While researching my first book on AI in mid-sized companies, I read a stack of guidebooks on the subject. They cover prompts, workflows, tools, the usual promise that the right setup will save you time. Every few chapters the argument arrives at the same place: sooner or later, human work goes away. The claim itself did not surprise me. What I could not get past was how calmly it stood there.
One sentence stayed with me. The era of autonomous AI companies that operate without human involvement is imminent. I read it twice, then again, because I could not believe it was standing there with no comment and no context around it. The author moved straight on to the next practical tip.
Take the sentence literally. A company is a legal person that decides, signs contracts and earns money. Now remove the people. The legal person still does all three. Nobody gets up in the morning to go to work. This is the end of the idea that the economy has anything to do with people. Automation was never the right word for it. The author set it down the way you would mention the weather.
The equivalent in urban planning would note in passing that cities will soon no longer need residents, then turn to street lighting and parking. You would stop reading, wanting to know whether the author had understood what they had just written. Here the same move passes without a pause.
I have spent thirty years self-employed, and in that time a lot of technology promises have crossed my desk. Blockchain was going to remake how we hold property, sign agreements, trust each other. Most of it never arrived. That is the usual pattern: less changes than the announcements say. But not always. Sometimes something does shift quietly, and you notice only once it is done.
So I am not arguing the prediction is wrong. It might be right. What unsettles me is the register. These guidebooks describe the autonomous company like an operating system upgrade. Install it and everything runs better. Better for whom, though. For the owner it means no salaries, no personnel risk, an operation that runs on an algorithm that wants nothing. The people who used to make up the earlier version of the company are gone from the sentence within a clause.
This is the part of the whole AI question that occupies me most. The large changes tend to go unremarked. The authors package something existential so neatly that the consequences for actual people read like side effects you are expected to accept. I keep wondering whether they do it on purpose, or whether the logic of optimisation has become so ordinary that nobody stops to think. Once efficiency is the goal, a person is a cost. Removing the cost reads as a technical question with no ethical weight: you solve it, and you book the result as a gain.
I do not think any of this is meant to be evil. The people writing these books are trying to be useful. But some sentences cannot be written in passing. In the future there will be autonomous AI companies that operate without human involvement. A line like that should make the reader stop and sit with it. If the sentence no longer gives anyone pause, that is the harder fact: it no longer reads as news.