AI Will Take Jobs. It Will Also Make Ones We Can't Name Yet.
There is a pattern to how transformative technology lands. It arrives, it disrupts, people panic, and then — slowly, messily — the economy reshapes itself around the new thing. Jobs that existed before disappear. Jobs that nobody had a word for start appearing in hiring posts.
We are in the panic phase right now with AI.
We have been here before
The ATM was supposed to kill the bank teller. Instead, teller numbers grew for two decades after widespread ATM adoption. Cheaper branches meant more branches, which meant more tellers. The job changed — less cash counting, more customer relationship work — but it did not vanish.
The spreadsheet was supposed to make accountants obsolete. It made them more productive, which made the demand for financial analysis go up, which created more accounting jobs than existed before Lotus 1-2-3.
The internet was supposed to kill retail. It killed some of it. It also created logistics coordinator, UX researcher, SEO specialist, growth hacker, community manager — none of which were job titles in 1995.
The pattern is not that technology creates jobs identical to the ones it destroys. It creates different jobs, often better ones, usually after a painful transition period.
What AI is actually doing
AI is compressing the cost of certain cognitive tasks — writing a first draft, summarising a document, writing boilerplate code, answering a tier-one support question. Tasks that required a human hour now require a human minute with the right prompt.
That compression does eliminate some roles. If one person with AI tools can do what three people did before, two of those positions do not get posted again. That is real and it is already happening.
But compression also changes what is economically viable. Things that were too expensive to attempt become affordable. Markets that did not exist get created.
We cannot name the jobs AI will create because they depend on applications of AI that do not exist yet.
The jobs we cannot name yet
In 2005, nobody was posting for a role called "data scientist." The job existed in fragments — statistician, analyst, researcher — but the specific combination of skills the modern data science role requires did not crystallise until the data existed to justify it.
The same is happening now. The new roles will likely cluster around a few themes: people who understand AI well enough to direct it on complex problems, people who can verify and audit what AI produces, people who handle the human side of decisions that AI informs but should not make alone.
There will also be entirely new categories we genuinely cannot anticipate. The economy has a way of finding work for human effort when the frontier of what is possible expands.
The uncomfortable middle
None of this makes the transition painless. The people whose jobs compress first are not automatically the ones who find the new roles. There is friction, retraining, and real hardship in between.
The honest version of the "AI creates jobs" argument is not that everyone will be fine. It is that the economy will eventually find a new shape — it always has — but the path from here to there is not smooth and the people who bear the transition cost deserve more than a historical footnote.
The technology is not the villain in this story. Neither is it the hero. It is just the next wave — bigger than most, faster than most, and like every wave before it, impossible to fully see from the shore.