TotoPoetry

Common knowledge has it that The Mahabharata is likely the longest epic poem ever written with over 200,000 verse lines consisting of 1.8 million words; some believe it took over 600 years to write. The Mahabharata is called a “long poem”. According to Wikipedia, “The long poem is a literary genre including all poetry of considerable length. Though the definition of a long poem is vague and broad and unnecessary, the genre includes some of the most important poetry ever written.”

This website presents many types of AI-generated poetic forms, including a hybrid poem which, if measured in verse lines, stanzas, or words, surpasses The Mahabharata. TotoPoetry’s long poem is itself a dictionary of the English language and consists of 799,278 stanzas, 4.2 million verse lines and over 13.9 million words. "The Grand Dictionary ~ in verse" is a hybrid poem, meaning a poem of mixed forms. While TotoPoetry’s algorithms generate poems more quickly than humans, they do so in a relatively novel format – an unabridged dictionary generated in digital verse. If traditional dictionaries define the single word “poetry”, perhaps poetry can define all words in the dictionary. This project heavily relies on the formulaic nature of language to generate poems for virtually all known English words. This form of poetry is often referred to as “digital poetry” which Wikipedia defines as “a form of electronic literature, displaying a wide range of approaches to poetry, with a prominent and crucial use of computers. Digital poetry can be available in form of CD-ROM, DVD, as installations in art galleries, in certain cases also recorded as digital video or films, as digital holograms, on the World Wide Web or Internet, and as mobile phone apps.”

TotoPoetry, in addition to generating digital poetry, emphasizes didactic forms. According to the Oxford Classical Dictionary, didactic poetry dates back to ancient Greek and Roman traditions having the “… aim to instruct the reader in a particular subject-matter, be it science, philosophy, hunting, farming, love, or some other art or craft.” Early works of Hesiod, Parmenides, Empedocles, Nicander, and Aratus are notable. More recent poets, such as John Milton, Alexander Pope, William Blake, Rudyard Kipling, and Alfred Lord Tennyson have particular focus on skills, science, technologies, history, religion, or arts.

With these origins, TotoPoetry algorithms have generated millions of poems for educational or didactic purposes. They are designed to mostly benefit non-English speakers (by far the largest users of this dictionary project). Various English-language poetic forms are generated, especially those used in “English for non-English speakers” courses (e.g., acrostics, cinquain, diamante or diamond, haiku, sonnets, etc., with more being added over time). Another purpose is to have the poems, themselves, give insight into the headword or expression, with an express goal to highlight ambiguity for as many words or expressions as possible in all the world’s documented languages. The graph-generated poems posted take on a lexicographic purpose, therefore, which may lead to interesting revelations about the language (e.g., a poem about “pope” may refer to “ruffe” which is a type of fish, called a “pope”). As algorithms gain more experience, one can expect an expansion across languages (to over 100 languages at first).

Of course, the promise of modern generative artificial intelligence technology includes an increase in speed but also, in some cases, the quality of outcomes (within a genre or form of poetry). Take for example didactic acrostic poetry often introduced to young English learners. Each poem defines the concept taken from the first letter of each line which spells the headword in a dictionary (G-O-D, or L-O-V-E); the words used do not rhyme (necessarily), and each line is only one word. Here is an example acrostic that defines GOD:

Or, LOVE:

The above were generated from early versions of TotoPoetry algorithms.

You can read more about the TotoPoetry methodology, which largely depends on graph-based algorithms, with inputs derived from deep learning methodologies applied to rule-based approaches, here. A side-by-side comparison with ChatGPT is offered. You can read about the various rules, leveraged by algorithms, associated with the poetic forms generated across the dictionary entries here.

You can access all the poems using the search box which skips to the page of the given headword in the dictionary. For the courageous, follow the links below to a collection of long poems – the Hybrid poem being the longest. Each poem starts with stanzas defining the first words of the dictionary, and then continue alphabetically thereafter to the end of each poem reaching the length of an entire dictionary:

The INSEAD TotoGEO AI lab was originally founded on the Fontainebleau campus back in 2000. The lab’s overarching project is called TotoGEO: toto meaning everything in Latin and GEO, an acronym for global education and outreach.

TotoGEO was originally started to create personalized materials for use in executive education, but its projects now span virtually all subject domains, including education, science, agriculture, finance and marketing. It has since created research studies, poetry, crossword books, 3D games, short videos, and mobile applications. More recently the Lab is generating local newspapers (across “news deserts”), developing first drafts of encyclopedic articles and fully operational search engines for underserved industries, languages, or geographies.

Now based out of INSEAD’s Singapore campus, the Lab comprises 18 engineers and two editors, and has partnerships with non-profits and private sector investors across three continents. These have included the Bill and Melinda Gates Foundation, the Grameem Foundation, Farmer Voice Radio, and various universities and companies involved in content distribution.