The scenario of robots who become a threat to humans has been explored in fiction, but is there reason to fear a real-world rise of the machines?
‘Frankenstein stories’ have changed a lot since Mary Shelley published Frankenstein: Or, The Modern Prometheus 200 years ago.
It was arguably the first work of science fiction and is the cultural reference we reach for when we think of human-made creations turning on their creators.
Tales of anthropomorphic robots (androids) escaping control and running amok have long been popular with filmgoers, from Fritz Lang’s robot femme fatale in Metropolis (1927) to Ava in Alex Garland’s Ex Machina (2014); from RoboCop and the Terminator films through to I, Robot and Autómata.
But how realistic are fears of a robot apocalypse? Are we facing a Frankenstein moment? If so, what form will the monster take?
Shelley’s Victor Frankenstein is an obsessive scientist who secretly animates an artificial human made from bits of corpses. But, for many people, ‘Frankenstein’ brings to mind the monster, not the creator, and especially Boris Karloff’s character in the 1931 film: a lumbering giant with a rectangular head and neck bolts who’s barely capable of stringing two words together.
Readers who turn to Shelley’s novel are often surprised to find the monster, though horrifying in appearance, is sensitive and articulate. Tormented by exclusion from the human race, he persuades his creator to make him a female companion.
When Victor destroys the female creature, fearing they would breed a ‘race of devils’ that could wipe out the human race, the monster starts a murderous campaign of revenge.
Many modern Frankenstein stories involve androids who turn on their human creators.
In I, Robot, androids escape the control of Isaac Asimov’s famous ‘Three Laws of Robotics’ that prevent harm to humans, and threaten to enslave the human population.
In Westworld, the human-looking ‘hosts’ of a Wild West theme park gain sentience and react against their abusive mistreatment by human tourists.
In Ex Machina, a female experimental android uses psychological manipulation to escape her male human captors. But the creation of realistic androids lies decades, if not centuries, in the future. The real threats come from elsewhere.
In July 2015, Stephen Hawking, Elon Musk and more than a thousand leading AI experts called for a ban on lethal autonomous weapons, such as drones.
The Australian military has been extensively involved in drone warfare in joint operations with the US and is now preparing to purchase its own lethal drones. There is currently a legal requirement for humans to authorise decisions to kill, but the line is becoming increasingly blurred.
For a plausible fictional version, check out Charlie Brooker’s ‘Metalhead’ episode in Black Mirror, where a woman desperately tries to escape a robot guard dog that uses heat-sensing, acute hearing and vision, and formidable machine intelligence to track down and kill its human targets.
But there is another side to the Frankenstein problem: the unintended consequences of more mundane AI.
A lot of attention has focused on the first deaths caused by self-driving vehicles, but autonomous algorithms have played havoc with human lives for quite some time.
The US stock market ‘Flash Crash’ in 2010 was triggered by high-frequency trading algorithms – in 36 minutes, it wiped a trillion dollars off the market before human traders regained control.
In 2012, Facebook published the results of an experiment in which it secretly altered the newsfeeds of 700,000 users. Tweaking the algorithm to show users an abnormally high or low number of positive stories produced significant changes in people’s moods.
Since 2017, Centrelink’s ‘robo-debt’ program has caused widespread financial and emotional distress, with tens of thousands of people aggressively pursued for debts they do not owe, due to misplaced confidence in automated data-matching.
In other words, not only do autonomous algorithms already operate in ways that are independent of human control, but they increasingly influence human behaviour in ways we are often not aware of, and certainly do not understand.
Frankenstein’s monster will almost certainly not look human. Killer robots are the least of our problems.
After an incredible FIFA Women's World Cup 2023, ANU experts have chosen 'Matilda' as their Word of the Year.