What A Time To Be Alive

Background

What is the technological singularity?

The technological singularity is the moment when machines become smarter than humans and keep improving themselves faster and faster. From that point, technological progress moves so quickly that we can no longer predict what the world will look like. It is not an invention that is finished on a particular day, but a tipping point towards which many developments may converge together.

A utopian city of white and gold towers on a river, beside a green valley with waterfalls and grazing deer; balloons with gardens float above the city.

Where does the word come from?

In physics, a singularity is a point at which the ordinary rules no longer apply, such as at the centre of a black hole. What happens there cannot be seen from outside.

The mathematician John von Neumann was already speaking in the 1950s of ever-accelerating technological progress that seemed to be heading towards a "singularity" in the history of humankind. The writer and mathematician Vernor Vinge popularised the term in 1993 with his essay The Coming Technological Singularity. The inventor Ray Kurzweil brought it to a wide audience in 2005 with his book The Singularity Is Near.

The idea in brief

The core comes from the British mathematician I. J. Good. In 1965 he wrote: suppose we build a machine that is cleverer than the cleverest human at everything. That machine could then also design better machines than we can. Those new machines would in turn design better ones, and so on. Good called this an "intelligence explosion". The first truly super-intelligent machine would then be the last invention that humans ever need to make.

Why might it happen?

  • Computing power has been growing very fast for decades. Chips contain ever more components, and data centres for AI keep getting bigger.
  • AI improves predictably as it is scaled up. Researchers observe that language models get better the more computing power and data they are given.
  • AI is starting to help with research. AI systems predict the shapes of proteins, find new algorithms and write code that is used to build better AI.

When?

Opinions differ widely on this. Vinge wrote in 1993 that it would happen within thirty years. Kurzweil expects human-level AI around 2029 and the singularity around 2045. Others think it will take far longer, or never happen.

Criticism and doubts

  • Intelligence is more than computing power. Faster chips do not by themselves make a machine clever or creative.
  • Growth runs into limits. Energy, raw materials, data and the laws of nature set boundaries. Much that looks exponential at first flattens out later.
  • The risks are great. Many researchers, including the Nobel laureate Geoffrey Hinton, warn that an AI that is smarter than humans could be hard to control.

Why WATTBA writes about it

WATTBA follows the news that bears on this question: AI, robotics, computing power, biotech and energy. Beneath each article it says what the news means for the road to the singularity. Not to claim that it is certain to arrive, but to show which steps are being taken right now and which are still missing.