Will AI replace your job? Look at what QuickBooks did to accountants
There is a fear underneath most conversations about AI at work: that the tool which does part of your job is the tool that ends it. It is worth looking at the last time a whole profession faced exactly that fear, because the story did not go the way everyone expected.
When accounting software arrived, everyone thought accountants were finished
When affordable accounting software like QuickBooks reached small businesses, the prediction was obvious and widely shared. The software could keep a ledger, reconcile accounts, and run the reports that bookkeepers and accountants had always done by hand. If a fifty dollar program did the work, the reasoning went, who would pay a person to do it?
The fear was reasonable. The math looked like subtraction. Take the manual work, hand it to the machine, and the people who did that work are no longer needed.
What actually happened
The profession did not shrink. It grew, and the work moved up.
Two things happened at once. The cost of doing a single client’s books fell, because the software did the mechanical part in minutes instead of hours. And the capacity of one accountant went up, because the hours that used to go into data entry were freed for something else. A bookkeeper who could carry a handful of clients by hand could now carry many more.
Lower cost per job plus higher capacity did not destroy the work. It expanded the market. Small businesses that could never have afforded a bookkeeper could suddenly afford the help. And the accountant’s own role shifted toward the part the software could not do: telling the owner what the numbers meant, where the business was leaking money, what to do next. The data entry became cheap. The judgment became the product.
Why the same pattern applies to AI
AI lands on knowledge work the way accounting software landed on bookkeeping. It does the mechanical middle of the job fast and cheaply: the first draft, the summary, the lookup, the boilerplate, the routine analysis. That is the part people assume defines the job. It rarely does.
What it actually does is the same two-part move. It drops the cost of producing a unit of work, and it raises the amount of work one person can carry. The person who learns to direct the tool does not get replaced by it. They get the capacity of a small team, and their value shifts to the parts the tool cannot do alone: deciding what is worth doing, judging whether the output is any good, owning the result.
The people who lose ground are not the ones whose tasks got automated. Everyone’s tasks get automated eventually. The people who lose ground are the ones who defined their value as the task itself, and did not move up when the task got cheap.
What this means for how you work
The useful question is not whether AI can do part of your job. It can, and the part it can do will only grow. The useful question is what you do with the capacity it gives back.
The accountants who did well did not compete with the software on data entry. They used it to serve more clients and to sell the judgment the software could not offer. The same opening is in front of most knowledge workers right now. The cost of the mechanical part of your work is falling. The value of knowing which work matters, and whether it was done well, is rising.
The tool is not the threat. Standing still while the tool resets the economics of your work is the threat.