The Great Waste in the Age of AI

An organization can put AI on every desk and still waste the intelligence of the people sitting there. People can be fully occupied all day while much of their knowledge, judgment, and ability goes unused.

In 2014, I spoke at the Northeast Lean Conference about something I called the Great Waste: the underutilization of human potential. One of the parables in that presentation involved employees arriving at work to find red dots on some of their desks. Consultants had been around. People were uneasy. Nobody had explained the dots, so employees began trying to work out whether they identified who was about to lose a job.

Facilities had marked desks that needed to be moved for carpet cleaning. Someone had forgotten to remove the stickers. A routine maintenance decision had become a threat because people had so little confidence in what was happening around them.

Consider that story alongside an AI rollout. We ask employees to explain how their jobs work, share the judgment they have developed, and help identify what can be automated. What they are willing to contribute will depend partly on what they expect us to do with it. If the reward for sharing what they know appears to be the elimination of their job, holding something back is understandable.

The Great Waste begins well before someone stops contributing. It can be there in a busy, capable person who knows exactly where a process breaks and has learned to work around it. The organization gets the workaround every day. The person's idea for fixing the problem never gets a hearing. AI gives that person some interesting new possibilities.

Imagine a customer service team that spends part of every morning explaining delivery delays. One employee notices a pattern: customers are being promised a date before a scheduling decision is complete. The employee knows which cases tend to go wrong because they keep taking the calls. Their performance is measured by how quickly they clear the queue.

An AI assistant could help write those explanations faster. It could also help investigate why the explanations are needed. In this hypothetical, the employee gets an approved set of case notes, promised dates, and scheduling confirmations. They use AI to group recurring complaints, compare the timing, and look for cases that contradict their theory. They check the underlying records because a convincing summary can still be wrong.

With help from a coding assistant, they might build a small demonstration of a different handoff: hold the promised date until the scheduling confirmation arrives, and flag exceptions for review. They could test it against past cases before anyone changes the live process. The idea now has something people can examine, question, and improve.

That is a substantial change in what the employee can bring to the discussion. A person who once needed someone else to analyze the records or build a demonstration has a way to make progress. They still need access, help, and approval. They also need time that belongs to this work. If every available minute is assigned to closing today's tickets, tomorrow's customers will keep calling about the same problem.

Research released this week gives us a reason to pay attention to that distinction. Charter's AI Pulse surveyed 5,002 US desk workers who already use AI. Being involved in how AI fits their job was the condition most strongly associated with reporting benefits. Only 37 percent of individual contributors said they were extremely or meaningfully involved in figuring that out. These are self-reported associations; they do not establish cause and effect. They do raise a useful question about an AI program: how much influence do the people doing the work actually have?

We can give a software agent considerable freedom while giving an employee very little say in how their job changes. I would want to understand that decision. The employee may know why the formal procedure fails on Tuesdays, which customer needs an exception, or why another department cannot use the information being sent to it. That knowledge matters when we decide what an agent should do.

There is a more creative side to this, too. In an October 24 profile, architectural designer Shotaro Tsuda described using AI while working on a pavilion entrance in Tokyo. The generated images did not feel right. He changed the conversation and asked for architectural references to help him rethink the design. In his account, the exchange sharpened his own design judgment. He was using the tool to develop his thinking about the work, with the decision still his to make.

That possibility interests me. A person can explore alternatives, question an assumption, or produce a demonstration that makes an idea easier to discuss. An organization can get more from the people it already employs when it makes room for that kind of contribution.

There is a difficult management decision underneath all of this. If the service employee helps remove a recurring problem, what happens to the capacity that creates? Perhaps the team can improve another part of the service, absorb growth, or spend more time on difficult cases. Perhaps management intends to reduce staffing. People will draw conclusions from those decisions, whatever the AI announcement said.

Leaders should be candid about that. An invitation to share expertise needs a credible account of what the contribution will mean for the people involved. Promising participation while quietly treating every improvement as a headcount calculation gives people a reason to distrust the next invitation. The red dots will have an explanation long before leadership supplies one.

I would start with one recurring problem and someone close enough to understand it. Give that person an approved way to use AI, time to investigate, and access to a manager who can act on what they find. Ask them to show the evidence, the proposed change, and what might go wrong. In the service example, fewer repeat calls would be useful only if the new handoff also worked for scheduling and customers.

Then follow the idea through the organization. Did anyone test it? Did another department identify a problem? Was the proposal improved, rejected for a clear reason, or simply left waiting? What happened to the person who brought it forward? Those answers say a great deal about whether the organization is making better use of human potential.

An employee who can finish a report faster may also have an idea for making the report unnecessary. I want our AI investment to give that idea a fair hearing.

---

Also published in The AI Catalyst on LinkedIn.