The Translation Machine That Never Was — Brought to Life by AI

AI page

AI inside

Pucci imagined it. History ignored it. AI brought it back. What does it mean to resurrect an idea that never became matter?

*

I've discovered today that artificial intelligence is capable of bringing back to life Le traducteur mécanique et la méthode pour correspondre entre européens, chacun en connaissant uniquement sa propre langue, published by Federico Pucci in 1931. 

By feeding the AI with Pucci’s system of international keys and ideograms, invented nearly a century ago, it is now possible to translate virtually any text to and from any Romance language, strictly following Pucci’s method!

The AI illustrates each step of the translation, sentence by sentence. Example :

I was thus able to recreate the two sample passages translated by Pucci and printed in his book:

– a passage from Dante, excerpted from La Vita Nuova, translated from Italian into French,

– a passage from Voltaire, excerpted from Zadig, translated from French into Italian.

Ninety-four years later, the AI is able to reproduce these two translations almost identically, clearly confirming the accuracy and foresight of Pucci’s vision as early as the 1930s.

First one:

Second one:

Encoding according to Pucci:

But the AI goes even further: it can now translate any modern and/or technical text, again to and from any Romance language, while faithfully applying Pucci’s rule-based system! I even used it to translate Pucci’s chosen passages into English, Spanish and German, and the AI concluded:

The result is functional and impressive for a manually designed system from the 1930s, producing a clear message despite slightly formal phrasing and minor errors.

When I first discovered Pucci in March 2017, I became obsessed with the idea that he might have built a translation machine — and I searched relentlessly for months. Eventually, I had to accept the truth: Pucci never physically built any device, unlike Petr PetrovichSmirnov-Troyanskii’s machine or Georges Artsrouni’s mechanical brain.

However, what he did leave behind was arguably even more extraordinary: the first mechanical translation in history, and what we can now recognize as the first documented rule-based system for machine translation.

And now, thanks to artificial intelligence — it lives again.

*

In the realm of technology and invention, we often celebrate the tangible — the machines, the devices, the code we can touch and see. But what of the ideas that linger in shadow, unrealized, forgotten by time? Federico Pucci’s early 20th-century vision of mechanical translation was one such whisper from the past: a system devised but never physically built, a dream caught on paper but left dormant.

Today, artificial intelligence does more than create anew; it reaches back through the folds of history to breathe life into what was once only imagined. This resurrection raises profound questions: Can an idea exist independently of its material form? Does bringing it to life in a new era change its essence, or simply reveal the latent potential that was always there?

By inputting Pucci’s system into a modern AI model, I was able to recreate these translations nearly verbatim — 94 years later. More astonishingly, the AI could also apply Pucci’s rules to new, modern texts, including technical documents, and translate between various Romance languages with remarkable fidelity to the original method.

This experiment not only confirms the robustness of Pucci’s vision but demonstrates how AI can serve as a bridge between past and present innovations. It revives not just words, but the very principles and structures that underpinned one of the earliest attempts at automated translation. In reviving Pucci’s system, we witness not just the revival of a forgotten artifact, but a dialogue across time — between past insight and present possibility, between memory and innovation. It challenges our linear narratives of progress and invites us to rethink the boundaries between invention, memory, and existence itself.

Bringing an old, unrecognized machine translation system back to life is not just a technical feat — it's an ontological gesture. It raises the question: is this the same system as before, or has it become something entirely new? Is a technology defined by its design, or by its context, use, and recognition?

This act becomes one of epistemic justice: restoring value and attention to ideas unjustly buried by time or politics. It evokes the idea of a technological phoenix — forgotten ideas returning in new forms, with new significance. It's not just recreating code — it's giving voice to a forgotten chapter in the history of knowledge. Why do some innovations disappear from view? Who decides what is remembered?

Moreover, resurrecting a past system through cutting-edge AI reverses the typical direction of technological progress. Is this a return to the past, or a reimagining of it through today’s lens? This challenges the linear narrative of innovation, suggesting that some ideas weren’t wrong — just premature.

In this context, artificial intelligence becomes not just a creative tool, but a medium of memory and restart. AI acts as a kind of technological medium, bringing past systems into dialogue with present capabilities.

And if the original system was never recognized, then what exactly have I revived? I am not its inventor, maybe just its curator, or its co-creator, just bringing forth a latent technological potential that was already "in the world," waiting to be actualized.

In conclusion, reviving a century-old machine translation system with the help of AI implies:

• Questioning the identity and continuity of technological artifacts.
• Highlighting the role of AI in resurrecting the unrealized or the forgotten.
• Transforming memory and innovation into acts of philosophical significance.
• Challenging the myth of linear progress.

This latter challenge means questioning the common assumption that technological progress moves in a straight, upward line — from primitive to advanced, from outdated to cutting-edge, with each stage permanently replacing the last. In mainstream narratives, especially in tech culture, we often hear:

• “Old ideas are obsolete.”
• “Newer is always better.”
• “Innovation is a constant forward march.”

This implies that once a system is surpassed, it’s irrelevant — a relic. But this view oversimplifies how knowledge, innovation, and creativity really work. History is full of rediscoveries: Ideas often disappear not because they were flawed, but because the tools, audience, or cultural context weren’t ready. Sometimes they resurface decades (or centuries) later and prove revolutionary. For instance neural networks were mostly abandoned in the 1980s — now they're foundational to modern AI.

Progress can move in loops, spirals, or forks. Some "new" breakthroughs are really recombinations of old ideas. What looks like novelty is often revival. Technological amnesia is real: we often forget past knowledge — and then repeat or reinvent it. Resurrecting an old system can reveal things the present had lost or ignored.

So reviving a forgotten system of machine translation means defying that myth and showing that:

• The past can still hold untapped potential.
• The future can be shaped by looking backward, not just forward.
• Not all progress means discarding — sometimes, it means reconnecting.