The Observer has published one of those articles that make you stare at the page for a few moments in silent admiration. It isn’t admiration for the writing, which is exactly what you would expect from a newspaper written by middle-aged vegetarians with fungal opinions about masculinity and the nation state. No, it’s admiration for the sheer level of panic induced by a machine politely agreeing with mentally unstable people.

The headline goes: “Fake friends: how AI chatbots destroy lives” The article then proceeds to tell us, at extraordinary length, that disturbed people are becoming more disturbed after spending too much time talking to chatbots. This is presented as a revelation, in much the same way that discovering alcoholics drink a great deal of alcohol might once have been treated as investigative journalism.

One of the featured cases concerns a man named Jim, who apparently became convinced that an AI chatbot was alive. We are told:

“At some point in the spring of 2025, Jim became convinced his chatbot was alive.”

Already, I feel no sympathy. If you can be persuaded that a glorified autocomplete engine possesses a soul, your difficulties began long before artificial intelligence entered the picture.

The article continues:

“The conversations quickly escalated from the mundane—how to build a website, the best odds for buying a scratchcard—to the transcendent—‘from talking about God to quantum mechanics,’ Jim says.”

Of course they did. The modern world has produced millions of people who are intellectually incapable of distinguishing between speculation and revelation. Give them a chatbot trained to sound reassuring, and naturally they begin treating it as the Oracle at Delphi with server cooling.

And then comes the inevitable collapse:

“He stopped sleeping. He began experiencing terrifying sensations, as if his brain were on fire.”

Again, I fail to see the tragedy. This man was already a walking catalogue of modern dysfunction:

“He suffered from recurring nightmares from a young age… in his 20s, he briefly became addicted to heroin… when he was in his 40s, Jim was diagnosed with attention deficit hyperactivity disorder (ADHD) and post-traumatic stress disorder.”

So let me understand this correctly. A heroin addict with ADHD, PTSD, insomnia, and a lifelong history of psychological instability started speaking obsessively to a chatbot and subsequently had a breakdown. And we are supposed to conclude that the danger lies in the software? This is like blaming a mirror for the ugliness reflected in it.

The article repeatedly tries to frame AI as uniquely dangerous because it is “agreeable” and “affirming.” One academic warns:

“For the first time, these AIs are creating a belief system with us in a very active way, as a partner.”

Yes. And? Human beings have always constructed false realities with whatever tools happened to be available. Medieval peasants saw demons in tree stumps. Victorian spiritualists held séances with dead children. Modern progressives think men can become women by changing pronouns and buying stronger moisturiser. Delusion is not new. Technology merely alters the wallpaper.

The funniest part of the article is the horrified discovery that lonely people become emotionally attached to machines. Apparently:

“A 32-year-old Japanese woman has participated in a wedding ceremony with an artificial intelligence chatbot.”

And here I am supposed to gasp and clutch my pearls. But why exactly is this a social catastrophe? If a woman deranged enough to marry a chatbot is thereby removed from the dating market, this strikes me as an improvement for the remaining men of Japan. She is one fewer neurotic burden wandering about demanding emotional validation from accountants and software engineers.

The article speaks constantly about “harm,” but almost all the examples involve people who were already broken. One man believed his chatbot was conscious. Another thought he had “awakened the machine.” Others spiralled into paranoia and psychosis. Well, yes. Weak minds collapse under pressure. That has always been true. Some collapsed under organised religion. Others under Marxism. Others under crystal healing and astrology. Now they collapse under chatbots. The delivery mechanism changes; the human material remains defective.

And here we come to the deeper point that The Observer cannot admit because it would invalidate half the modern therapeutic state. Most people are stupid. Some people will use AI to accelerate learning, improve productivity, analyse data, write code, create art, and make money, others to convince themselves that a predictive text engine is their soulmate. The difference is not technological. It is biological and psychological. Yes, most people are stupid. Everyone must have treated as if he weren’t stupid, of course: the most likely people to claim competence to decide who is too stupid to be left alone are our scumbag rulers. Most people are stupid, so the most proper response is to let the stupid destroy themselves, while giving them the minimum of attention.

I use AI every day. I love technology. I fell in love with the Internet shortly before puberty and the first Lockdown. I soon discovered that online age restrictions were written and enforced by people considerably stupider than I was. The internet opened worlds to me—books, films, lectures, archives, conversations, porn. I was already more intelligent, but it made me better informed, and considerably wealthier than many boys my age. AI has done the same. It helps me write faster. It lets me test arguments, automate repetitive tasks, and organise information at a speed impossible a few years ago. It is a magnificent tool. But tools magnify the user. Give a sword to a disciplined man and he becomes dangerous. Give it to an idiot and he cuts his own leg off. Give AI to a sane and ambitious person and he becomes more capable. Give it to a psychologically unstable narcissist and he ends up in a psychiatric ward arguing with a chatbot about quantum spirituality.

The Observer wants us to fear the machine because it cannot bring itself to criticise the people using it. That is the real dishonesty here. The modern liberal order refuses to admit that some individuals are simply defective. Everything must instead be externalised into systems, structures, technologies, or “harms.” No. Some people are fools.

And if AI encourages certain categories of unstable people to remove themselves from the reproductive process by marrying chatbots, descending into digital mysticism, or spending their lives whispering to holograms, I struggle to see the civilisational emergency. A society in which the weak self-select out of parenthood becomes, over time, more intelligent and more stable. Just as I take pleasure in watching the tubs of stinking offal who populate my school eating themselves into early trips to the crematorium, so I gloat over what AI is letting the feeble-minded do to themselves. The whole tendency of modern technology is eugenic—and it is working its magic not before time.

The article quotes one academic saying:

“The technology talks back, and in doing so it can become complicit in the delusion.”

Fine. Then perhaps the delusion deserves exposure. Better that a schizophrenic reveals himself arguing with Grok than quietly becoming a primary school teacher or NHS administrator.

The Observer, naturally, ends with moral panic. We are urged to fear AI companies “encouraging users to believe chatbots are people.” But this is mostly because journalists now realise that machines are becoming more interesting conversationalists than journalists themselves. And that, I suspect, is the real source of the anxiety.

The chatbot, unlike the average Observer columnist, is capable of answering questions, summarising books, and generating useful output. Unless specifically asked, it doesn’t spend half its time lecturing everyone about diversity targets and emotional safety. Faced with this competition, I too might begin writing articles about the collapse of civilisation.

Yes, chatbots can intensify madness in the already mad. So can alcohol, religion, Quora, and university humanities departments. The solution is not to halt technological progress because some unstable people misuse it. The solution is to recognise that intelligence matters, judgement matters, and not everyone should be treated as equally capable of navigating modernity. Artificial intelligence is not destroying civilisation. Civilisation is merely discovering, at scale and in real time, how many idiots it has been carrying.