TL;DR: The doomers say AI will destroy civilization, your job or your career. The hypers say it’ll replace every human task, save your business on its own, or make your strategy work without the inconvenient work of adopting it. Both framings serve the people producing them more than the people listening, and they’re doing damage right now, so neither shows up in what I tell clients about AI.
Both camps wear me out. My own position is dull and I’m keeping it: AI is a tool. Treat it like one and you’ll do fine. Let it do your work for you and you’ll have problems. Neither camp can sell that, so neither camp says it, and the people listening to them end up paying for the silence.
The doomers think AI is going to destroy everything and the only sane response is to refuse it. The hypers think AI is going to replace every human task and the only sane response is to lean in completely.
Both camps get a lot of airtime. They sell books, run podcasts, and produce content that people share when they feel they have now formed an opinion on AI.
Each camp is wrong in a different way, and the people who pay for it are the ones who take them seriously: the freelancer who swears off the tools and loses clients to someone faster, the owner who lays off staff because a podcast told him the machines were ready. That second one is shameful, because the advice came from people who’ll never see the bill.
Why are the AI doomers wrong?
The doomer position is that AI is a deep threat to humanity, work, art, truth, democracy, or all of the above. The right response is to oppose its development, refuse to use it, and warn everyone you can reach about the catastrophe coming.
Your refusing to use AI doesn’t stop your competitor from using it.Share on X
There are sophisticated versions of this position and there are shallow versions. The shallow version is the one that’s getting the most attention, and the shallow version is wrong for five reasons.
First, the technology already shipped. Whether you think AI is dangerous is irrelevant to whether AI exists. It exists. It’s deployed. It’s in production at every major company. The conversation about whether to allow it is over. The only conversation left is how to use it well. Doomers who haven’t accepted this are still arguing the last war. (See also Mo Gawdat’s 2027 AI dystopia, and which of his warnings are already here.)
Second, the catastrophe narratives keep failing. Doomers have been predicting imminent civilizational collapse from AI for several years, and the dates keep moving. Each new prediction is presented with the same urgency as the last one even though the last one didn’t come true. This is proof that the people making the predictions don’t have a reliable basis for the timing of the predictions. That means the predictions aren’t practical.
Third, refusing the tool doesn’t make the tool go away. Your refusing to use AI doesn’t stop your competitor from using it. Your refusal doesn’t stop the next generation of workers from being trained on it. Your refusal doesn’t slow the development of the technology by any measurable amount. It only changes whether you, personally, get the benefit of the productivity gain or watch other people get it.
Fourth, the doomer position is convenient for people who profit from inaction. Every week you spend reading articles about how AI is going to destroy your industry is a week lost. You didn’t spend it learning the tools that would have made you more valuable. The doomer content is satisfying. It feels like analysis. It’s not action. The producers of doomer content make money whether you act or not. You only benefit if you do, and the doomer framing actively discourages action.
Fifth, the actual risks are different than the ones doomers talk about most. The real risks of AI are the immediate, boring, already-happening risks. Hallucinated facts in published documents. Edge case failures in customer service that destroy customer relationships. Drift in long deliverables that produces wrong recommendations. Workforce restructuring that displaces unaugmented workers. These are real risks, and the doomer framing distracts from them by aiming attention at speculative future scenarios.
The shallow doomer position is a pose. It lets people feel they’ve taken a serious stance on AI without learning how any of it gets deployed, and meanwhile the real risks, the hallucinated fact in a published document, the chatbot that loses a good customer, go unwatched by the very people who say they’re worried.
Watch: AI Debate: Doomers vs. Accelerationists, AI News Bulletin
Why are the AI hypers wrong?
The hyper position is the inverse and equally wrong. AI is going to replace every human task. Your job is going away. Embrace the future. Buy the course. Subscribe to the newsletter. Adopt the structure. The transformation is total and the only question is how fast you can ride it. Wrong for five mirror-image reasons.
Every week you spend reading articles about how AI is going to destroy your industry is a week lost.
First, AI doesn’t replace jobs. People using AI replace people who aren’t. “AI is replacing humans” is a marketing line. It lets the people who profit from AI hype avoid the awkward truth. What’s happening is one human outcompeting another. The “AI replacement” framing obscures that this is a workforce restructuring driven by humans, not a technological replacement driven by machines. I walked through that distinction in AI Isn’t Coming for Your Job. Other People Using AI Are.
Second, the productivity gains are narrower than the hype claims. AI is useful for the work in What AI Is Actually Good At. It’s bad at the work in Six Places AI Will Break Your Work. The hype lumps both together and promises blanket transformation. That doesn’t match what the technology delivers when you deploy it.
Third, the most public AI deployments keep reversing. Klarna said its chatbot was doing the work of 700 customer service agents, and now it’s hiring humans back. Air Canada’s chatbot got the company sued and lost. The doctor’s offices that deployed chatbots that trap patients are quietly walking them back. The hyper framing predicted these deployments would succeed. They didn’t. The hyper framing’s track record on its own predictions is bad and getting worse.
Somebody at each of those companies believed the hype, skipped the thinking, and let customers and patients absorb the damage. Nothing else in this debate is worse. The executives who made those calls rarely feel the consequences, and the person stuck in a chatbot loop at the doctor’s office feels all of them.
