The web is full of articles of the anxiety and fear of AI taking over, killing humans. If you are the ones who are building it, I get your dilemma.

You have heard the reassurance – “People said the same thing about calculators. About television. About computers. Every generation panics, and every generation turns out fine.”

If you work in AI and you have felt genuinely uneasy about what you are building, someone has handed you the above argument. You have probably handed it to yourself at two in the morning. And it probably did not work.

Here is why it did not work, and what I think is actually going on.

The problem with the reassurance

The argument has a hole in it. If you have felt that without being able to name it, this is the name: survivorship bias.

We remember the panics that turned out to be wrong. Nobody opens this argument with asbestos, or leaded gasoline, or CFCs. And nobody leads with the handloom weavers, who were not wrong at all.

They were skilled craftsmen, among the highest-paid workers in Britain, and the power loom ended them. Their numbers collapsed from around 250,000 in 1800 to roughly 7,000 sixty years later. The surviving petitions and handbills show men asking for a minimum wage and being refused. Their fear was accurate and their trade did not come back. We just do not tell their story, because it does not have the ending the argument needs. We use their name as an insult instead.

So when someone offers you the calculator comparison, some part of you notices that the deck has been stacked. That noticing is correct. Trust it.

I am not going to tell you the worry is unfounded. Since 2023, when Challenger, Gray & Christmas began tracking it as a distinct category, employers have cited AI in 184,538 announced job cuts. That is not a forecast. Those are people.

Though I want to be careful with that number too, because it measures what employers say rather than what is true. Glassdoor’s chief economist has pointed out that a company can name AI as the reason for layoffs without that being the actual reason. Challenger’s own July report describes counting twelve nursing positions at a Bronx hospital as AI cuts because the nurses’ union said so, while hospital leadership called the characterisation misleading. “AI took the jobs” is sometimes a real observation and sometimes a convenient story for a decision made for other reasons.

Either way, the reassurance that new jobs will appear is historically well supported and completely useless to a fifty-two-year-old whose role disappeared last quarter. Both things are true at the same time. Pretending otherwise is how you lose someone’s trust in a single paragraph.

What I think is actually happening

Here is a pattern I have not seen anyone name.

Line up the technology panics in order and look at what each one was actually about.

Machinery: jobs. Photography: creativity. Radio and television: behaviour. Calculators: cognition. Computers: jobs and cognition. Smartphones: cognition, health, and relationships. AI: jobs, cognition, creativity, education, relationships, and autonomy.

Each of those links is worth opening. The 1961 TIME piece is my favourite, because it could run today with three words changed. It quotes a business expert calling automation the nation’s second most important problem, behind only peace. In 1986, maths teachers picketed their own professional body’s conference over calculators, warning of tremendous potential for destruction of children’s arithmetic. In 1954 the Senate ran televised hearings into whether comics and television were producing juvenile delinquents.

The fears do not replace each other. They accumulate.

Each technology inherits every anxiety that came before it and contributes one of its own. Which means the “we panicked before and it was fine” comparison is not comparing like with like. You are not being asked to weather the calculator panic again. You are being asked to weather the calculator panic and the television panic and the smartphone panic at once, plus something new that none of them touched.

That is why this one feels different. It is different. Not because we are more fragile than our grandparents, but because the surface area is larger than it has ever been.

I find this more useful than reassurance, because it explains the feeling instead of arguing with it. You are not being irrational. You are responding to something with more dimensions than any previous version of this.

What the pattern does not prove

It does not prove we are doomed. Accumulating fears are not the same as accumulating harms, and the historical record does show something real. Machines have repeatedly taken a task, and humans have repeatedly found another one. When radio was all we had, we had voice artists. Television produced actors, camera operators, editors, writers, an entire economy that did not previously exist. Technology tends to redistribute human work rather than delete it.

But it does not prove we are fine either. I do not know which way this goes, and neither does anyone telling you confidently in either direction. The people I trust least on this question are the certain ones.

Where spirituality actually comes in

Not as a sedative. I want to be clear about that, because “meditate and stop worrying” is the calculator argument with incense on it.

Here is the more honest version.

You will be here for another thirty, forty, fifty years. Our species has been here a few hundred thousand. The planet, four and a half billion. Zoom out far enough and your entire working life is a rounding error, and so is this technology, and so is the argument about it.

That is not meant to make the worry disappear. It is meant to give it somewhere to sit. There is a difference between a fear that runs your life and a fear you can hold at arm’s length and examine. The zoom-out is what makes the second one possible.

And there is one thing I am reasonably confident about. Whatever these systems become, the faculty in you that notices all of this, that sits with uncertainty, that cares about strangers you will never meet, that asks whether the thing you are building is good, that part is not threatened by better prediction. It is the oldest thing in you and the least automatable.

So sit with it. Meditate, if that is your practice. Walk, if it isn’t. Not to feel better about AI, but to think clearly about it, which is a different thing and a more useful one.

And if the unease is heavy enough that it is affecting how you sleep or work, talk to someone qualified. That is not a failure of practice.

The practical part

Do something today. Learn how these systems actually work, because informed unease is more useful than vague unease and considerably easier to carry. Build something.

The people best placed to shape how this goes are the ones already inside it, feeling uncomfortable about it. Follow that to ask for better governance, controls, make it safer for humanity.

That discomfort is not a problem to be solved. It is what a conscience feels like when it is working. Let it work.