Myths and Machines, part 9

Pontus Wärnestål
6 min readMay 20, 2024

Welcome to the final part of “Myths and Machines,” a series where we’ve woven together the profound narratives of ancient myths with the challenges and ethical considerations of modern artificial intelligence (AI). From the illuminating tale of Prometheus’s fire to the cautionary descent of Orpheus into the Underworld, each myth has offered insights (or at least, interpretations) into our technological practices. Today, we conclude with the stark and powerful myth of Saturn, a tale that resonates deeply with the contemporary devouring of the Internet as we know it. Personally, I enjoy reflecting on the cycles of consumption and creation that define our digital age and the ethical imperatives that should guide us through these transformative times.

Previous texts in the series: part 1, part 2, part 3, part 4, part 5, part 6, part 7, part 8.)

Saturn Devouring His Children: The Self-Destructive Cycle of Generative AI

In the Greek mythology, the story of Cronus, also known as Saturn in Roman mythology, stands as a grim testament to the perils of unchecked power and the self-destructive nature of consuming one’s own progeny. Saturn, driven by the fear of a prophecy that foretold his downfall at the hands of his offspring, consumed each of his children at birth. This desperate act of self-preservation ultimately ensured the very fate he sought to avoid, as his youngest son, Jupiter, escaped and eventually overthrew him.

Peter Paul Rubens — Saturn devours his son. Oil on canvas, 1636–1638, Museo del Prado, Madrid

This mythological tale provides a parallel to the trajectory of generative AI today. As these models increasingly consume AI-generated content rather than fresh, original human input, they risk a cycle of degradation that mirrors Saturn’s cannibalistic folly. Moreover, some of today’s generative AI, both text and image generators, often builds upon the work of human artists without proper compensation, potentially stifling the very creativity that feeds its evolution. This process, akin to a snake eating its own tail, or a father killing off his own offspring, leads to what is known as model collapse — where the quality and diversity of AI outputs diminish over time.

The Cannibalistic Cycle of Generative AI

Generative AI models, such as those used for creating text, images, and music, are trained on vast datasets that include an array of human-created content. This initial wealth of input allows these models to generate outputs that are innovative and, at times, remarkably human-like. However, as these models increasingly feed on AI-generated content, the richness and originality of their outputs begin to wane.

This cannibalistic cycle is a parallell to Saturn devouring his children. As AI models consume their own creations, they gradually lose the diversity and depth that come from fresh, human-generated input. Over time, the outputs become repetitive, derivative, and less creative, leading to a decline in quality known as model collapse. Without new and varied input, the AI’s ability to innovate and produce high-quality content deteriorates, much like Saturn’s lineage weakened by his own actions.

The Ethical Dilemma of Uncompensated Artistic Work

In addition to the technical risks, generative AI also poses significant ethical dilemmas, particularly in the realm of visual arts. Image generators, for instance, are often trained on the works of human artists. These models draw inspiration from countless hours of creative labor, yet the artists themselves are rarely compensated or acknowledged. This exploitation not only undermines the value of artistic labor but also threatens the sustainability of creative professions. This is often not the goal, but an effect of overambition — see the tale of Daedalus and Icarus.)

By failing to support and reward the human creators who provide the raw material for AI training, we risk stifling the very source of innovation that these models depend on. Without new and original works of art, literature, and music, generative AI models will increasingly recycle and degrade existing content, leading to a barren creative landscape.

The Self-Destructive Path and the Way Forward

The myth of Saturn devouring his children serves as a warning for the future of generative AI. To avoid a similar fate, we must recognize and address the self-destructive tendencies of these technologies. Here are key lessons and actions to consider:

1. Ensuring Fresh Input: To prevent model collapse, it is crucial to continuously incorporate fresh, diverse, and high-quality human-generated content into AI training datasets. This approach maintains the richness and creativity of AI outputs.

2. Fair Compensation for Creators: Ethical practices must include compensating artists, writers, and musicians whose works contribute to AI training. This support not only honors their contributions but also ensures a steady influx of new, original content.

3. Transparency and Collaboration: AI developers should work transparently and collaboratively with the creative community to establish fair usage practices and promote mutual benefits. This collaboration can foster innovation while respecting the rights and contributions of human creators.

Take-Away Message

The story of Saturn devouring his children is a powerful allegory for the risks facing generative AI. If models consume their own outputs and exploit human creativity without compensation, they set the stage for their own decline. To sustain the vitality and innovation of AI, we must prioritize fresh input, fair compensation, and ethical collaboration. By doing so, we can prevent the self-destructive cycle that threatens the legacy of generative AI and ensure a future where technology and human creativity thrive together.


There are many articles and podcasts on this topic, too many to list. But here is a recent article on the topic of how we risk destroy the complex online ecosystem that allows writers, artists, and other creators to reach human audiences.

Epilogue: The Endless Endeavor of Ethical AI

I originally set out to explore the three first myths. Then it grew to nine. I have called this ninth episode the last one. Perhaps to avoid hubris and suffer the same fate as another myth that we haven’t touched upon yet: in the realm of myth, few stories resonate with the relentless pursuit of progress quite like the tale of Sisyphus. Condemned by the gods for his hubris, Sisyphus is sentenced to an eternity of rolling a massive boulder up a hill, only to watch it roll back down each time he nears the summit. This ceaseless labor, both daunting and unending, also serves as a powerful metaphor for the modern quest to develop ethical and responsible artificial intelligence.

The stories of Prometheus, Pandora, Icarus, and others have illuminated the myriad challenges and ethical quandaries we face in the age of AI. From the need for caution and foresight in harnessing transformative technologies, to the ethical imperatives of transparency and accountability, these ancient myths provide timeless wisdom for our contemporary struggles. Yet, the journey towards truly ethical AI is not a destination but an ongoing process, much like Sisyphus’s eternal task.

Developing AI that is fair, responsible, and beneficial to all of humanity is an endeavor fraught with complexities and setbacks. Each advancement in AI brings with it new challenges — biases that must be mitigated, ethical dilemmas that must be navigated, and unforeseen consequences that must be addressed. The work is never truly complete. Just as Sisyphus returns to the base of the hill to begin his task again, we too must continuously strive to refine, improve, and regulate our AI systems. However, as opposed to Sisyphus, I believe the stone doesn’t roll all the way down each time. I believe we can make progress — if we genuinely take warning from these stories and myths.

Our myths are there to guide us, to highlight the importance of maintaining a balance between innovation and responsibility, and warning us of the perils of neglecting ethical principles. They remind us that while the tools we create can elevate us, they can also bind us if we do not wield them wisely.

Thanks for reading, and please feel free to spread the word. And please, make sure you hire an Arts and Humanities person in your dev team.



Pontus Wärnestål

Deputy Professor (PhD) at Halmstad University (Sweden). Father of two. I ride my bike to work.