Artificial intelligence is not as close as its fans make out
John Parrington looks at artificial intelligence, and says it will be a long time before computers are smarter than humans
We don’t have to worry yet about Terminators taking over (Pic: Wikimedia Commons)
Will a computer ever be able to think like a person? Could such a machine even surpass human intelligence?
And if so, should we be delighted or terrified by such a scenario?
For those of us worried about making ends meet, the prospect of job cuts, or even the threat posed by global warming, these may seem like abstract questions.
Yet IT entrepreneur Elon Musk recently warned that super-intelligent computers are not only close to becoming a reality, but also pose a grave potential threat to humanity.
Musk made billions from internet company Paypal and was an early investor in the artificial intelligence (AI) firm DeepMind, acquired recently by Google for £255 million.
Yet now he claims that, “With artificial intelligence we are summoning the demon”. He warns of a situation like in the Terminator films, in which Skynet, a military AI system, turns on its makers and destroys human civilisation.
How seriously should we take Musk’s claims?
One problem with AI enthusiasts who say that conscious computers are coming soon is a tendency to underestimate the human brain’s true complexity.
For example, some describe the brain as “wetware”, analogous to computer hardware. This obscures the fact that it’s a far more sophisticated structure than any computer.
Our brains contain 100 billion nerve cells. But each connects to an average of 1,000 others, creating a quadrillion connections, all exchanging electrical impulses. At the moment, no computer has anywhere near this level of circuitry.
But brains are far more than just giant circuit boards. We also can’t ignore the structures of the nerve cells themselves, encoded in our DNA but also shaped by our experiences as individuals.
Ideas
Our brains represent a unique interface between a biological entity that has evolved over millions of years.
These include the unique human traits of self-conscious awareness and an ability to share ideas and thoughts with others.
All of this is powered not only by electrical impulses but also by diverse chemical messengers.
Currently even understanding how this complex object functions as a connected whole lies beyond us, never mind trying to mimic it in a machine.
The differences between humans and computers are reflected in what the latter are still unable to do.
A lot has been made of some computers’ prowess at chess. But this reflects their number-crunching abilities rather than real intelligence.
They have proven useless at games like poker or the board game Go, which don’t involve calculations, but do require intuition, creativity, and even empathy. I’m sceptical about claims that a computer will soon think like a person, or even become more intelligent.
Where I would agree with Musk, though, is the need for scrutiny of new developments in science and their potential risks.
But who will carry out such scrutiny? Ordinary people have little input into the development and application of new technologies, particularly those with military potential.
Only in a socialist society will we be able to realise the potential of new technologies for improving the lives of ordinary people, and to properly assess their risks.
A character in the Terminator films says, “There’s no fate but what we make for ourselves.”
Ordinary people can decide whether new technologies make Earth a paradise or a hell. But we must first reclaim the world from the minority who rule us.
