// The accelerator
AI & the Singularity
Artificial intelligence is the engine of transhumanism: it accelerates every other technology and raises the question of when machines will surpass us at everything. Here is the state of play, without hype and without downplaying.
01
Where we stand: the steepest curve in the history of technology
Modern AI models pass bar exams and medical licensing tests, write production-grade code, discover new protein structures, and carry out multi-step tasks on their own. What was considered decades away in 2020 is done today by freely available tools in a browser. The benchmarks once used to measure AI have been saturated one after another. Researchers keep inventing harder tests, and those fall too.
The fuel behind it: compute, data, and algorithmic breakthroughs. The computing power used for AI training has grown exponentially for years, tech companies' investments have crossed the hundred-billion threshold, and with each generation the models become not only more capable but also more efficient and cheaper to run.
For everyday life this means: AI is already a cognitive prosthesis for billions of people. It writes along, thinks along, researches along. The merger of human and machine did not wait for the first brain implant. It has long been underway, just still via the detour of screen and keyboard.
02
AGI: the next plateau
AGI, artificial general intelligence, means AI that handles virtually every cognitive task at least at human level: not a specialist for chess or proteins, but a universal thinker. Definitions vary, but the direction is clear, and it is the declared corporate purpose of the world's largest AI labs.
On timelines, precision pays: the heads of leading labs, among them Sam Altman, Dario Amodei and Demis Hassabis, name windows from a few years to around 2030. Broad surveys of AI researchers spread much wider, with medians sometimes decades later. Both belong in an honest answer: the people building the systems are more optimistic than the academic average, and both groups have been wrong before, in both directions.
The consensus is something else by now: it is no longer a question of if. Hardly any serious researcher still disputes that general machine intelligence is possible. The fight is about when, how safely, and who will control it. These same three questions also decide how the merger of human and machine plays out.
03
Why merge instead of compete?
A central transhumanist argument goes: if machines become smarter than us, the best survival strategy is not competition but integration. A tool you operate from the outside can be lost, taken away, or turned against you. A capability that is part of you cannot.
Neuralink was founded on exactly this reasoning: Elon Musk's declared argument was that in the AI age, humans need a direct, high-bandwidth interface to the machine to stay relevant, since otherwise we become a footnote to our own creation. You don't have to like Musk to take the argument seriously, but you also shouldn't adopt it unexamined.
The counter-position says that integration doesn't solve the control problem but only relocates it: whoever controls the interface then controls not just your tools but you. Both arguments have substance, and that is precisely why the governance question of open standards, neurorights and independent oversight is not a sideshow but the core of the debate. My personal weighing of it is in the manifesto.
04
The Singularity: 2045?
The technological singularity is the hypothetical point at which AI improves itself and progress becomes so fast that human prediction falls behind. The mathematician Vernor Vinge popularized the concept, and Ray Kurzweil attached the dates, naming the same two for decades: human-level AI around 2029, the Singularity around 2045.
Kurzweil's track record is better than his reputation: many of his predictions about computing power and AI development came remarkably true, while others, nanobots in the bloodstream for instance, lag far behind his schedule. Critics call 2045 marketing with a countdown, while supporters call it conservative. The only serious statement is: nobody knows, and expert estimates range from 'within ten years' to 'never'.
On this platform we document both sides, with sources, names and dates. Because for the practical questions of your life, the exact year of the Singularity is secondary anyway: the road there alone, through AI assistants, brain interfaces and longevity therapies, changes work, health and society so fundamentally that it pays to be prepared rather than surprised.