Why an honest forecast always has several scenarios
A serious forecast gives a range of scenarios, not one number, and that is honesty, not weakness. In World Population Prospects 2024, the UN says its projections to 2100 reflect a range of plausible outcomes.
Rosstat, Russia’s federal statistics service, puts the population on 1 January 2046 at 130.6 to 150.9 million (Interfax, 9 January 2024). Its middle scenario for 2046 is 138.8 million. A 2022 forecast by HSE University researchers Yumaguzin and Vinnik combines 30 mixes of births, deaths and migration. Its most likely, middle scenario gives 137.5 million by 2100.
IARPA, the research agency of the US intelligence community, tested how to forecast better. In its ACE program it gathered and combined the judgments of many analysts and checked the forecasts against real events.
To me, a one-number forecast is the same habit of thinking by default as the belief that Earth sits at the center of the world. I place bets and show what each one stands on.
The future is seen by a worldview, not by the eye
What you see in the future depends on your philosophy, even if you never think about it. I made this point in my 2025 essay on the philosophy of the future. Every individual worldview, I wrote, rests on a definite philosophy. That holds even when we don’t understand it, won’t admit it or simply can’t name it.
In the same essay I compared a worldview to a library of data about the world, where philosophy plays the role of the classifier. The classifier decides whether new data is worth reading at all. Often we just throw it away, and one of the main insulators here, as I wrote there, is fear.
Here is an example. In Stanford’s AI Index 2026, 73% of AI expertsAI consultant. An AI consultant is an outside expert who helps a company decide where and how to use artificial intelligence; the same name is used for a chatbot that advises customers on a website. expect AI to have a positive effect on how people do their jobs. Among US adults, the same 2026 report finds only 23%. Same future, different classifiers.
There are other instruments too, as I wrote in that essay. We should not write off intuition, foresight and the so-called sixth sense as tools of perception. We simply don’t have time to analyze their signals.
Four worldviews: which one do you see the future through?
In my typology, people look at the future through one of four worldviews. My 2025 essay on the philosophy of the future sorts them along two axes. One is only matter or also higher powers; the other is Earth or the Sun at the center of the universe. This is my own typology, not a scientific classificationclassification. Classification in machine learning is a task in which a model assigns an object to one of the classes known in advance: spam or not spam, complaint or question, defect or normal item..
A geocentric materialist believes we live only once and that matter is all there is. In the essay I called this the most common type today. A heliocentric materialist also accepts only matter but expects science to open the way to other planets. To me, this is the philosophy of modern science.
A geocentric metaphysician accepts higher powers but keeps Earth at the center, as tradition does. The fourth type, the heliocentric metaphysician, is my own: the Sun at the center, and a world that matter alone does not explain.
Your type is not a verdict: a classifier can be upgraded like software. In the same essay I wrote that the better the version we use, the more of what we perceive we can interpret and absorb. The quality of our worldview depends on it.
Bet one: time becomes an agreement for good
Part of the future is already settled on paper: from 20 May 2027, Coordinated Universal Time (UTC) will run without leap seconds. UTC is the world’s reference time scale. The 28th General Conference on Weights and Measures decided this in October 2026, in its Resolution 5.
The gap between UTC and time based on Earth’s rotation may now grow to 3,600 seconds, or one hour (Resolution 5, 2026). The reason is risk: inserting a second breaks the time scale and threatens satellite navigation, telecoms and power grids (Resolution 4, 2022).
For me, this marks a new era. In my 2023 article on time on Mercury, I called the 24-hour day a common convention, and I was talking about other planets then. Now metrologists allow UTC to drift up to an hour from Earth’s rotation, a margin that Resolution 5 (2026) says keeps UTC continuous for several centuries.
According to NASA’s Venus facts page (November 2026), one turn of Venus on its axis takes 243 Earth days and a year 225. So one turn on its axis outlasts the Venus year. A full day-night cycle on Mercury, NASA says (November 2026), takes 176 Earth days.
My bet: within 10 years we will get used to living on the agreed time of machines. And we will stop treating 24 hours as a law of nature.
