When talking with Chatgpt, it may be a kind of intelligence beyond the human, although its model is simpler and more simple than it seems.
At the general level, the OpenAI chatbot, regardless of the version-also with the latest model, GPT-5-is based on 2 basic concepts: The understanding of natural language and the generation of answers.
At the first point, entries or Prompts become numerical representations that make Chatgpt understand us, what is known as the embedding; In the second, it combines this sequence and predicts the answers according to your training model.
Due to this operation, the training is key, as well as the data sets that Chatgpt has been able to learn, mostly with all the information available on the Internet, something that has not happened without criticism for OpenAI and other companies.
But,What happens if we ask Chatgpt with extremely rare prompts? Is it still reliable or is you able to understand what we are asking for?
We have done the test and, although Chatgpt is no divine or metaphysical entity, it has even been able to revive dead languages.
Rebuild a hypothetical ancestral language

Computer today
Protoindoeuropeo is a language more than 3,000 years ago that, supposedly, gave rise to the different Indo -European fragmentations – Sánscript, Latin, classical Greek, germic, etc. – and, based on reconstructions, it has been hypothetized about it.
This reconstructed language It does not have established fixed grammarbut refers to comparative methods based on other languages that have been studied and have survived until today.
Interestingly, Chatgpt knows this and has been able to rebuild a text in protoindoeuropeo with 4 stanzas, as if it were a kind of poem loaded with symbolism, with the characters that would supposedly be part of their writing.
Zero-G chess game

Computer today
For this test, I have asked Chatgpt that, directly, imagine a chronicle halfway between poetry and technology that illustrates a space chess game – zero gravity – between a telepathic octopus and a robot with logical capacity.
Through different symbols and Latin writing, has turned this game of chess into a kind of enigma to decipherwith a detailed glossary that explains everything that has been invented.
In this game of intergalactic chess, Chatgpt has chosen to leave the end open, in what has come to call an «infinite orbital resonance.» A surreal outcome that demonstrates that any type of code resists this AI.
Latin exorcism that goes wrong

Computer today
Another characteristic of Chatgpt, and that has improved with GPT-5, is the structure of fiction stories. In this case, he has been able to write a medieval liturgical story in Latin ecclesiastical tone about an exorcism that ends badly.
Including satire and humor, chatgpt 5 has structured this kind of story in a Latin story with enough comedywith a priest as the protagonist of this ritual that does not end well, with a closure that acts as a parable.
Of course, with the initial prompt in Latin and a story completely in the same language.
Instructions in Klingon and the heart of the AI

Computer today
Klingon is the language of the homonymous race that acts as a villain of the Star Trek universe. Invented by Marc Okrand, this language was built by a fairly simple structure: object, verb and subject.
To complicate the instructions that I have given to chatgpt, I have asked for that offers different lines in Klingon, but also built in X86-64. The result has been a code with comments in Klingon.
Interestingly, Chatgpt 5 has chosen a structured code in 14 different parts in the purest style of an epic epic.
The Language of the Valinor elves

Computer today
Jrr Tolkien was able to invent a complete language to demonstrate his skills as a philologist, and the final result – with his variations – was the Qenya, the language of the Valinor elves in the Lord of the Rings, the most widespread next to the Sindarin, one of the 15 languages created by the author.
Chatgpt 5 has also had the ability to invent a language derived from it, with touches of others, to perform a composition in 10 sections that show different syllogisms «Logical propositions locked in themselves.
Although the result shows that, at the logical level, Chatgpt 5 is not yet able to refine very well what it shows, it represents a creation in an invented language beyond that of Tolkien, but without forgetting the original of the high elphical.
Know How we work in NoticiasVE.
Tags: Artificial intelligence