Yesterday, 08:40 AM
Has AI ever been applied to the decipherment of the Voynich?
Yes, artificial intelligence and machine learning have been applied multiple times to analyze and attempt to decipher the 15th-century [Voynich Manuscript](https://www.ualberta.ca/en/science/news/...cript.html), though none of these attempts have achieved a universally accepted translation. [1, 2, 3, 4]
## Major AI and Machine Learning Approaches:
* University of Alberta Hebrew Hypothesis (2018):
* Method: Computer scientists Greg Kondrak and Bradley Hauer used natural language processing (NLP) and machine learning algorithms trained on the Universal Declaration of Human Rights across 400 languages to identify the underlying statistical fingerprints of the script.
* Result: Their model concluded that the most likely language of origin was Hebrew (assuming the text used alphagrams where consonants were reordered and vowels suppressed). When run through Google Translate, the opening lines produced a rough grammatical sentence.
* Criticism: Medievalists and linguists heavily criticized the method, noting that training on modern languages to decode a 600-year-old scrambled cipher easily led to hallucinated or forced translations. [2, 5, 6, 7, 8]
* Statistical and Structural Analysis (2025–2026):
* Method: Newer AI audits and transformer models (such as machine learning tools comparing cross-language embeddings or structured procedural analytics) have shifted focus away from direct word-for-word translation.
* Result: Rather than finding a hidden message, recent AI-driven structural studies suggest the text behaves like a procedural or algorithmic notation system (similar to a technical or botanical manual) rather than a standard human language or a traditional linguistic cipher. [9, 10, 11]
* Other Machine Learning Tests:
* Researchers have deployed supervised learning algorithms like Support Vector Machines (SVM) and clustering techniques to map character frequencies, evaluate scribal hands (identifying multiple authors), and test hypotheses ranging from Old Turkish to Neo-Latin or constructed dummy systems. [3, 12, 13]
[1] [https://siloe.es](https://siloe.es/en/can-artificial-intel...anuscript/)
[2] [https://www.medievalists.net](https://www.medievalists.net/2018/01/usi...anuscript/)
[3] [https://medium.com](https://medium.com/@alexandrafiona/the-w...c974bd259f)
[4] [https://www.youtube.com](https://www.youtube.com/watch?v=42SJuSniDfg)
[5] [https://www.facebook.com](https://www.facebook.com/groups/25202935...145731868/)
[6] [https://www.youtube.com](https://www.youtube.com/watch?v=P5F-peZ0qu4&t=283)
[7] [https://www.youtube.com](https://www.youtube.com/watch?v=BB5edbbrwTM&t=288)
[8] [https://news.artnet.com](https://news.artnet.com/art-world/ai-dec...pt-1241105)
[9] [https://www.facebook.com](https://www.facebook.com/aurizn/posts/th...531403250/)
[10] [https://www.science.org](https://www.science.org/content/article/...found-them)
[11] [https://wordlift.io](https://wordlift.io/blog/en/unraveling-t...anuscript/)
[12] [https://www.academia.edu](https://www.academia.edu/129055807/_The_...anuscript_)
[13] [https://www.quora.com](https://www.quora.com/Isn-t-it-possible-...ryone-else)
Yes, artificial intelligence and machine learning have been applied multiple times to analyze and attempt to decipher the 15th-century [Voynich Manuscript](https://www.ualberta.ca/en/science/news/...cript.html), though none of these attempts have achieved a universally accepted translation. [1, 2, 3, 4]
## Major AI and Machine Learning Approaches:
* University of Alberta Hebrew Hypothesis (2018):
* Method: Computer scientists Greg Kondrak and Bradley Hauer used natural language processing (NLP) and machine learning algorithms trained on the Universal Declaration of Human Rights across 400 languages to identify the underlying statistical fingerprints of the script.
* Result: Their model concluded that the most likely language of origin was Hebrew (assuming the text used alphagrams where consonants were reordered and vowels suppressed). When run through Google Translate, the opening lines produced a rough grammatical sentence.
* Criticism: Medievalists and linguists heavily criticized the method, noting that training on modern languages to decode a 600-year-old scrambled cipher easily led to hallucinated or forced translations. [2, 5, 6, 7, 8]
* Statistical and Structural Analysis (2025–2026):
* Method: Newer AI audits and transformer models (such as machine learning tools comparing cross-language embeddings or structured procedural analytics) have shifted focus away from direct word-for-word translation.
* Result: Rather than finding a hidden message, recent AI-driven structural studies suggest the text behaves like a procedural or algorithmic notation system (similar to a technical or botanical manual) rather than a standard human language or a traditional linguistic cipher. [9, 10, 11]
* Other Machine Learning Tests:
* Researchers have deployed supervised learning algorithms like Support Vector Machines (SVM) and clustering techniques to map character frequencies, evaluate scribal hands (identifying multiple authors), and test hypotheses ranging from Old Turkish to Neo-Latin or constructed dummy systems. [3, 12, 13]
[1] [https://siloe.es](https://siloe.es/en/can-artificial-intel...anuscript/)
[2] [https://www.medievalists.net](https://www.medievalists.net/2018/01/usi...anuscript/)
[3] [https://medium.com](https://medium.com/@alexandrafiona/the-w...c974bd259f)
[4] [https://www.youtube.com](https://www.youtube.com/watch?v=42SJuSniDfg)
[5] [https://www.facebook.com](https://www.facebook.com/groups/25202935...145731868/)
[6] [https://www.youtube.com](https://www.youtube.com/watch?v=P5F-peZ0qu4&t=283)
[7] [https://www.youtube.com](https://www.youtube.com/watch?v=BB5edbbrwTM&t=288)
[8] [https://news.artnet.com](https://news.artnet.com/art-world/ai-dec...pt-1241105)
[9] [https://www.facebook.com](https://www.facebook.com/aurizn/posts/th...531403250/)
[10] [https://www.science.org](https://www.science.org/content/article/...found-them)
[11] [https://wordlift.io](https://wordlift.io/blog/en/unraveling-t...anuscript/)
[12] [https://www.academia.edu](https://www.academia.edu/129055807/_The_...anuscript_)
[13] [https://www.quora.com](https://www.quora.com/Isn-t-it-possible-...ryone-else)