Advertisement

What Do You Stop Doing When AI Arrives? – Lenni Laukkanen

Illustration of an empty meeting table, reflecting the article’s core argument: real AI adoption shows up in the recurring routines leadership chooses to end. *Photo by Michael Fousert on Unsplash

One question has come up in practically every conversation I have had with executives this autumn. It does not originate with them. It comes from their board, and it arrives in much the same words each time: where does all this AI work show up, when nothing seems to have changed?

It is a fair question, and it lands on people who have done what was asked of them. The licences are bought, and the training has run. There is a coordination group, and usually a network of AI ambassadors. People are using the tools, many of them with real enthusiasm. And from the boardroom, nothing seems to have changed.

My argument is that this is not a measurement problem. Most organisations bring AI in as an addition: a new tool, new training, new expectations, and a new task called “experiment and learn”. In my experience, very few decide, at the same time, what stops. When nothing stops, nothing visibly changes. What an organisation stops doing is the part of an AI programme that a board can see without a dashboard, and deciding it is a leadership decision, not something a team or an individual employee can settle alone.

Two reasons nothing looks different

The boards are not imagining things. In PwC’s 29th Global CEO Survey, 56% said AI had delivered neither higher revenues nor lower costs over the past year; only 12% reported both.

The first reason is time, and executives should say so without apology. My own company is two years old, and even the client we have worked with continuously the longest has been with us for less than a year. In that time the organisation has started training its people, set up its ambassadors and begun to build governance. Only recently have the clients furthest along started to go through how their work is actually done, looking for places where AI could change the process itself. People are still in training, in the athletic sense of the word: practising how to bring the tools into daily work. Expecting business metrics to move at that stage is premature.

The second reason does not cure itself with time. We have assisted dozens of AI adoptions, and in not one of them did anybody raise, at the point of adoption, the question of what the organisation would stop doing. Where the hours for coordination groups and ambassador work come from has typically not been agreed either. Nor had anyone considered where the time that AI frees up should go, until we asked. That is understandable, because it is not the first thing that comes to mind with AI. Yet unless leadership decides what the freed time is for, AI produces no benefit you can count in money. Without that decision, the time drifts to wherever each person sees fit. The usual explanation is a backlog long enough to absorb it, which may well be true, although I have never seen it tested.

Advertisement - [email protected]

The pattern is not peculiar to my clients. According to Finland’s Working Life Barometer 2025, among wage earners whose workplace uses AI, 6% said AI had brought many and about a third some new tasks over the past 12 months, whereas for 82% it had replaced no tasks at all. In Humlum and Vestergaard’s Danish data, about 85% of chatbot users said they put the time they save into other job tasks, and the same researchers found no effect on earnings or recorded hours: their estimates rule out effects larger than 2% two years after ChatGPT’s launch. Time is being saved. It is going nowhere a board would notice.

What stops is what shows

Here is the claim in its plainest form. AI becomes visible through what it ends: a report nobody compiles any more, a meeting that no longer takes place, an approval step that has gone. A board member can see those without being shown a chart.

Take an example close to the boardroom. If anyone in the executive team can ask for the sales figures in plain language, or have a dashboard drawn on the spot, I see no sense in still requiring the sales director to bring a slide deck and a set of tables to the meeting. Yet the deck survives in most organisations, because nobody has decided that it stops.

When I was recently asked what my own company had stopped doing since AI arrived, my first answer was: probably nothing. It took me a while to see why. We are young enough never to have started. Compared with my earlier workplaces, a great deal of manual work in prospecting, sales reporting, invoicing and writing up meetings simply does not exist with us. The most concrete absence, and probably the largest in hours, is that we have no weekly or other recurring meeting whose only purpose is to pass information around and keep everyone on the same page. In fairness, that is not always a good thing. Sometimes the technology lets us down and people lose track of where a client project stands. But it has worked well enough that we have never considered introducing one.

Now set that beside a team that had been feeling too busy. The whole team sat in a weekly meeting of general announcements, and we suggested that one member attend and brief the rest. The team preferred to keep the meeting, because it was also the only time they went through shared matters among themselves. Same kind of meeting, opposite outcome. A young company gets for nothing what an established one has to decide.

