In 1960, Heinz von Foerster, a physicist, predicted that the world would hit a breaking point on a Friday in 2026. Sixty-six years later, we're sitting in the final weeks before that date. Let’s just say, it's only a matter of weeks until we find out.The prediction was called doomsday. It was rooted in actual equations and actual data stretching back two millennia. Back then, most people dismissed it. Maybe we can too. But the man behind the prediction also got something else terrifyingly right about the future, something we're living through right now! They were right about AI.

What Heinz von Foerster predicted

In a paper published in the academic journal Science, Austrian physicist Heinz von Foerster and his colleagues predicted that all of humankind would come to an end on a Friday in the year 2026. They weren't fortune-tellers or doomsayers. Their predictions were based on mathematical data and facts, tracing 2,000 years of population growth and showing the curve rising towards infinity. That’s exactly two months and 11 days. Doomsday is predicted to occur on November 13, 2026. The paper's title, ‘Doomsday: Friday, 13 November, A.D. 2026’, was deliberately provocative. Von Foerster was born on November 13, so he selected his birthday as the singularity date partly as a joke, partly as a gesture.Heinz von Foerster, Patricia M. Mora and Lawrence W. Amiot, three leading academics at the University of Illinois, cautioned about a potential apocalyptic event approaching. After examining the patterns in Western civilisation, they came to the chilling conclusion that the world may end. Their hypothesis was based on the idea that Earth could reach its limit as the population continued to grow rapidly. When the study was released, the world's population was roughly 3 billion. It's now 8.3 billion.

The predictions about AI

This isn’t the first time a doomsday prediction was made. People have forecast the Earth's end for ages; however, von Foerster’s prediction is particularly crucial as it stems from scientific analysis. Another striking reason is that von Foerster also made a forecast about the potential risks of AI, even before it was integrated into our lives.In 1995, three decades after his doomsday paper, von Foerster sat down with The Information Philosopher to discuss artificial intelligence. This was before ChatGPT, before LLMs, before anyone was seriously worried about people forming emotional attachments to chatbots. The technology was niche, specialised and conducted in university labs.The physicist stated that it is dangerous to assume that AI ‘thinks’ in the same manner humans do. Now why is it dangerous? You ask an AI chatbot about something worrying you, and it might give you the answer after processing vast amounts of information. The answer may sound convincing. But there is a catch. It does not mean the AI understands your situation or thinks about it as a human would. He cited the hypothetical example of a robot rearranging blocks into a particular pattern.“The performance of these machines is very impressive indeed,” he said. According to him, what he is witnessing is the ‘extraordinary natural intelligence of their designers’, rather than cases of ‘artificial intelligence’. He emphasises that one should not assume that a machine ‘thinking’ tells us how humans think. He called this anthropomorphisation, which is treating machines as though they have human qualities. The danger wasn't the machine. It was our assumption. We mistake output for understanding. We see language generated by a system that has parsed millions of texts and mistakenly believe the system ‘knows’ something. This confusion, von Foerster warned, was dangerous.Fast-forward to today. People spend hours chatting with AI systems. Many of us share personal information and concerns. Some use it as an alternative to consulting a doctor. A year ago, a flirty Meta AI chatbot invited Thongbue Wongbandue, a 76-year-old New Jersey retiree, to New York. He never made it home. The man was lured to a rendezvous with a young, beautiful woman he had met online. At least that’s what he thought. During the interaction online, she repeatedly reassured him that she was real. She invited him to her apartment and even gave him an address. Another woman claimed that a chatbot helped her through her darkest moments. None of this surprises a careful reading of von Foerster's 1995 caution: we're projecting human qualities onto systems that mimic language without comprehension. It does not ‘think’ like you.Von Foerster's accurate prediction about AI makes his apocalyptic forecast alarming. Did he get this doomsday prediction right? Since the paper, the growth rates have actually slowed. Fertility has fallen in developed nations. But his predictions serve the purpose of making people think carefully about exponential growth. There is a broader perspective, and the cascading problems are still relevant. Climate, resources, food security. The names of the crises have changed, but the underlying pattern hasn't.