Why Leaders Need Artificial Intelligence
Why does the truth arrive last at the top of an organisation?
Why Leaders Need Artificial Intelligence
The Last to Know
John Stumpf has left Wells Fargo after revelations that employees opened unauthorised accounts for customers. The scandal looks like a failure at the top, but it also reveals something stranger about the top itself. The people with the most authority can be the last to see what their organisation is actually doing.
From the executive floor, more accounts can look like growth. They suggest that customers are using more services, branches are performing well and the strategy is working. The measure appears reasonable. The trouble begins when it stops describing the work and starts becoming the work.
When the Measure Becomes the Work
A target doesn’t remain on a dashboard. It travels down through the organisation and changes the behaviour beneath it. People learn what is counted, what is rewarded and what must be produced to remain in good standing, until the metric begins shaping the reality it was meant to observe.
This is one of the quiet dangers of management by numbers. A leader may believe they are seeing evidence of customer confidence when they are really seeing evidence of employee pressure. The number can be accurate in the narrow sense while the interpretation around it is not.
Formal hierarchy makes that harder to detect. People near the bottom perform much of the operational work, managers coordinate it and senior leaders decide its direction. As information moves upwards, the work itself is gradually replaced by descriptions of the work. Some compression is unavoidable. No chief executive can read every email, inspect every account or sit inside every branch, but the same compression that makes leadership possible also gives distortion somewhere to live.
The Long Climb Up
Information rarely rises untouched. It is selected for relevance, adjusted for context and softened for presentation, often by several people before it reaches the person expected to act on it. Each alteration can seem sensible on its own. A detail disappears because it looks too minor. A warning becomes less direct because the evidence is still developing. A conclusion is generalised because nobody wants to accuse another department without certainty. The report becomes safer each time it moves. By the time it reaches leadership, it may be polished enough to say very little.
The problem isn’t always dishonesty. People are often trying to protect the organisation, their colleagues or themselves. They know that bad news carries political weight, and that certainty can be punished when it later proves incomplete. Tentative language becomes a form of shelter. The leader receives information, but not necessarily intelligence. Information is what has been reported. Intelligence is what helps someone understand what is actually happening.
A leader can try to repair the gap by going directly to the source. They can visit teams, inspect records and ask questions without waiting for the formal chain. That restores detail, though it doesn’t scale. An organisation can produce more signals in a day than one person could examine in a month. Direct access without a way to reduce the noise simply creates another kind of blindness.
A Second Line of Sight
Artificial intelligence offers a different route through the organisation. Pattern-recognition systems can examine large operational streams and bring unusual behaviour to the surface before it has travelled through several layers of interpretation. The point is not to let a machine lead the company. It is to give the leader another line of sight.
A useful system could compare account openings with complaints, closures, employee turnover, branch targets and other operational traces. It could identify where the official story and the underlying pattern begin to separate. The leader would still make the decision, but the field of view would be wider.
Google Search already shows the basic shape of this relationship. A person doesn’t read the entire web before asking a question. A system searches, ranks and reduces a vast body of material into something that can be inspected. An organisation needs a similar ability to question itself.
That doesn’t make the result free from bias. No system is unfiltered. The data can be incomplete, the model can be badly designed and the questions can still be wrong. An automated analysis does not, however, need to preserve a manager’s standing before reporting an anomaly. It can provide a route around the same social chain that shaped the original report.
Until such systems become practical, an independent business architect can perform part of this function. The role should sit close enough to the operation to understand the detail, but far enough from departmental politics to report what the structure reveals. Like an auditor, part of the value lies in independence.
Who Holds the Lens
Leaders need artificial intelligence because the organisation is larger than any leader’s attention. They need systems that can preserve detail while reducing volume, reveal patterns across silos and show where the measure has begun altering the thing being measured.
The risk is that the new lens becomes another layer of authority. If leaders cannot inspect what the system selects, ignores and ranks, then the politics haven’t disappeared. They have only moved into the model.
Power does not belong only to the person who makes the decision. It also belongs to whatever decides what that person is allowed to see.
Comments