How to Prepare for AGI (For Dummies)

Okay. You’ve heard the word. AGI. Artificial General Intelligence. Smart people keep saying it with a look on their face like they’ve seen something in the woods.

You don’t work in tech. You don’t have opinions about transformer architectures. You just want to know if your job is going away and what, if anything, you’re supposed to do about it.

Good news: this post is for you.

Bad news: some of your colleagues are cooked.

Let’s begin.

Tip #1: Use the thing. Like, actually use it.

This sounds so stupid I almost didn’t write it. But the number of people who are vaguely scared of AI while having used it approximately twice is genuinely staggering.

ChatGPT, Claude, Gemini — pick one. Use it every day for a month. For everything. Write your emails with it. Argue with it. Ask it to explain things to you. Ask it to explain things to you like you’re eight. Ask it to explain things to you like you’re a bored Victorian nobleman.

Here is what will happen: you will develop taste. You’ll start to feel — viscerally, in your body — where it’s brilliant and where it’s confidently, elaborately, completely wrong. You’ll stop treating it like magic and start treating it like a very fast intern with no memory and occasional brain damage.

That intuition? That’s the skill. That’s what you’re actually building.

People who have it will manage AI. People who don’t will be managed by people who do.

Tip #2: Figure out the one thing you do that’s actually hard.

Not the thing on your resume. Not your job title. The specific, unreasonably difficult, learned-from-experience thing that you do that nobody who just started would be able to do.

Because here’s the thing about AI: it’s extraordinarily good at the average version of almost everything. Give it a task that millions of people have done, written about, documented, and trained into it? It’ll do a version of it in four seconds.

But the weird specific stuff? That still needs a person. At least for now.

The nurse who can tell from across the room that a patient is about to deteriorate before any monitor goes off — before she can even explain how she knows. The account manager who understands that this particular client says “that works for us” when they mean “we’re three weeks from pulling the contract.” The contractor who looks at a job quote and immediately knows the subcontractor is either sandbagging or hasn’t accounted for the soil conditions on that side of town.

None of that is in a textbook. None of it can be Googled. It’s pattern recognition so compressed by experience that it doesn’t even feel like thinking anymore — it just feels like knowing.

AI is not close to that. AI has read every textbook but has never stood in a room.

So: what’s your version of that? The thing you know in your body, not your head? Protect it. Deepen it. Stop spending time on the generic parts of your job that you could now hand off to a chatbot and spend it becoming more unreasonably good at the irreplaceable part.

If you can’t think of one, that’s also useful information. Move quickly.

Tip #3: Be extremely suspicious of jobs that are basically “moving information from one place to another.”

Junior analyst. Copywriter. First-year associate. Junior anything. Research coordinator. Content manager.

These aren’t bad people. These are jobs that are, structurally, the process of taking information from one format and putting it in another format. Summarizing. Drafting. Compiling. Reformatting.

AI does not get tired. It does not need health insurance. It does not have feelings about the brand brief.

This is not doomsaying. It’s just that if your entire job description could be summarized as “moves and reshapes information,” it’s worth having a plan B that doesn’t involve being the person who does that.

The people who will do well are the ones who decide what information matters and why, not the ones who handle the logistics of it.

Tip #4: Learn to give good instructions.

The ability to tell an AI what you actually want, precisely and efficiently, is already a legitimate skill. It has an embarrassing name — “prompt engineering” — but underneath the embarrassing name is a real thing.

Clear thinking → clear instructions → good output.

Fuzzy thinking → fuzzy instructions → plausible-sounding garbage that you then have to sort through.

The people who get the most out of AI are the people who already knew what they wanted. Which means the underlying skill is knowing what you want — being able to decompose a problem, articulate the constraints, describe the thing that would count as success.

You know where you learn that? By actually thinking hard about things. By writing. By being forced to explain yourself to someone who doesn’t share your context.

So ironically, the best preparation for the AI era is getting better at being a clear-headed, articulate human.

Tip #5: Pick up something physical.

Plumber. Electrician. Massage therapist. Woodworker. Chef.

Not because the robot apocalypse is coming and you need to know how to fix pipes in the bunker. But because AGI disruption will be extremely uneven, and the parts of the economy that require a licensed human to show up with hands will be disrupted last and least.

Also it’s just nice to be good at something that exists in the physical world. There’s a dignity to it. Touch something real. Make something that doesn’t live in a tab.

Tip #6: Get your financial life boring.

Not exciting. Boring.

The next decade will have more economic turbulence, more industry disruption, more “wait what happened to that entire job category” moments than the last three decades combined. The people who weather turbulence are not the ones who predicted the exact turbulence correctly. They’re the ones with low expenses, some savings, and no catastrophic single point of failure in their income.

An emergency fund is not a vibe. An emergency fund is a time buffer between you and a panicked decision. In a period of rapid change, time buffers are the most underrated asset class on earth.

Tip #7: Don’t panic. But do move.

The window for “I’ll figure this out later” is closing. Not today, probably not this year. But the gap between people who are fluent with these tools and people who are not is compounding weekly.

The good news is you don’t need to understand how it works. You don’t need to learn Python. You don’t need to attend a single conference with the word “summit” in the title.

You just need to start.

Pick up the tool. Use it badly for a while. Get a feel for it. Figure out what you’re actually irreplaceable at. Keep some money saved. Learn to do one thing with your hands.

That’s it. That’s the whole guide.

You’re already less of a dummy than you were 900 words ago.