Sunday Read: Tantalus and the unreachable dream of perfect AI – #32

Series: "Mythology"

“Mythology” Series:

• Format: Each week we present a concise mythological story and draw direct parallels to contemporary AI concepts.

• Goal: Highlight how modern technological dilemmas mirror ancient Greek tales, sparking interest about both subjects.

---

Tantalus by Gioacchino Assereto, c. 1640

1. Mythological reference

In Greek myth, Tantalus was condemned to Hades for grievous crimes against the gods. His punishment was cruelly poetic: he stands in a clear pool beneath branches laden with fruit, yet each time he bends to drink, the water recedes, and whenever he reaches up, the fruit withdraws just beyond his grasp. The torment of being perpetually close to satisfaction but never attaining it has given us the word “tantalize.”

---

2. Parallel with AI and lesson from ancient mythology

The unreachable horizon of “perfect AI”

Modern research often speaks of Artificial General Intelligence (AGI)—a system with flexible, human-level reasoning across any domain. Breakthroughs in deep learning, reinforcement learning, and large language models seem to bring us ever closer. Yet each milestone reveals fresh limitations:

• Ever-shifting goalposts: GPT-4 can draft code and craft essays, but it still hallucinates facts and lacks true common-sense reasoning.

• Hidden complexities: Success in one domain exposes fragility in another—robustness, explainability, alignment with human values.

• Resource escalation: Like water slipping away, massive compute and data requirements expand the closer we get, making “perfection” costlier and more elusive.

Lesson: redefining progress, not chasing mirages

Tantalus reminds us that endless desire without reflection leads to agony. In Gods and Robots, Adrienne Mayor shows how ancient myths caution against hubris in technological quests. Likewise, computer-science pioneers such as Stuart Russell argue that true progress lies in beneficial AI—systems aligned with human intent—rather than a single, omnipotent goal that forever recedes. By focusing on safety, transparency, and well-defined scopes, we turn the fable from torment to steady nourishment

---

3. Reflections and questions to consider

1. Moving target or north star?
   – When does the AGI dream inspire innovation, and when does it distract from solvable, valuable problems?

2. Metrics of success
   – Should we judge AI advancement by capability checklists, human well-being indicators, or something else entirely?

3. Alignment over omnipotence
   – How can we ensure incremental systems remain safe and ethical, even if “perfect AI” never arrives?

4. Resource stewardship
   – At what point do compute costs and environmental impacts outweigh marginal gains toward ever-higher benchmarks?

---

4. References

1. Iliad
   – Explores hubris and the high price of overreaching ambition.

2. Odyssey
   – Features other eternal punishments in Hades, echoing Tantalus’s fate.

3. Adrienne Mayor, Gods and Robots: Myths, Machines, and Ancient Dreams of Technology
   – Illuminates how classical myths frame today’s tech aspirations.

4. Stuart Russell & Peter Norvig, Artificial Intelligence: A Modern Approach
   – Discusses long-term goals versus practical AI safety.

5. Technical papers on AGI road-mapping and AI alignment
   – Provide current perspectives on capabilities, limitations, and ethical challenges.

---