Frankenstein, AI and humanity’s love of fearing technology
In 1818, the first copy of Frankenstein; or, The Modern Prometheus was published. Two hundred years later, it’s still our go-to monster story, even if the cultural images we associate with it owe more to Boris Karloff’s portrayal of the monster than Mary Shelley’s original novel.
Only a handful of books maintain relevance beyond a decade, let alone 200 years – yet Frankenstein endures to this day and still offers instant shorthand for cultural touchstones. Even the name Frankenstein conjures up images of a frightening hotchpotch concoction that isn’t natural and shouldn’t exist: Frankenfoods, Frankenbabies, and even Frankenalgorithms.
That latter of these is important. Artificial intelligence algorithms are silently changing lives, but not in the dramatic (and abrupt) way a serial-killing monster might. And while plenty of high-profile people can’t help but draw comparisons between Victor Frankenstein’s monster and AI, we can definitely read too much into it.
Dr Beth Singler is an anthropologist researching artificial intelligence at Cambridge University (“I think about what you think about machines that might think” is her one-line summary at New Scientist Live). To her, there’s a synergy – but it’s more in the way we talk about artificial intelligence than any actual narrative thread. “There isn’t actually a how-to-do list in Mary Shelley’s Frankenstein of how to create this creature and yet it’s one of the tropes that’s been picked up by media, saying this is a direct analogy for what we’re trying to do with contemporary technology,” she tells me.
Shelley, she continues, is very vague about Dr Frankenstein’s aims, but adds that it’s clear that the protagonist was trying to create life in stark contrasts to the minds working on AI today. “There are very few people who actually say they’re trying to create life in artificial intelligence,” she continues. “They’re exploring the capacity of intelligence and trying to replicate and magnify in particular directions.”
In other words, Frankenstein is interesting not because it offers an apocalyptic cautionary tale into what might be, but an insight into how we express our fears about the prevailing technology of the time, be that AI or the telegraph. In Shelley’s era that was galvanism: the belief that electricity could be the key to life.
“When Mary Shelley was writing, there were literally people on the street who would use electricity, this new resource, to electrify dead frogs and they would sort of spring them back to life in a sense,” she continues, referring to the way amphibian limbs would jerk to the sudden jet of current. It’s no wonder it caught imaginations at the time, even if Shelley was somewhat vague in exactly what Dr Frankenstein’s process entailed (we have Hollywood to thank for the more explicit explanations).
“I think Mary Shelley was very clever not to be too detailed in her description of how the creature is resurrected because we can then take our 21st century knowledge and kind of fit it into the gaps that she left,” says Dr Kathryn Harkup, a science communicator who literally wrote the book on the science of Frankenstein. “So when we read ‘the instruments of life’ and ‘the spark of life’, we can make an awful lot of assumptions.”
And would these vague terms have alarmed technophobes of the time? “I think there would have been a little niggle at the back of the mind when reading this, thinking ‘Oh God maybe this is just around the corner’,” Harkup reckons.
Shelley’s vague description of the monster’s assembly means that even if you want to view Frankenstein as a prophetic allegory for artificial intelligence, you have limited sources to go on. The mind just isn’t mentioned at all – and it’s only film that refers to a convict’s brain to cheaply explain the eventual murderous intent. We do see a little bit of how the monster learns in the novel, however, as it reports back to Dr Frankenstein that it picked up speech by listening to humans, and mastered reading by absorbing literature… and doesn’t that sound a lot like machine learning?
“Yeah, people try to draw out this analogy, but the kind of learning that Frankenstein’s monster does is very embodied,” says Singler. “That’s not how machine learning runs successfully because it’s quite slow. Just sitting down an AI system to listen to humans to learn would take an incredibly long amount of time.”
To Singler’s mind, this is actually closer to human learning, while machine learning involves constantly “reforming internal algorithms through repetition” – and “that’s not analogous.” Lee Se-Dol, the human Go champion beaten by AI has played something like 5,000 games in his life, she explains, while AlphaGo, the robot that beat him, has played several million matches against itself. “The scale is so grand that it’s almost an entirely different way of learning.”