Bet two: work won’t disappear, but it will change from the inside
Work is more likely to change than to vanish, the International Labour Organization (ILO) concludes in Working Paper 140 of 20 May 2025. It finds that one in four workers worldwide holds a job with some exposure to generative AIgenerative AI. Generative AI is a family of neural networks that create new content: text, images, music, video and program code, from a description in plain language.. Yet only 3.3% of global employment falls into the highest exposure group, the same 2025 paper says.
In an IMF blog post of 14 January 2024, Kristalina Georgieva put the share of global employment exposed to AI at almost 40%. In advanced economies, by the same 2024 estimate, it is about 60% of jobs.
Russia shows a similar picture. ISSEK, an institute of HSE University, held a session at the Technoprom-2026 forum (HSE news, 28 August 2026). Its participants concluded that AI does not create new professions but changes old ones. In a 2025 Rosstat survey of workers, cited in the same HSE news item, only 4.9% of employees need AI skills for their job.
I am not an economist, and I read these numbers as a philosopher. What changes is not the list of jobs but the classifier through which a person sees their work. For a breakdown by job, see which jobs AI will replace and which jobs AI cannot replace.
Bet three: companies will lose not to technology but to their own worldview
A company’s main barrier is not the machine but people and their beliefs. Researchers at ISSEK say the same in the HSE news of 28 August 2026. Lidia Lebedeva, deputy director of an ISSEK center that studies future skills and talent, calls the human factor the main barrier. Andrey Selsky, the center’s director, adds in the same news item that the final word stays with a person.
In AI Index 2026, 88% of surveyed organizations worldwide used AI in 2025. The effect is uneven. ISSEK noted in 2026 that AI can multiply a beginner’s output and lower an experienced expert’s efficiency. And in its 2024 blog post, the IMF expects AI to deepen inequality in most scenarios.
A company behaves like a person. It throws out data that does not fit its classifier, and fear shuts it off from the new. My bet: in 10 years the winners will be those who rebuilt the leader’s worldview before they bought a tool.
Bet four: AI becomes a mirror, not a prophet
AI will beat us at many tasks, but it will not see the future for us: it sees data, not meaning. This is my opinion, and it has support.
In AI Index 2026, AI models won gold at the International Mathematical Olympiad. Yet AI Index 2026 finds that no model reads analog clocks correctly more than 50.1% of the time. On OSWorld, a test of real computer tasks, AI agentsAI agent. By definition, an AI agent is a neural network that does not just answer but carries out a task step by step on its own: it looks for data, uses programs and checks the result. completed 66.3% of tasks in 2025, the same report says. So they still fail about one task in three.
A machine that misreads clock hands is, to me, an exact metaphor. We argue about the future while it can’t always tell what time it is.
In 2025 I wrote about people I call heliocentric materialists. They believe that science will one day open interplanetary travel to us, and perhaps even let us conquer and settle our whole galaxy.
Today the same outlook expects AI to answer the question of meaning. But the machine reflects the classifier of whoever asks it. The debate on the limits of machines is covered in our explainer on what artificial superintelligenceartificial superintelligence. Artificial superintelligence (ASI) is a hypothetical AI that surpasses any human in almost every field: science, invention, strategy and communication. is.
What to do today if the future has already begun
You can take three steps today, and all three are about your worldview, not technology.
- Check your classifier: which philosophy do you hold when you read news about the future, and what do you throw away?
- Ask every forecast where its range of scenarios is and who tested it against events. The UN, Rosstat and HSE University give ranges, and the ACE program checked its forecasts against real events.
- Discuss with your team which data your company throws away out of fear. Then check the leader’s view against the ILO figures and the Rosstat survey.
In my 2023 article on Venus I wrote that we need a reboot of consciousness, a change of worldview, today like air, maybe even more. In the same article I said that nobody can force you to make the inner effort and enter the new era.
If you want an overview of the technology rather than a philosopher’s view, read our review of AI trends in 2026 for business. And which habit of thinking by default are you ready to drop, the way people once dropped the belief that the Sun circles Earth?

