How to not be a Useless Person in this Age of AI
Some honest words over cold beer, from someone who'd like you to still have a job in five years.
Because I’m going to use the word “useless” a lot in this piece, let me define what it means before you start talking about how that guy called ‘Kay’ is such a rude boy…
Useless, in this case, is when a person can no longer do anything an AI can’t already do. It’s the day the whole of you fits inside the output token limit of an AI model, with nothing extra left for anyone to pay you for. You’ve basically become a more expensive and emotional version of a free, open-source AI model sitting on GitHub. And the annoying thing about these models is that they don’t go on annual leave or get pregnant. Anyway, I digress.
Now, the part that should keep you slightly alert is that being useless is now the easy default setting because AI makes us so. Which means you have to actually do something to avoid it.
So here’s what I’d tell you if we were having this conversation over a bottle of cold beer, as a friend who’d like you to still have a job in five years.
1. Stop using AI like a vending machine…
The most common way people use AI is also the worst way. You paste a task in, it answers, and you copy the answer out. Paste, copy, input, output, like a vending machine. You never read why it did what it did, you never argue with it, and you never push back.
It feels productive, but that’s the trap, because it’s somehow making you dumber. Every time you let the machine think so that you don’t have to, you hand over a little more of the exact thing that made you worth hiring in the first place. If you do that for a year, then you’ve personally trained yourself into the first useless person AI replaces.
Now, I’m the loudest person in the room telling people to use AI more. But I’ve learned I have to define what I mean by that, because “use AI more” is the kind of advice that gets misunderstood the second it leaves my mouth. People hear it and simply do more paste, copy, paste, copy, faster. But that’s not what I mean by using AI more. That’s being a vending machine operator who wants to hit a higher daily target. You’re closer to the machine, sure, but you’re getting dumber at speed, which is the opposite of the point.
Using it more means thinking with it more. Understanding how it works more. Maximizing its advantages while deepening your understanding of the subject and the tool. You should treat it like a sharp colleague that you can always ask why and understand how it thinks, so it makes you a better thinker.
In essence, you should know more, do more, and be smarter because you use AI, not otherwise.
2. Know the difference between your task and your responsibility…
Your task is the thing you do all day. Your responsibility is the outcome you actually own. They are not the same thing, and confusing them is how to become a useless person in the age of AI.
A developer’s task is writing code. A developer’s responsibility is shipping software that solves a real business problem and doesn’t fall over the moment 100 users log in at the same time.
A copywriter’s task is writing words. A copywriter’s responsibility is words that make a complete stranger reach for their wallet and pay.
A designer’s task is designing things. A designer’s responsibility is making a confused human stare at the screen and instantly know what to do, without a single arrow pointing at the button.
A customer support agent’s task is replying to inquiries. A support agent’s responsibility is to make a customer walk away feeling heard, rather than one who walks away and goes on to write a long paragraph on Trustpilot about how shitty your service was.
AI is brilliant at tasks, it can probably already do it better than you. And if you spent your whole career believing the task was the job, you’ll watch the tasks walk out the door because the task finally realised you’re useless.
You should put more energy into the responsibility, which is the messy part full of context, taste, and “this is technically correct but it’s wrong for us.” That’s the part that is hard to hand to a machine, because it needs you to understand the business, the humans, and a hundred small things AI can almost never master.
3. BYOB i.e Bring your own brain…
What makes a human better than the AI isn’t speed, and for sure isn’t effort. You are not faster than the AI I use, even if you’re Usain Bolt (that doesn’t make sense, I just felt like adding it 😂).
Back to the point.
What you have is subjectivity, context, judgment, and taste. The weirdly specific knowledge of what actual humans want this particular week.
Copywriting is the cleanest example. If you sit down and produce templated rubbish, the kind of copy that could be advertising car insurance or a dating app or a funeral home, and you genuinely couldn’t tell which, then AI already does your job better than you, and I’m talking of the free version of ChatGPT.
Your edge was never the typing. Your edge is that you’re a human who understands humans. You know the slang that’s funny today and cringe by next month. You know the joke that lands in Lagos and the slightly different one that lands in Cape Town.
If you bring none of that, then you’re not even competing with the machine, you’re applying to be replaced by it. And in fairness, robots will need someone to sew their clothes one day, so you should probably start learning tailoring.
4. Don’t ship nonsense…
If your output is mid, you are in the first row of the useless people AI is waiting impatiently to replace.
Let me be specific, because I live this daily: this Codex AI I use writes code that, at times, doesn’t even need testing because it just works. So if you’re a developer handing me sloppy, untested code you pasted from somewhere you didn’t fully understand, the comparison is not going to flatter you. It’s the same in every field. If your work looks like what a model produces on its first attempt off a lazy prompt, you’ve already lost the argument for keeping you around.
