Predictions of office automation tend to be phrased as jobs disappearing. What is observably happening is different and in some ways harder to respond to.
The compression pattern
A task that took a team of five now takes a team of three using tools that handle the routine parts. Nobody’s job was automated. Two positions were not refilled when people left.
That is how it has arrived in customer service, legal document review, financial analysis, and marketing. The role survives; the headcount shrinks; the remaining people do more with assistance.
Why this is harder to see
Because there is no announcement. No factory closes. The org chart looks similar and the hiring plan is quietly smaller. The effect appears in employment statistics years later and is hard to attribute.
Which tasks compress first
Routine, well-specified, verifiable. Drafting, summarising, extracting, formatting, first-pass review. Anything where the output can be checked quickly by someone who knows what good looks like.
Not judgement, negotiation, relationships, or anything where the answer depends on context that is not written down.
What this means for a career
The person who can check output is worth more than the person who can produce it. Expertise shifts from doing to evaluating, and the path to expertise – which traditionally ran through years of doing – has to be rebuilt.
That is the genuine problem and no organisation has solved it well.
The honest forecast
Gradual, uneven, concentrated in specific functions, and larger in aggregate than any single announcement suggests. Not a cliff. A long slope.