Somewhere in the last two years, many of us were quietly promoted.
There was no letter. No new title. No increase in salary. One day you were writing the report yourself, and the next you were reading one a machine had written, hunting for the sentence that sounded right but was wrong.
You did not apply for this role. But you are doing it, possibly for hours every week.
The new workplace vocabulary has a name for it: botsitting. It means supervising artificial intelligence: feeding it context, checking what it produces, catching its confident mistakes and cleaning up after it. The more I think about it, the more I believe it is one of the most misunderstood changes in modern work. Not because of the time it takes, but because of what it quietly demands of us.
The Job Nobody Assigned You
The numbers are worth sitting with. Glean's Work AI Index 2026, a survey of about 6,000 workers in the United States, the United Kingdom and Australia, found that people save roughly 11 hours a week using AI, but spend an average of 6.4 hours botsitting. The net gain is closer to four and a half hours. Real and useful, but far smaller than the headline promise.
Most conversations stop there, at the arithmetic. Is AI worth it? How do we reduce the hours?
I want to look at what sits underneath the arithmetic. When your work shifts from making something to checking something, it is not only your timetable that changes. It is the kind of person your work requires you to be.
From the Mason to the Site Supervisor
Anyone who has built a house in Nigeria knows there are two very different jobs on a construction site.
There is the mason, who lays the blocks. And there is the person who stands back and says, "That wall is not straight. Take it down before the mortar sets."
In years of designing spaces and visiting sites, I have come to see that the second job is not easier than the first. It is different. The supervisor does not carry blocks, but he carries something heavier: a precise, almost physical sense of what right looks like. A good supervisor can see a wall leaning from across the compound. He can tell, from the way the mortar sits on the trowel, that the mix is too weak. He does not need to measure everything, because his eye has been trained.
Here is the question nobody seems to be asking about botsitting: where did that eye come from?
Almost always, it came from the hand. The best supervisors once laid blocks themselves. They mixed mortar, got it wrong, and saw what happened next. Their judgment was built from inside the work.
Concrete Hides Its Mistakes
There is something else every builder knows. Concrete is generous to the eye and unforgiving to the structure. Once the slab is poured and smoothed, you can no longer see the iron rods inside it. Were they the right size? Properly spaced? Was anything left out? The surface looks perfect either way. The truth only appears later, under load.
AI output behaves like fresh concrete. It arrives smooth. The grammar is clean, the tone is confident, the layout looks professional. And that polish covers whatever is, or is not, underneath: the invented figure, the misread instruction, the summary that drops the one point that mattered.
This is why botsitting is so draining. You are not proofreading. You are inspecting a finished-looking surface for faults that, by the nature of the tool, do not look like faults. Fluency is the plaster. Your job is to know what is behind it.
You Cannot Inspect What You Never Learned to Build
This is the idea I keep returning to. AI has not removed the work. It has moved it from the hand to the eye. And the eye is trained by the hand.
If you spent ten years writing reports before these tools arrived, you can spot a weak one in seconds. Your judgement is load-bearing, and it was built long before the machine came along. But what about the young graduate who begins their career as an inspector? Who has never written the report from a blank page, never wrestled the numbers into a budget, never drafted the difficult letter and felt exactly where it went wrong?
They are being asked to approve a wall they have never built.
This is not a criticism of young people. It is a structural problem, and it applies to all of us in any area where we have stopped practising. Every skill we hand over completely is a skill whose quality we slowly lose the ability to judge. The danger is not that AI does the work. The danger is that one day nobody in the room can tell whether the work is good.
The Supervisor Who Signs Without Looking
Botsitting has a cruder cousin, a term now used alongside it: botshitting. It describes oversight abandoned. The output goes out as it is. The checking is skipped. When something breaks, the tool takes the blame.
Every construction site has a version of this. The supervisor who is tired, behind schedule, and under pressure glances at the reinforcement and signs the form. Most days, nothing happens. That is exactly what makes it dangerous. The habit is formed on the days when nothing happens.
I do not think most people skip the checking because they are lazy. I think there are invisible beliefs holding the habit up.
"If I check everything, I will look slow." "Everyone else is just sending it." "The machine is smarter than me anyway."
That last one deserves attention. It is a quiet surrender of authority, and it often arrives dressed as humility. But humility that stops you from asking questions is not humility. It is abdication.
If You Run a Small Business
For a small business owner in Lagos, Aba, or Abuja, this matters even more, because there is often no second pair of eyes.
Imagine a woman who runs a growing catering business. She starts using AI to write her quotes, reply to customers on WhatsApp, and draft a proposal for a corporate contract. It saves her hours. Then one proposal goes out with a delivery promise she never made and prices that no longer reflect what foodstuffs cost in the market. The client accepts. Now she must either lose money or lose face.
The tool did not fail her. The structure did. She replaced a step in her business without building an inspection point around it.
Technology should serve your purpose, not quietly run your business. That means deciding, deliberately, which outputs can leave with a light glance and which must never leave without someone who understands the work reading every line: anything with money, promises, legal terms, or a customer's trust attached.
If you want help putting simple structures like this in place, that is the kind of work SIBTECH AFRICA exists for.
Building Your Inspector's Eye
This week, try a small audit. Take a sheet of paper and draw two columns.
- In the first column, list every task you now hand to AI, even partly.
- Beside each one, write honestly: could I still do this well without the tool? Yes, partly, or no.
- Circle every task with real consequences: money, health, reputation, legal commitments, someone's trust.
- For each circled task marked "partly" or "no", choose one way to keep your hand in. Do it yourself once a month. Write the first draft before you ask the machine. Check the output against a source, not against the feeling that it sounds right.
- Finally, set your sign-off rule. What must you personally read, line by line, before it leaves your desk?
You are not trying to use AI less. You are trying to remain the kind of person who can tell when it is wrong.
Questions to Sit With
Which skills have I quietly stopped practising since these tools arrived?
When I approve AI output, am I checking it, or simply recognising that it sounds right?
Where in my work does a polished surface hide something I have not inspected?
If the tool disappeared tomorrow, which parts of my work could I still stand behind?
Keep Laying a Few Blocks
The machines will keep getting better at laying blocks. That is not the threat. The threat is a generation of supervisors who have forgotten what a straight wall looks like.
So use the tools. Use them well. But keep laying a few blocks with your own hands.
One day, the straightness of everything you sign may depend on it.




