Is it going to take my job?

That's the first question everyone asks, and it deserves better than a "yes" or "no". The honest answer starts with a shift in framing: an AI doesn't take a job, it takes tasks.

The distinction that changes everything

A job is a bundle of tasks. An accountant enters journal entries, reviews documents, calls a client who hasn't sent their invoice, advises on a tax choice, reassures someone panicking before an audit.

An AI is good at some of these tasks and useless at others. What shifts, then, isn't "the accounting profession" as a block: it's the proportion of time spent on each task. The job deforms before it disappears, and most of the time it deforms without disappearing.

Which tasks are affected

The criterion is neither difficulty nor prestige. It's tasks that meet three conditions:

• They repeat with variations, not identically.
• The result is quickly verifiable by the person assigning it.
• Mistakes aren't costly, or can be corrected before consequences arise.

Drafting a first version of a meeting minutes meets all three. Delivering a diagnosis to a patient meets none. And many highly skilled jobs contain tasks of the first type, while many low-paid jobs contain none: that's one of the surprises of recent years.

What we're seeing so far

Two things, and they don't point in the same direction.

On one hand, job cuts are being announced citing AI as the reason, in banking, insurance, and customer service. These are facts, dated and quantified.

On the other, major usage studies show widespread but shallow adoption: many people use it, few use it deeply, and most usage remains writing and summarising. The gap between what's announced and what's actually automated remains significant.

Both are true at the same time. Also worth remembering: a company cutting jobs has an interest in citing modernisation rather than a bad quarter, the stated reason isn't always the real one.

The only strategy that counts

Don't try to figure out whether your job "will go". Instead, look at: which tasks in MY week meet the three conditions? Those are the ones that will shift first, and they're also the ones you'd do well to hand over yourself, rather than defend.

Do this right now: list five tasks from your last week. For each, tick the three conditions. The result will tell you more than any prediction.