The Technological Singularity: A Sixty-Year-Old Idea That Needs Scrutiny

In 1965 the British mathematician Irving John Good wrote that an "ultraintelligent" machine, one that far surpasses all the intellectual activities of any man however clever, could design even better machines, so that there would be an "intelligence explosion" leaving the intelligence of man far behind. Then he added a clause that quoters tend to drop: the first such machine is the last invention man need ever make, provided that the machine is docile enough to tell us how to keep it under control.

Three names, three estimates

In March 1993 Vernor Vinge presented his paper at a NASA-sponsored symposium and opened it by saying that within thirty years we would have the technological means to create superhuman intelligence. He said he would be surprised if this happened before 2005 or after 2030, and he called the turning point a "singularity": a point where our models must be discarded and a new reality rules. In 2005 Kurzweil published The Singularity Is Near, setting the date at 2045 and expecting a computer to pass the Turing test around 2029. His follow-up, The Singularity Is Nearer, appeared in 2024. All of these are their authors' estimates and should not be presented as facts.

Three silent assumptions

First: that accelerating computing means accelerating intelligence. In 1965 Gordon Moore observed that the number of transistors per unit area on integrated circuits would double every year, and in 1975 he revised this to every two years. That describes chip density, not machine intelligence. Summaries of Kurzweil's book note that he himself concedes that computing capacity alone will not create AI, and that understanding human intelligence and good algorithms are also required. The 2025 AI Index reports that the compute used to train notable models doubles roughly every five months; but more input does not prove more meaning in the output.

Second: that the exponential curve does not bend. Critics, among them Paul Davies and Theodore Modis, object that much that looked exponential, such as population and oil production, turned into a logistic curve that flattens. No one has shown that intelligence is an exception, so asserting that acceleration will last is a guess.

Third: that performance is understanding. John Searle's 1980 "Chinese Room" argument holds that running a program is not enough by itself to produce understanding, because manipulating symbols by their form does not supply their meaning. It is a widely debated argument, so we do not treat it as a final verdict, but it reminds us that "behaves as if intelligent" and "understands" are different phrases.

The risk that does not wait for a date

Intelligence does not need to "explode" for the labour market to change or for power to concentrate. According to the AI Index, industry produced nearly 90% of notable models in 2024 (55 models from industry and none identified from academia), and the United States produced 40 models against 15 for China and 3 for Europe. Whoever owns the compute and the data holds the keys to the decisions: when a system is released, with what values it is tuned, and for whose benefit it works.

Here we offer a reading: the nearer "singularity" may be a singularity of power rather than of intelligence. It creeps in when, out of laziness or a hunger for speed, we hand decisions that touch people, such as hiring and loans, to systems whose logic we do not understand, and accountability evaporates between developer, operator and user.

What should we do?

• Read estimates against the evidence: who said it, on what data, and what does the speaker gain from the attention?
• Keep a responsible person behind every fateful decision, able to object and correct.
• Invest in what we do well: moral judgement, grasp of context, and asking the right question.

Whether or not Kurzweil's estimates come true, what will be tomorrow is known to God alone, and what is asked of us is to take the means: to understand what we build, to hold its builders to account, and to keep our decisions in our own hands.