October 27, 2021

I, Science

The science magazine of Imperial College

As computers have become more developed and intricate they have been given responsibility for many tasks that people used to carry out. However, how close are we to them taking over everything we do, resulting in a Hollywood-esque robotic revolution?

Turing intelligence

Scientists have been asking this question for a very long time. The famous mathematician Alan Turing said in the 1950’s that “at some stage … we should have to expect the machines to take control”. He also devised the Turing test for robotic intelligence. In this he proposed that true robotic intelligence would only occur when a human, having a conversation with another human and a computer (both of whom are trying to pretend to be human), is unable to correctly judge which is a computer. In the 1960s, Turing’s ideas were expanded upon by Professor IJ Good, who theorised that if computers were to ever, even slightly, become more advanced than humans. They would then be able to continually redesign and improve themselves resulting in an “intelligence explosion”. He gave this the term “technological singularity”.

Back when these theories were being established, the world of robotics and computing was just beginning, with the supercomputer of the time (IBMs Naval Ordnance Research Calculator) running at a speed of 15,000 operations per second. Fast-forward to today and our fastest computer, Tianhe-1A, can operate at 2507 trillion calculations per second. And the rate at which computers work is set to increase even further.

Moore’s Law

The current trend in computing advancement follows Moore’s Law, which states that the number of transistors which can fit in a given area on a chip doubles every two years. It has been accurate since its inception in the 1960s and is believed will still apply up until at least 2020.

The combination of Moore’s law and technological singularity theory gives humanity a potential problem. Could computer development (following Moore’s law) reach a point of singularity and will it result in a technological apocalypse? Some people think so.

The author and futurist Raymond Kurzweil outlined this in his book The Singularity is Near, (2005). He predicted that by 2029 the computers will be able to beat the Turing test and by 2045 we will have reached the nightmare situation Professor Good envisaged.

His forecasts for post-2045 are rather ominous. He states that Moore’s law will no longer apply as we will reach the limit of how small transistors can become. As this point he theorises that as a result of this, computers will increase in power by expanding in size until the Earth becomes a ‘Hitch Hikean’ planetary computer. These predictions have been criticised by many, who cite the previous failures of similarly outrageous predictions.

It seems highly unlikely that we will end up living in the world Kurzweil predicted. But, it seems prudent that we should consider where the continued rapid advancement in computing might lead us.