Why it lands on the executive team

The tempting conclusion is to ask the people who do the work, and there are cases where that has worked. In October 2017 Hawaii Pacific Health asked all employees to nominate anything in their electronic health record documentation that was poorly designed, unnecessary or useless; nurses and physicians submitted 188 nominations.

My experience in AI workshops has been different. Participants find it hard to say that something should not be done at all, and harder to get it onto the list. Drawing out what not to do takes more of our effort than anything else, and still it does not seem to turn up, although from the outside we regularly see reporting that is done because it always has been, and information meetings that are too large and too long. From the inside they are harder to see, because they have become part of how everyone works. My reading, and it is only a reading, is that people lack permission less than they lack a safe way to propose a stop, and that taking part in a meeting can matter for a person’s standing as much as for the information shared. Psychology offers a compatible finding: Adams and colleagues showed in 2021 across eight experiments that people default to searching for additive changes and therefore systematically overlook changes that subtract. Those were laboratory experiments on individuals; applying them to teams is my inference.

Two more things keep the decision at the top. Stopping has a price, as my own company’s occasional confusion shows, and only someone with the authority to accept that price on others’ behalf can make the call. And the direction will not come from above. In the conversations I have had, boards have not given AI a direction, goals or metrics; they expect management to supply them. I hear this from executives rather than from boards, so it is one side of the story.

The question arrives anyway. In my experience, about a year into a deliberate adoption at the latest, someone asks where the time for all this is supposed to come from and what it is allowed to cost. 

What is left when the routine goes

There is a quieter reason to decide. My own view is that when AI takes the easy routines, what remains for people is heavier and vaguer, and it lacks the sense of completion that routine provides. Clearing an inbox produces nothing, but it leaves you feeling that something got done. The idea is old. Writing about industrial process control in “Ironies of Automation”, Lisanne Bainbridge argued that when automation takes over the easy parts of a job, the hard parts left to the human can become harder. I use it as a parallel.

The current evidence is mixed. In a survey study by Lee and colleagues, work shifted from information gathering to information verification and from task execution to what the authors call task stewardship, yet most examples were rated as requiring less cognitive effort with generative AI, not more. And where a tool removes a task outright, the picture changes: in one before-and-after quality improvement study across six US health systems, the share reporting burnout fell from 52% to 39% after 30 days of using an ambient AI scribe to document patient visits.

The reading that survives is conditional. When the same old work is simply done faster and in greater volume, and the freed space fills with checking, the load grows. When a task is taken away, or when the nature of the work changes and AI assists with it, the load can shrink. Which of the two happens is, once again, a decision.

Start here, this quarter

1. Answer the board with a stop list. Not adoption figures. Name three things the organisation will no longer do by the end of the quarter, and say what the time is for.

2. Begin with what you yourselves demand. The decks, tables and status rounds the executive team requires of others are yours to end. The first deletion is more convincing when it comes from the top.

3. Give every planned use of AI a named stop. For each planned use of AI, write down what it ends, who owns ending it, and where the freed time goes. A plan without those three lines is just an addition.

4. Pay for the learning curve. People adopting the tools are doing two jobs. In our survey material, employees describe training that is on offer but that they have no time to attend, and practice for which no working time exists. Stop something for a fixed period to make room.

5. Say the price out loud, and set a date to look again. State what might be lost when something stops, and review the decision after a quarter. A stop that can be revisited is easier to agree to than one that feels final.

The decision under the question

When a board asks where the AI shows up, the tempting answer is a usage chart. The better answer is a short list of things the organisation no longer does, and a sentence about where that time now goes. People further down can propose what goes on that list, but only the executive team can agree it, and the tools will not write it for you. What you stop is how AI shows up.

About the author

For nearly three decades, Lenni Laukkanen has helped organisations develop and grow through technology. He argues that AI is not a technology question but a leadership one: it rarely fails as a tool — it exposes ways of operating built for a world that no longer exists.

As founder of Astu Labs, trusted advisor to executive teams and keynote speaker, he helps leaders navigate the organisational changes required to capture value from AI. A runner, cyclist and adventurer, he holds that organisations, like athletes, change one deliberate step at a time. More at lennilaukkanen.fi.

*The views and opinions expressed by the author do not necessarily state or reflect the views or positions of Hyperight.com or any entities they represent.   

Add a comment

Leave a Reply