Fourth, the hyper position is also convenient for people who profit from your inaction. Same trick as the doomer position, in reverse. Every week you spend reading articles about how AI is going to transform everything is a week lost. You didn’t spend it doing the boring, specific work of integrating AI into your own workflow. The hyper content is satisfying. It feels like you’re on the right side. It’s not action. The producers of hyper content make money whether you act or not.
You only benefit if you act, and the hyper framing swaps specific action for vague enthusiasm. That swap costs people real money and real months: courses they never finish, subscriptions they forget, tools bought for a strategy nobody sat down and thought through.
Fifth, the opportunities are narrower than the hype claims. AI doesn’t make your existing strategy work better. AI makes you faster at the parts of your strategy that AI can accelerate, which is some of them, but not all. Adopting AI well requires understanding your own work and figuring out which parts to hand off. That’s not a podcast subscription. It’s thinking about your job, and most people who buy into the hyper framing skip that step because the framing makes it feel like the transformation will happen without it.
The hyper position lets you feel like you’re on the leading edge without doing any of the boring work of figuring out where AI fits in your specific role. It’s a satisfying position. It’s not a useful one.
| Reason | Why the doomers are wrong | Why the hypers are wrong |
|---|---|---|
| First | The technology already shipped; the only question left is how to use it well | AI doesn’t replace jobs; people using AI replace people who aren’t |
| Second | Catastrophe predictions keep failing and the dates keep moving | Productivity gains are narrower than the hype claims |
| Third | Refusing the tool doesn’t make it go away or slow it down | The most public deployments, like Klarna and Air Canada, keep reversing |
| Fourth | Content producers profit whether you act or not | Same trick: vague enthusiasm replaces specific action |
| Fifth | The real risks are mundane: hallucinated facts, customer service failures, drift | Opportunities are narrower and need thinking about your own job |
What is true about AI, between the doomers and the hypers?
The third position is boring, so nobody wants to platform it.
AI is a useful tool with real limits. It absorbs some categories of work very well, breaks badly on others, and most professional roles are going to evolve in ways that require their occupants to learn the tool without becoming dependent on it. The transformation is real but bounded. The productivity gains are real but specific to certain kinds of work. The risks are real but mostly mundane. The opportunity is real but requires actual work to capture.
Anyone over 15 and under 81 has time to learn the tools and integrate them into their work. The blacksmith analogy from The Retraining Question applies. The blacksmiths who learned the new trades after the factories arrived came out ahead. The ones who refused to adapt had the hardest time. The doomers told the blacksmiths to refuse the new trades. The hypers told them that becoming a factory worker would transform their lives.
The boring truth was that they had to learn specific new skills, slowly, and the ones who did the boring work got the better outcomes.
The same is true now. The third position is to identify the routine, repetitive parts of your work, find AI tools that handle those parts, learn to use them, free up the time, and put the time into the parts of your work that AI can’t do. Augmented Beats Replaced. Every Time. walks through what this looks like in practice. The Birth of the Augmented Human walks through what it looks like across professions.
Why the wrong positions are louder
The doomers and the hypers are both producing content for an audience that wants to feel like they’ve engaged with the AI question. The third position requires the audience to do something specific in their own work. That’s harder and less satisfying than reading another opinion piece.
Your inaction benefits both camps. The doomers benefit because every reader who agrees with them is a subscriber, a book sale, a podcast listener, a member of the camp. The hypers benefit the same way. Neither camp benefits if you stop consuming content about AI and start doing specific work to integrate it.
The third position has no industry behind it because it doesn’t sell anything. It’s the same boring, competent response that worked for every previous shift, and it won’t top any bestseller list. It’s also the one that produces an augmented human in 2031 instead of a casualty, and I know which of those I’d want to be.
The shadow side of getting the framing wrong, what happens to a culture when the loudest voices in the room are both wrong about the actual problem, is in The Death of Thinking. Both the doomers and the hypers are providing examples of that failure mode in real time.
My take
Both camps are selling something, and you’re the product. The doomers want subscribers and the hypers want buyers, and neither gains a dime if you stop reading about AI and start working with it. I’d tune both out. Find the routine parts of your own work, learn a tool that handles them, and spend the freed time on what the machine can’t do. It’s boring advice. It’s the same advice that worked for the blacksmiths.
The third position, stated cleanly
AI is here. It’s useful for some categories of work and not others. It absorbs the boring parts of most jobs and leaves the parts that require human judgment, voice, lived experience, and relationship. Anyone who learns to use it well becomes more valuable. Anyone who refuses or who tries to use it for everything becomes less. The transformation is real but boring, and the boring response is the one that works.
Pick one tool this week. Spend an hour. Spend two weeks using it. Repeat next month.
That’s the whole position. The hypers will call it timid and the doomers will call it naive, and I don’t care what either camp calls it. Both have built an audience by swapping specific work for abstract opinion, and I’m angry about how many working people they’ve talked out of learning a tool that would have made their jobs easier, when neither camp will lift a finger for those people once they fall behind.
Between doom and hype sits the useful path. The Birth of the Augmented Human is my book on AI that amplifies your thinking instead of replacing it.
