As AI makes individuals more productive, leaders face a new challenge: ensuring technology strengthens, rather than weakens, the collaboration and collective judgment that drive organisational performance.
Twenty years ago, when mining companies introduced more sophisticated machinery, the expectation was that the more powerful equipment would make workers more productive.
But that is not how it played out in practice.
Studies revealed that the new equipment disrupted established relationships and patterns of collaboration between workers. The technology changed not only how each worker performed their tasks, but also how they communicated, coordinated and worked together.
Mining companies discovered that improving the efficiency of individual tasks doesn’t always improve the effectiveness of the system as a whole.
That principle holds true today as organisations embrace artificial intelligence to improve productivity.
Individual productivity is not collective performance
Much of the current conversation about AI focuses on individual productivity. How much time can employees save? How many tasks can be automated? How much more can one person accomplish?
These questions remain important, but organisations don’t create value simply by aggregating individual outputs. They create value through coordination, the integration of different perspectives, sound judgment and collective action.
AI is dramatically expanding what individuals can accomplish. But what happens when the same technology begins to change how people interact with one another?
Imagine a team member who needs advice on a complex problem. Previously, they might have approached a colleague, discussed the challenge, exchanged perspectives and developed a solution together.
Today, they can ask AI and receive an answer in seconds.
The immediate productivity gain is obvious. But what may get lost is the opportunity to exchange knowledge, strengthen trust and develop a shared understanding of the problem.
Multiply that interaction across hundreds of employees and thousands of decisions, and a different organisational picture begins to emerge.
We may be making individuals more productive while unintentionally weakening the connections that make organisations effective.
The hidden cost of AI productivity
Recent research published in Harvard Business Review highlights this emerging risk.
In a study of 1,033 US knowledge workers across more than 20 industries, researchers found that greater AI use was associated with employees perceiving more specialised expertise within their teams.
That sounds encouraging. But the research revealed a troubling counterpoint.
When employees increasingly turned to AI instead of their colleagues, team coordination and trust suffered. Teams characterised by this behaviour reported 22.3% lower coordination and 21.9% lower trust.
The researchers describe this phenomenon as relationslipping: the gradual weakening of working relationships as people substitute AI interactions for human ones.
The distinction is important. The problem is not necessarily how much AI people use, but which human interactions it replaces.
High-performing teams rely on an often-invisible network of relationships and shared knowledge. People know who has expertise, whose judgment they can trust and how to draw on one another’s strengths.
These connections develop through everyday interactions: asking questions, seeking advice, challenging ideas and solving problems together.
When AI replaces too many of these exchanges, the relationships that underpin effective teamwork can gradually weaken.
And the consequences extend beyond team relationships.
When trust and coordination deteriorate, decisions can take longer, expertise becomes fragmented, work is duplicated and execution suffers. The productivity gains achieved by individuals risk being eroded by inefficiencies created between them.
The emerging risk is that organisations become more productive within functions but less effective across them.
The paradox of greater independence
One of AI’s most powerful benefits is that it reduces our dependence on others to complete tasks.
An executive can develop a financial analysis without involving finance. A marketing leader can generate sophisticated customer insights without engaging an analyst. A manager can prepare a strategy paper without seeking input from colleagues.
This greater independence can be enormously valuable.
But it also raises an important question: what happens to collaboration when people no longer need one another in the same way?
Consider an executive team preparing to make a major strategic decision.
Each functional leader uses AI to develop sophisticated analysis, test scenarios and prepare recommendations. Individually, they arrive better informed and more thoroughly prepared than ever before.
Yet if the team lacks shared understanding, trust and the ability to challenge one another constructively, more sophisticated individual analysis may simply produce more entrenched positions.
The quality of the decision still depends on what happens when those executives come together.
This is the paradox leaders need to recognise.
AI may reduce our dependence on one another to complete tasks, while increasing our need to work together to create value.
The more capable individuals become, the more important it is that organisations can integrate their expertise, reconcile competing perspectives and translate decisions into coordinated action.
Without that capability, AI risks reinforcing the functional silos and fragmented decision-making that many organisations are already trying to overcome.
Why leadership matters more, not less
The HBR research offers another important insight.
Employees reporting low psychological safety were more than twice as likely to turn to AI instead of a colleague.
This makes intuitive sense. AI is available, patient and generally non-judgmental. For someone reluctant to expose uncertainty or seek help, it offers an appealing alternative.
But the conversations people avoid are often the very interactions through which trust, mutual understanding and psychological safety develop.
AI may not create the underlying dysfunction in a team, but it can make that dysfunction easier to avoid rather than resolve.
This presents leaders with a challenge that goes well beyond encouraging AI adoption.
Most organisations are investing in technology, developing individual AI capability and measuring productivity improvements. Far fewer are systematically examining how AI is changing the way their teams collaborate, share knowledge and make decisions.
Indeed, leaders may be unintentionally rewarding the very behaviours that weaken collective performance.
If success is measured primarily through individual output, AI adoption and time saved, employees have little incentive to consider how their use of AI affects colleagues or the broader organisation.
What appears rational and productive for an individual may ultimately be counterproductive for the team.
AI transformation is therefore not simply a technology or individual capability challenge. It is a leadership and organisational effectiveness challenge.
Three questions every executive team should be asking
The opportunity is not to slow AI adoption or preserve unnecessary human interactions. It is to deliberately redesign how people and technology work together to create value.
Executive teams need to consider three crucial questions:
Hours saved and tasks automated are useful measures, but they tell us little about whether an organisation is becoming more effective.
Leaders need to consider whether AI is improving the outcomes that matter – better decisions, faster execution, stronger cross-functional collaboration and improved customer and commercial results.
This may require changes to performance measures, incentives and the way teams are held accountable for shared outcomes.
Not every interaction needs to be preserved. AI should eliminate unnecessary work and reduce friction wherever possible.
But some interactions are fundamental to organisational performance – challenging assumptions, resolving competing priorities, sharing expertise, developing judgment and building commitment to decisions.
Leaders need to identify these moments and design ways of working that strengthen them.
AI can help people prepare more thoroughly for a discussion, explore alternatives and test their thinking. The objective should be to improve the quality of the subsequent human interaction, not necessarily replace it.
As AI offers an increasingly convenient alternative to seeking advice from colleagues, trust and psychological safety become even more important.
Leaders need to create environments where people are comfortable asking questions, acknowledging uncertainty, challenging assumptions and seeking help.
They also need to make collaboration intentional: establishing clear decision rights, encouraging knowledge-sharing and ensuring teams have opportunities to develop shared understanding.
These are not new leadership responsibilities. But AI makes them more consequential.
Making organisations smarter, not just faster
The mining research offers a useful reminder.
Introducing more powerful tools doesn’t automatically improve the performance of the system. Technology can change the relationships and patterns of coordination on which effective work depends.
AI creates a similar challenge, but at a much greater scale.
We can use it to make individuals extraordinarily capable while allowing the connections between them to weaken.
Or we can use it to strengthen how people share knowledge, combine expertise, challenge one another’s thinking and solve problems together.
The next phase of AI transformation will not be won simply by organisations with the most sophisticated technology or the highest rates of adoption.
It will be won by those that learn to combine the extraordinary capabilities of AI with the distinctly human capabilities of trust, collaboration and collective judgment.
For leaders, that requires a fundamental shift in focus: from making individuals more productive to making teams more effective.
Because an organisation in which everyone works faster, but no one works better together, is not necessarily a more productive organisation.
