What Greek myths can teach us about the dangers of AI

We might think that the conception of robots, AI, and automated machines is a modern phenomenon, but, in fact, the idea had already appeared in Western literature nearly 3,000 years ago. Long before Isaac Asimov conceived the Laws of Robotics (1942) and John McCarthy coined the term "Artificial Intelligence" (1995), Ancient Greeks myths were full of stories about intelligent humanoids.
The fact that these mythical humanoids meet the criteria of modern definitions on robotics and AI is impressive in itself. But what's even more astonishing is that these old tales can provide us with valuable teachings and insights into our modern discourse on artificial intelligence.
Such stories "perpetuated over millennia, are a testament to the persistence of thinking and talking about what it is to be human and what it means to simulate life," historian Adrienne Mayor, writes in her book Gods and Robots: Myths, Machines, and Ancient Dreams of Technology.
Get your tickets for TNW Valencia in March!
The heart of tech is coming to the heart of the Mediterranean
In other words, the desire to reach beyond humans and create non-biological life by endowing intelligence into machines seems an innate part of our nature. And this is why we can find wisdom to inform contemporary discourse in age-old myths.
Through the dread, hope, and moral dilemmas they express, these stories can provide us with an alternative way to process some of the most pressing questions regarding intelligent machines: how far should we go with AI? And what are the looming moral and practical implications of this technology?
To revisit these questions, we'll look into three intelligent humanoids in Greek myth: the Golden Maidens, Talos, and Pandora.
The Golden Maidens: the inherent need for labor-saving technology
The Golden Maidens were built by Hephaestus, the god of fire. They're described as female assistants made of gold who look like living young women, and can anticipate and respond to their maker's needs.
But most importantly, "they're endowed with the hallmarks of human beings: consciousness, intelligence, learning, reason, and speech," Mayor remarks in her book.

There's an immediate parallel we can draw between ancient myth and modern society: one of the main reasons for the creation of intelligent, automated machines is economic, or rather, labor-saving.
The idea that robots and self-acting devices could act as servants (or slaves) was a point also stressed by the famous Greek philosopher, Aristotle. In the first book of his Politics, he contemplates:
If every tool could perform its own work obeying or anticipating the needs of others, [...] and if [...] shuttles could weave and picks could play the lyre by themselves, then craftsmen wouldn't need servants and masters wouldn't need slaves.
The idea, though well ahead of its time, is actually very simple: a society in which people don't have to do the drudge work, and instead, delegate it to machines. And as much as Greek society depended on the institution of slavery to function, we are now creating a new class of mechanical servants.
Think of the vacuum bot cleaners that may stroll across your floors, the surgical robots that perform complex surgical procedures, or the military bots designed to disarm bombs.
This does raise an interesting question though. While we have personal robots that can help us with small tasks, the real wealth of automation will come when entire industries are destroyed and replaced by free workers. Think of self-driving cars removing truckers from work for instance. But unless the money generated from such moves go to the dispossessed, the privileged and wealthy (i.e. those similar to Hephaestus) will benefit most.
This idea is explored further in the continuing myths.
Talos: intelligent machines in the hands of tyrants
Unlike the Golden Maiden, Talos was created to cause harm (as was Pandora, but more on that later).
Talos was a bronze giant robot, again made by Hephaestus. He was gifted by Zeus to his son Minos, the mythical king of Crete, to guard and protect the island.
The guardian robot would throw huge rocks at foreign ships approaching the island, and enemies managed to get on land, he would hug them and burn them alive thanks to his ability to heat up his bronze body.



Talos doesn't seem to possess a human level of intelligence, but he's able to interact with his environment and perform various tasks.
In fact, some mythological variations of his demise hint significantly at the possibility that he's conscious of his existence and that he has a kind of agency.
As Mayor notes, "In these versions, Talos is portrayed as susceptible to human fears and hopes, with a kind of volition and intelligence."
If one looks at the relevant mythological corpus, they will notice that all such machines used to cause harm belonged to tyrannical rulers; in our example, King Minos of Crete, and Zeus, the father of gods and men, in the case of Pandora.
And there's a notable moral in these stories: superior technology can help exercise control.
Think of military robots, for instance, which have been in use since World War II. To give a more recent example, the war in Ukraine has become the largest testbed for AI-powered autonomous and uncrewed combat vehicles, highlighting the excitement of military leaders about the potential usefulness of artificial intelligence technology.
And it's not just war that AI can serve those in power.
It can also be used by authoritarian regimes to track citizens, influence the flow of information, and marginalize dissident voices, as the example of China shows.
At the same time, AI's breadth of applications (in healthcare, finance, e-commerce and so on) is shaping a new battleground of geopolitical power -- as big technological breakthroughs have historically done.



