For most of modern organisational history, information has been expensive.
Not necessarily in monetary terms, but in the effort required to find it, understand it and move it around an organisation.
Data had to be collected. Reports had to be produced, analysis took time and expertise was concentrated in particular teams. Information moved upwards through management structures before decisions moved back down again.
We built organisations around those constraints.
AI changes them.
It can search, summarise, analyse and synthesise information in seconds. It can identify patterns across datasets, interrogate documents and produce plausible answers to questions that once required hours or days of human effort.
The obvious conclusion is that organisations will become more productive. But perhaps something more interesting may happen.
When information becomes cheap, information itself stops being the advantage.
Judgement becomes the scarce resource.
The organisation was designed for information scarcity
Many of the structures we take for granted in organisations emerged partly because information was difficult to move.
Managers gathered information from their teams and passed it upwards. Analysts converted data into reports and specialists became gateways to particular kinds of knowledge.
Meetings existed partly to bring information held by different people into the same room. Approval processes developed because those with authority often possessed context that those doing the work did not.
None of this was necessarily bad design - it was a rational response to the constraints of the time.
But those constraints are changing.
An employee with access to the right systems and AI tools can increasingly interrogate information directly. They can summarise hundreds of pages, compare competing arguments, analyse datasets and explore scenarios without waiting for someone else to package the information for them.
That raises an uncomfortable question:
How much of the modern organisation exists because information used to be difficult to find, understand and distribute?
More answers don’t necessarily mean better decisions
There is an understandable assumption that better access to information leads to better decisions.
Sometimes it does but organisations rarely suffer from a complete absence of information. They suffer from competing priorities, unclear ownership, conflicting incentives, risk aversion and an inability to decide what matters.
AI doesn’t remove those problems - it may make some of them more visible.
For example, if ten people can produce ten sophisticated analyses of the same problem in minutes, the organisation hasn’t necessarily become better at deciding. It may simply have acquired ten more things to argue about.
The bottleneck moves and the difficult question is no longer:
Can we get enough information to make this decision?
Increasingly, it becomes:
Which information matters, what are we prepared to believe, and what are we going to do about it?
Those are questions of judgement.
Knowing the answer is different from knowing what to do
Generative AI is remarkably good at producing answers but organisations don’t create value by producing answers.
Organisations don’t create value by producing answers.
Value is created through decisions:
Should we enter this market?
Should we stop this project?
Should we invest now or wait?
Which customer problem matters most?
What risk are we prepared to accept?
What should we deliberately not do?
AI can contribute evidence, identify patterns, challenge assumptions and model possibilities but… eventually somebody has to choose.
Naturally choosing requires context that is difficult to reduce to information alone - it requires an understanding of consequences and experience.
It also requires knowing when the data is incomplete but sufficient and recognising when an apparently rational answer conflicts with something the organisation knows but cannot easily quantify.
Most importantly, it requires accepting responsibility for what happens next.
AI can make recommendations but it cannot remove accountability.
The value of expertise may change
This doesn’t mean expertise becomes less important - it may simply mean we need to reconsider what expertise actually is.
In an information-scarce environment, expertise often meant knowing things other people didn’t know and that still matters.
But when knowledge can increasingly be retrieved on demand, another form of expertise becomes more valuable:
Knowing which questions to ask, which answers to distrust and recognising what is missing.
Organisations still need expertise in connecting information from apparently unrelated areas and an understanding of the organisational context in which a technically correct answer must operate.
Most importantly, organisational leaders need to know when to stop analysing and when to act. This shifts the value from possessing information towards interpreting it.
From knowledge alone towards judgement.
This has implications for leadership
Leadership has often been associated with having answers. The senior person was expected to know more, understand more and therefore decide more.
That model becomes harder to sustain when information is available throughout the organisation.
If employees can access much of the same information as their managers, leadership cannot derive its legitimacy simply from controlling knowledge.
The role starts to change. Leaders become responsible for providing context, clarifying priorities and for defining the boundaries to empower their people to make decisions.
Essentially, it creates an environment where people can exercise judgement without requiring permission for every action, while leaders step in when a decision genuinely requires authority rather than merely information.
That could make organisations less hierarchical but only if leaders are willing to give up some of the control that information scarcity once gave them.
AI may expose unnecessary organisational complexity
This is where the impact of AI could become much more profound than automation.
The question most organisations are currently asking is:
What work can AI do?
A more interesting question might be:
What parts of our organisation were necessary only because information was difficult to access?
Perhaps some reports don’t need to exist; perhaps some meetings are primarily information-distribution mechanisms.
Perhaps some approval layers exist because people lower in the organisation historically lacked sufficient context; perhaps some specialist functions spend significant amounts of time answering questions that people could increasingly answer themselves.
Most importantly - perhaps management layers designed to collect, interpret and transmit information need to evolve.
That does not mean eliminating people - it does mean reconsidering why the structure exists in the first place.
AI may automate tasks but its larger organisational impact could come from removing some of the assumptions those tasks were built around.
The next constraint
Technology rarely eliminates constraints entirely.
It moves them. Computing made calculation cheap, the internet made distribution cheap and cloud computing made infrastructure easier to access.
AI is beginning to make certain forms of knowledge work and information processing dramatically cheaper.
Each shift changes where value accumulates. If information becomes abundant, the ability to produce more of it becomes less interesting.
The ability to make sense of it becomes more important and when analysis becomes easier, deciding what deserves analysis in the first place becomes more valuable still.
The organisations that benefit most from AI may therefore not be those that generate the most content, automate the most tasks or deploy the most models.
They may be the organisations that understand where the constraint has moved.
Because when information becomes cheap, judgement becomes valuable.