But there is no need to panic. The answer is to be better than the first draft.
AI now gives everybody a decent starting point, which means the value has shifted to whoever can take that starting point somewhere it couldn’t reach on its own.
If you stay mid, then you’re just the first draft, except you also collect a salary and occasionally take sick leave.
5. Don’t be a naysayer…
There’s always some proud naysayers. Let’s call him John.
You already know John (everyone knows a John). John has been doing his job for ten years, and he will find a way to summarise all ten years within five minutes of meeting you. He’s good at what he does, to be fair. Or at least he was. He’s the kind of man who still prints his emails so he can “read them properly”, and who treats every new AI tool as a personal insult.
So at dinner, the moment someone brings up AI, John sets down his fork, folds his arms across his chest, and announces, with the calm authority of a man who has never lost an argument because he always leaves before they finish, that AI will never do what he does. “It’s just hype” and “Glorified autocomplete”.
For the record, John has used AI a few times, asked ChatGPT some lazy questions back in 2023, got some wrong answers, and has been retelling that story at every gathering since as proof that the whole thing is a bubble that’ll soon burst.
I’ll say this gently, John, because I’m on your side: the model I work with every single day is already better than most people at the one thing they do, and it does not care how senior you are or how many years you’ve put in. Your CV is not a force field. The AI hasn’t read it and wouldn’t be impressed.
By the way, you should not have an opinion on how good or bad AI is if you’re judging based on the free version of any AI model. AI is significantly more capable than what you’re getting, and that’d cause a judgment bias because you can only speak based on what you know, and what you know is what a limited free version of ChatGPT shows you. If you stay close enough to AI and use the latest paid models, you will understand why Kaki is not Leather.
Whatever you do, don’t be a naysayer when it comes to AI.
6. Understand that if your work can be automated, it will be automated, eventually
If your job is routine, fixed, follow-the-steps work with no real judgment living anywhere inside it, you’re already standing on the tracks, and that sound in the distance is the train coming to pack all the useless people.
Picture the work that’s pure process. Copying numbers from one sheet into another form. Sending the same three-line template reply to every customer who asks the same question. Reconciling two lists that are supposed to match and flagging the rows that don’t. The kind of work you could hand to a brand new intern with a one-page instruction sheet and zero context. If a human can do it straight off a flowchart, a machine can do it without a flowchart, except the machine does it at 3 am, for free, without complaining, and without asking when it’s getting promoted.
So the rule is, and I’d tattoo it on the inside of your eyelids if I could: if your work can be automated, assume it will be. Maybe not today, maybe not this year, because your boss doesn’t know anything about AI yet, but eventually, his friend, Abdul, will tell him. And “eventually” has a habit of arriving earlier than everybody planned for.
The way out is the same as it’s been all through this piece: put judgment into the work. Own the part that doesn’t fit on the flowchart, the bit that needs taste, context, and a human deciding what matters. The steps will eventually be automated either way. Just make sure you were never only the steps.
7. Get really good at spotting the problem…
There’s a certain kind of person every company refuses to let go of. The person who sees the crack before it becomes a flood, and shows up with a fix nobody asked for. That person is proactive, thinks in systems, and is very hard to replace, because being told what to fix is what everyone waits for, but figuring out what to fix on your own is rare.
Now this is the part where I’m supposed to comfort you and say AI can’t do this. It can. There’s an AI called Viktor that lives in one of my Slack workspaces, watches how the team actually works, and proactively flags problems before anyone on the team notices them, then proposes a fix on its own. So… No, “spotting issues” is not some sacred human-only gift anymore. Sorry.
But… the AI is only ever as good as the context you give it. Viktor can see what flows through Slack. It cannot see the conversation you had over dinner, the reason that one client must never be put on hold, the quiet politics in the marketing team, or the thing everyone in the company knows but nobody has ever typed anywhere. You are that context. The person who both spots the problem and carries the thousand undocumented things that decide whether the “fix” is genius or a disaster, that person is not getting replaced by a tool that only knows what it was told.
1, Don’t fix problems only when you’re told to. 2, don’t wait to be told which problems are worth fixing. And that’s one way to not be a useless person in the age of AI.
To sum it all up…
Useful, in this age, is a small and specific thing: you reliably add something an AI can’t. If you do that, you should be fine. And if you don’t do that, then I’m not afraid to say AI will eventually take your job, and it’ll be because you chose to be useless.
Anyway, that’s my advice. What do I know? I might be wrong. But if we all end up being useless anyway, then I’ll see you at the carpentry workshop, or wherever every useless person then chooses to gather.