Domination of AI by powerful nations is expected to deepen structural inequalities and contribute to new forms of social and economic imbalance. Similarly, as AI is mostly centralized (meaning that it's limited to the ownership of a single entity), it will further empower the leading Big Tech companies creating it, enabling them to pursue their own agendas.
But have these consequences been debated enough by the nations' regulatory bodies or the companies that are currently developing AI?
"I think the Silicon Valley and Big Tech companies and billionaires control the narrative over AI so much that it creates little space for that kind of debate that's necessary for a technology that grows so massively," George Zarkadakis, AI engineer, futurist, and writer, told TNW.
Unless controlled, legally regulated, and removed from the individual, AI tools won't benefit society in the way we envision. And the danger of them falling into the hands of nefarious actors who could use them to assert dominance is highlighted by the following myth where Zeus' fear of losing his ruling power led to the creation of a perilous weapon: Pandora.
Pandora: surpassing the limits
Pandora was created as an instrument of punishment. After the Titan Prometheus (his name means "foresight") stole fire from the gods and gave it to mankind to help it create technology, Zeus commanded Hephaestus to fabricate Pandora.
The mythical humanoid was designed to be an evil disguised as a gift, something that would make humankind pay for reaching closer to the divine level, as up until then, fire and technology were unique privileges of the gods.
Hephaestus fabricated Pandora, molding earth and water into the shape of a beautiful woman. She was also endowed with treachery, deceit, and seduction. At the end, Zeus gave her a mysterious jar.
After her completion, Pandora was sent to Prometheus' brother, Epimetheus (his name meaning "hindsight"), who forgot the warning never to accept a gift from Zeus. Once on Earth, Pandora opened the jar unleashing all kinds of evil that would plague humankind forever. Following Zeus' instructions, she sealed the jar right before hope could escape, trapping it inside.



"It's unclear whether Pandora has the ability to learn, choose, or act autonomously," Mayor notes. "Her only mission is to open the jar of all human misfortune." In modern terms, "she does what she's programmed to do."
There is a pressing question we can explore using this mythological context: are we suffering from a god complex? And, with AI, dealing with elements we simply don't understand?
An integral element of Greek mythology, which is fully expressed in Pandora's myth, is the notion of hubris. This refers to an act that violates the natural order by disregarding the divinely fixed limits on human action in the cosmos. Such an act is always followed by god-sent punishment to restore balance, as in the case of Pandora's jar.
According to Zarkadakis, there's a lot of hubris in AI as well.
"I think the purpose of God is to remind people that they're not gods themselves. And you know, in the absence of gods, we have this problem, right? We think we are gods because we don't need God anymore. So we're building our own gods that will be our gods in the future," he explained. And machines that would be almost impossible to discern from a human being could also be infinitely smarter, "they would be like a god," he added.
Zarkadakis believes that the ancient myths were trying to prevent us from going down that slippery path; but we're heading there anyway.
This recalls Steven Hawking's warning over the potential danger of AI. "The development of full artificial intelligence could spell the end of the human race," he said during an interview with BBC. "It would take off on its own, and re-design itself at an ever increasing rate. Humans, who are limited by slow biological evolution, couldn't compete, and would be superseded."



So should we open Pandora's jar?
Our choice doesn't differ too much from the one Epimetheus had to make. Much like the ancient humanoid, AI comes with a black box, the Deep Neural Networks (DNN) systems machine learning is based on. This means that while scientists have access to the inputs and outputs AI uses, they don't know how its decision-making process works.
We don't know what's inside the black box, the same way Epimetheus didn't know what was inside the jar. The moral of the myth is clear: think before you act, or act before you think -- and suffer the consequences. And to relate this to our modern debate, unless we seriously consider the possible negative outcomes, it's dangerous to rush into creating something we don't fully understand just because we can.
To avoid opening the jar recklessly, Zarkadakis suggests posing a vital question supported by ethical and philosophical considerations: "What's the end game?" And based on that, "what might be the cost and consequences of the technology?"
"A machine that has full autonomy and is conscious means it's completely free, it can think in any way, and, thus, that it can be potentially dangerous," he explained. "The number one risk is extinction, and it's bad enough in theory to try to build such AI and see what happens."



Zarkadakis remarked that the reason why we're fully autonomous is because as biological social beings we are equipped with ethics and morality.
But teaching ethics and morality to AI systems has been so far unsuccessful. Think of the racism scandals of Microsoft's Tay and ScatterLab's Lee Luda. Or, most recently, Meta's Galactica.
Zarkadakis believes that we don't actually need conscious AI. "I think what we need as human society is to live a better life and having more free time is a big goal to that," he added.
"There's a massive usefulness for artificial intelligence to help us reach that point. What we need for AI is social integration and we should absolutely rethink the autonomy of machines and revise their manifesto."
With this approach, AI could, in fact, be the inverse of Pandora -- a democratic tool that could help make ourselves and the world better. And shouldn't this be technology's mission?
From myth to reality
Thousands of years ago, these three myths illuminated the potential of intelligent machines to serve a good purpose (as in the case of the Golden Maidens) or cause harm (as in the case of Talos and Pandora) -- a potential we're already seeing materializing today.
Most notably, though, they bring forth a set of questions that are vital for our pursuit of AI: whose aspirations will it serve, from whom will it learn, what do we want it to be, and how far should we go with it before surpassing the limits?
Ultimately, AI is much like Pandora's mysterious jar. We don't know what's inside and we can assume it contains both good and evil. In the end, it's all about the role we'll play: will we be like Prometheus and demonstrate the required foresight, or will we be like Epimetheus and act before examining the consequences?
Ancient Greek myth has told us the dangers of AI, it's now up to us to listen.