Forget the Headlines. Make These Two Lists Instead.

Every week there's a new headline telling you AI is coming for your job. Every other week there's one telling you it's all overblown. Both can't be right, and staring at either one gets you exactly nowhere.

Here's the problem with the whole debate: it argues about "jobs" as if a job were one solid thing. It isn't. Your job is a bundle of maybe thirty different tasks, and AI doesn't care about the label on the bundle. It cares about the tasks inside it. Some of those tasks are already being handled by software. Some of them never will be. And the mix is what decides how your work actually changes.

So put down the think pieces for a minute. There's a faster way to see your own situation clearly, and it takes about fifteen minutes and two sheets of paper.

The exercise

Grab something to write with. You're going to make two lists.

List one: every part of your work you'd happily pay someone to take off your hands forever. The stuff you dread. The tasks that feel like friction between you and the actual point of your job.

List two: the parts you'd do even if they weren't in your job description. The work that feels like the reason you got into this in the first place. The stuff that, on a good day, doesn't feel like work at all.

Don't overthink the categories. Just be honest. Most people fill list one faster than they expect, and then sit staring at list two for a while. That pause is worth paying attention to. We'll come back to it.

Here's what the lists tend to look like for real people:

A shop owner's list one: reconciling invoices, chasing down late payments, retyping the same customer answers, formatting the newsletter, comparing supplier quotes. List two: talking to customers, spotting what's about to sell, training a new hire, deciding what the business should try next.

A sales rep's list one: logging calls into the CRM, writing the same follow-up email for the tenth time, building the weekly pipeline report. List two: reading the room on a call, figuring out what a hesitant buyer is actually worried about, closing.

A parent juggling a side business: list one is the meal-plan-grocery-list-budget-spreadsheet grind. List two is the part where they actually cook, or actually build the thing they're selling.

You see where this is going.

What the two lists are actually telling you

Here's the uncomfortable, and honestly kind of freeing, truth: AI is coming for list one. Not next decade. It's already chewing through a lot of it.

The tasks on list one share a family resemblance. They're repetitive. They follow patterns. They have a "right" way to do them that doesn't change much day to day. They're the tasks you could write down as instructions if you had to. That's precisely the kind of work software is now good at.

List two is different. That work resists instructions. Reading a nervous customer, knowing which risk is worth taking this quarter, deciding the newsletter should sound warmer this month, sensing that a supplier relationship is about to go sideways. You can't fully write those down, because you're not following a rule. You're using judgment built from years of paying attention. AI can assist with pieces of it, but it can't hold the context, the stakes, or the relationships the way you do.

That's not a comforting story about how humans are special and magical. It's a practical distinction. The tasks that are easy to explain are the ones most exposed. The tasks that are hard to explain are the ones that stay yours. We wrote about the flip side of this recently in You Can't Automate What You Can't Explain, and it's the same coin from the other face.

Why this beats reading another prediction

Predictions talk about averages. "40% of tasks in this industry." "Millions of jobs." None of that tells you anything about your Tuesday.

Your two lists do. They turn a giant abstract question ("is AI coming for me?") into a concrete one ("which of these specific things can I stop doing by hand?"). And concrete questions have answers you can act on this week.

Try it. Take the top item on your list one and ask yourself: could I describe this task clearly enough that a sharp new hire could do it from my instructions alone? If yes, that's a strong signal AI can take a big bite out of it right now. Drafting routine emails, summarizing long documents, turning messy notes into a clean first draft, comparing two options against a checklist, pulling the key points out of a contract. This is the boring, useful stuff that quietly eats hours.

If the honest answer is "no, you kind of have to be me to do this," congratulations. That task is on the right list.

The trap to watch for

There's a catch, and it's the one most people fall into.

When list-one work disappears, the instinct is to fill the gap with more list-one work. You automate the invoicing, and instead of pouring that time into the parts of the business only you can do, you find three new administrative fires to fight. The friction comes back in a different outfit.

This is how people end up busier than ever with tools that were supposed to free them up. They automated the task but never reclaimed the time. The whole point was to shift your hours toward list two, and if you skip that step, you've just made your treadmill faster. There's a related failure mode where businesses automate the wrong thing entirely and speed up a broken process. We got into that in Automate the Wrong Thing and AI Just Speeds Up Your Mess.

So the exercise has a second half. Once you've identified what's leaving list one, decide in advance where those hours go. Write it down next to the task. "AI drafts the follow-ups, I spend that hour actually calling the three biggest accounts." Otherwise the vacuum fills itself, and rarely with anything good.

What if your whole job is list one?

Fair question, and it deserves an honest answer instead of a pep talk.

Some roles really are mostly repetitive, explainable work. If that's you, this is worth knowing now rather than later, because it means the ground under that role is genuinely shifting. But notice something: even in those roles, list two usually isn't empty. It's just been crowded out. The person doing data entry all day often has sharp instincts about which records look wrong, which customers are unhappy, where the process breaks. That judgment was always there. It just never got room to matter, because the busywork ate the whole day.

When the busywork shrinks, that buried list-two skill is the thing to grow into. The move isn't to compete with AI at the stuff it's good at. It's to lean hard into the part of the work that needed a human all along, and to get deliberate about building more of it. This is roughly why the people getting the most out of AI often aren't the technical ones. They're the ones who know their domain cold and finally have time to use that knowledge. We made that case in The People Getting the Most From AI Aren't the Tech People.

The five-year version of the question

Do the exercise once and it's useful. Do it as a habit and it becomes a compass.

Every few months, redo your two lists. Watch what migrates. Over time, tasks slide from "only I can do this" toward "actually, I can explain this now, and something else can handle it." That migration is the real story of AI and work, and it's happening at the level of individual tasks, quietly, all the time. You'll feel it before any headline confirms it.

The people who'll be fine in five years aren't the ones who guessed the news right. They're the ones who kept honestly sorting their own work into these two piles and kept steering their time toward the pile that's actually theirs.

You don't need a forecast for that. You need two lists and the nerve to be honest about what's on them.

So before you read the next breathless take on the future of work, make the lists. Yours will tell you more than the headline ever could.