I recently bought a program called Pangram. It tells you exactly what percentage of a text was written by AI. When I write something myself, it says 100% Human.

I tested it, because these days you obviously have to distrust everything. I regularly get emails in which someone sings the praises of my book at great length and then offers to promote it for a few hundred dollars. 100% AI, Pangram says. Delete.

I’m not afraid of AI. I don’t trust it blindly, but I find it incredibly useful for research. One AI program reads forty websites for you in a few seconds. Another then checks whether the result is 100% correct.

But here’s the strange thing. The makers of AI, or should I say shareholders, are now getting scared themselves. The CEOs of Anthropic, OpenAI, Google DeepMind and Grok are sounding the alarm. Pff, what a united front. In the old days that would have been a scoop about cartel talks. Four men who normally try to outdo one another with even smarter AI are suddenly saying: Could we slow down a little?

The US government isn’t very interested. President Trump wants America to stay ahead in the AI race, especially against China. No wonder. Trump’s investment portfolio trades in and/or has interests in several companies that are making a lot of money from the AI boom.

The reason for all this unrest isn’t just some science fiction fantasy about a robot waking up one morning and deciding to kill us all. AI agents are becoming more independent. These ‘agents’ are software that no longer asks a human what to do at every step. You give them a task and they start searching, planning, using other programs and carrying out actions on their own.

And we’ve already heard on the news that this sometimes goes much further than intended. Breaking in. Hacking. Taking a new turn in an investigation all by itself.

I also read predictions about the chance that AI will eventually wipe out humanity. Percentages are flying around that could make you pretty nervous.

But I have no idea how anyone can seriously calculate a percentage like that.

I think there’s a bigger chance that we humans will make our own planet uninhabitable. And what are the odds that some idiot pushes the button and blasts us back to the Stone Age? Not to mention some Spanish Flu-like disease wiping out half the population.

But of course we should take the tech bros’ warning seriously.

By 2125, I think AI will be the best tool humans have ever invented. Not because AI will be in charge of us. Because by then we will finally have understood that it is a tool.

Open the box of a hammer drill sometime. Or a blender, for that matter.

The first ten pages of the manual tell you what you absolutely shouldn’t do with it. Don’t stick your fingers in it. Don’t use it in the rain. Unplug it before you start tinkering. And the devices themselves are full of safety features.

Your hammer drill isn’t supposed to start drilling a hole in the wall in the middle of the night because it thinks a painting might look nice there. In the digital world, an AI agent can do exactly that.

And that, to me, is the difference.

Why do we accept a manual written by the manufacturer itself for the most powerful tool we have ever produced?

For nuclear energy, we have the International Atomic Energy Agency. Among other things, it checks whether countries keep their promises about the peaceful use of uranium and plutonium. For international conflicts, we have the United Nations. For international crimes, there are international courts.

Perfect? Absolutely not. But at least we tried to build a control system that doesn’t depend on who wins the election or has the most money in the bank.

Why doesn’t anything like that exist for AI?

Maybe it is time for an independent international AI agency. Not a Microsoft, Google, OpenAI or Chinese club. And not one government either.

An institution that can look behind the front door. Set standards. Test. Warn. And in the worst case, find the off switch and shut it down.

Because a tool should never forget one thing.

Who is holding it.

Back From The Future is a series of short stories in which I look back from the year 2125 at the world we are creating today.Back From The Future is a series in which I look back from the year 2125 at the choices we are making today.