She’s also not entirely sold on the idea that humans are determined to make AI in their own image, like Frankenstein and his creation. “A lot of AI creators are not necessarily working towards the familiar and the human when it comes to certain domains,” she says. “So yes chatbots obviously need to present a reassuringly human conversation. But no one was really bothered about how familiar and human AlphaGo appeared, and no one’s really that concerned how AI in Machine Vision and facial recognition look.”
READ NEXT: The sci-fi legends who shaped today’s tech
Of course, it’s not so much what researchers are actually doing as what it’s imagined they’re doing, and it’s perhaps telling that Singler mentions fan art showing the very non-human AlphaGo in semi-humanoid form. In that respect Frankenstein’s monster – fictional though it is – still ticks the boxes.
“The creature is its own being, so it is in some respects an artificial intelligence that has been created by man,” says Harkup. “Victor Frankenstein assumes that his creature will love him unquestioningly and will just fall at his feet and worship him. When it doesn’t it’s clearly a shock to him, and I think we have the same fears today. What would that intelligence think of us? How might it behave towards us?”
Fear of the future
To be clear, Harkup isn’t an AI Doomsday evangelist, and she’s quite clear that this isn’t where we are in 2018. But it raises some interesting questions about our current batch of relatively dim consumer artificial intelligences. They are designed, like Frankenstein’s monster, to serve – and currently, they do just that no matter how rude you are. Indeed, Amazon recently produced a Kid’s Echo which will respond more positively when children remember please and thank yous.
Neither of these scenarios – machines that will respond to any order, or those that will only obey when spoken to respectfully – are ideal to Singler. “Both directions lead to either objectification of humans or de-objectification of machines,” she explains. You’re either objectifying the human voice, or giving a machine human qualities. “It’s not that Alexa, Cortana, Siri or whoever would give a damn if you said bad words to them or were impolite, but actually it says a lot about you that as a human being.”
Ironically, of course, this is what led to the downfall of Victor Frankenstein: the monster was abused by humanity, so showed abuse in return. Fortunately, just as the AI rebellion we fear isn’t a realistic fear today, bringing Frankenstein’s monster to life is only slightly less implausible than it was in 1818. Barring improvements to stitching, the main ground we’ve made in the intervening 200 years is in understanding how far away such a fantasy is.
Harkup’s book goes into this in great detail, and while some bits were entirely reasonable 200 years ago (“grave-robbing and anatomy and dissection were quite prevalent at the time”), others would have been enormously challenging. “Victor Frankenstein must have had extraordinary insight into tissue matching given when he was working, or he got extremely lucky with the pieces that he chose,” she explains. “Tissues was pretty much lashed together with thread so, you know the fine detailed stitching that we use today that was developed in the late 19th century long after Victor Frankenstein. So it’s quite likely that his creature sprung a few leaks when he infused it with blood.”Charles Byrne, also known as “the Irish Giant.” Byrne initially took advantage of his unusual proportions to make money, but as people lost interest he “turned to drink and died” at the age of 22. His body was displayed at the museum of anatomist John Hunter – a friend of Shelley’s father. It’s possible she got to see it there.
“Charles had an absolute terror of being dissected when he was dead, but because of his great height, he knew he would be a target for resurrection gangs and anatomists,” Harkup explains. “So he begged his friends to bury him at sea so he would be the beyond the reach of the scientists who wanted to examine him. But as is clear, he did not get his way.”
That horrible bit of trivia is probably the most haunting part of Frankenstein’s legacy today. We know the story’s largely implausible, but that’s with the benefit of 200 years’ worth of hindsight. It’s worth keeping that in mind a simple question when you read today’s scary stories of imminent AI annihilation: just what will the readers of 2218 make of our 21st-century hangups? More importantly, perhaps: will they still use the Mary Shelley’s most famous work as shorthand for the unnatural?