of Economics, 140(2), 889–942. https://doi.org/10.1093/qje/qjae044