AI & Your Job

The honest answer: AI automates tasks, not jobs

Start with the distinction that cuts through 90% of the noise. A job is not one thing — it's a bundle of twenty or thirty different tasks. An accounts officer doesn't just "do accounts": she chases missing invoices, drafts reminder emails, reconciles spreadsheets, explains a discrepancy to an anxious vendor on the phone, decides which of three ambiguous expense codes applies, and takes the blame if the month-end numbers are wrong.

AI is very good at some of those tasks — the drafting, the summarising, the first-pass number checking. It is nowhere near the others — the phone call that calms a vendor down, the judgement call on the ambiguous code, the accountability when something goes wrong. So what actually happens in most workplaces isn't "the job disappears." It's "the job changes shape": the routine slices get faster, and the human slices become a bigger share of what you're paid for.

That's not a comforting slogan; it's what the early evidence shows. Studies of workplaces that adopted these tools keep finding the same pattern — big time savings on writing, summarising, and routine analysis, with the person still firmly in the loop for decisions and relationships.

History rhymes — but it's playing faster this time

We've been here before, more than once. When spreadsheets arrived in the 1980s, they automated exactly what bookkeepers spent their days doing: adding up columns. Many people predicted the end of accounting. Instead, the number of accountants and analysts grew — because when calculations became cheap, everyone wanted more analysis, and the job shifted from "adding numbers" to "explaining what the numbers mean." When banks installed ATMs, teller jobs were supposed to vanish; instead branches got cheaper to run, banks opened more of them, and tellers shifted toward advice and sales.

So the pattern "tools change the shape of the role more than they delete it" has good historical form. But honesty requires two caveats:

• The pace is different. Spreadsheets took a decade to spread through offices. These tools reached hundreds of millions of people in months, and they improve every few months. Roles will change shape faster than the retraining systems around them are used to.
• Some roles really do shrink. Work that is mostly routine text or data handling — basic transcription, template-driven writing, simple translation, first-line "read the script" support — is genuinely exposed, because for those jobs the amber blocks in the diagram are most of the stack. Pretending otherwise would be cheerleading, and this chapter promised you honesty.

If your work sits close to that exposed zone, the message isn't "panic" — it's "start moving now, while you have time and options." The rest of this page is that plan.

The real competition isn't AI — it's people who use it

There's a line you'll hear everywhere right now, and it's quoted so often because it keeps turning out to be true: "AI won't take your job, but a person using AI might."

Inside almost every profession, a gap is opening. Two colleagues, same role, same salary. One drafts a client proposal in three hours; the other gets a solid first draft from AI in four minutes and spends her saved hours tailoring it, calling the client, and starting the next one. Over a year, the second person simply produces more, learns faster, and becomes the one everybody routes work through. Nothing dramatic happened — no robot walked in — but the competitive landscape shifted quietly, one afternoon at a time.

The good news hiding in this: the bar is astonishingly low right now. In most offices, "the person who knows AI" is currently anyone who has spent twenty focused hours with these tools. You're holding a book that gets you there. This is the rare moment where a modest head start compounds.

What to do about it — concretely, this month

Not "reskill for the future economy." Three specific moves:

1. Audit your own role. Write down the 10–15 tasks that fill your actual week — not your job description, your real week. Then mark each one: could AI draft it, speed it up, or check it? You'll typically find a third of your week is amber. That list is your personal automation map, and making it puts you ahead of most managers, who've never done this for their own teams.

Here's a list of the tasks that fill my typical work week (rough hours in brackets):
[paste your list — e.g. "answering customer emails (6h), preparing weekly sales summary (3h), supplier phone calls (4h), updating the stock spreadsheet (2h)…"]

For each task, tell me:
1. Can AI meaningfully help — yes, partly, or no?
2. For the "yes" and "partly" ones: exactly HOW, with one example prompt I could try this week.
3. Rank the top 3 tasks where I'd save the most hours, starting today.
Be realistic — tell me where AI would NOT help too.

2. Become your team's AI person. Once you've saved real hours on your own tasks, mention it. Show a colleague the trick with the meeting summary. Offer to draft the team's proposal template. In every workplace, someone is about to become "the one who gets this stuff" — the person consulted when the boss asks "should we be using AI for this?" That reputation costs nothing to claim right now and will be crowded in two years.

3. Reinvest the saved time in the green blocks. This is the step people skip. If AI saves you five hours a week and you spend them on more routine output, you've gained little. Spend them on what clients and employers actually pay premium for: judgement (knowing which answer is right for this situation), relationships (the client who trusts you, the colleague who owes you), taste (knowing good from merely fine), and accountability (being the person who says "this is correct — I checked"). AI has none of these. They were always the valuable part of your job; now they're becoming most of it.

And keep learning cheap: the same chatbot that drafts your emails is a patient, free tutor for whatever skill your role is shifting toward — see Learning with AI for how to set that up properly.

Whoever you are, there's a version of this move

Skills that appreciate, tasks that don't

Here's the honest sorting — which parts of everyday work are exposed, and which get more valuable as AI gets better:

Notice the pattern in the right column: four skills keep appearing. Clear communication — which, delightfully, is exactly what prompting is; every hour you spend learning to instruct AI precisely is an hour spent on a timeless skill. Domain judgement — knowing your field well enough to catch AI's confident mistakes (that's why the hallucinations chapter matters for your career, not just your homework). Checking and editing AI's work — a genuinely new skill, and scarce. And adaptability — because this page will need rewriting in three years, and so will everyone's job description.

Key takeaways

• AI automates tasks, not jobs. Your job is a stack of tasks — some are getting faster, and the human ones are becoming a bigger share of your value.
• History (spreadsheets, ATMs) says roles change shape more than they vanish — but this wave moves faster, and mostly-routine text and data roles genuinely shrink.
• The real competition is people who use AI well. Right now the bar is low; a modest head start compounds.
• This month: audit your weekly tasks with AI, become your team's AI person, and reinvest saved hours in judgement, relationships, taste, and accountability.
• The skills that appreciate: clear communication (that's prompting), domain judgement, checking AI's work, and adaptability.