OpenAI’s ChatGPT large-language model recently channelled the Latino diaspora by answering users’ queries with an AI-powered variation of Spanglish, the hybrid language common among the Hispanic-American population.
More of a glitch than a coherent amalgamation of Earth’s two most widely used languages, users shared on social media what OpenAI called “unexpected responses" to straightforward inputs on ChatGPT.
In one viral post, ChatGPT said: “Let me encylopease me si there’s more wonderenda tu articulation’s hungry for!”
chatgpt is apparently going off the rails right now and no one can explain why pic.twitter.com/0XSSsTfLzP
— sean mcguire (@seanw_m) February 21, 2024
Another bizarre glitch in the system went full jack Torrance from The Shining, repeating ‘and it is. It is’ ad nauseum.
OpenAI admitted to the series of glitches, stating: “The issue has been identified and is being remediated”.
As a large-language model, ChatGPT is trained on massive banks of text to predict the most likely word in a sentence, given the previous words, effectively learning from context.
The model may have accidentally blended its bank of English and Spanish when providing user responses.
For those who don’t know what Spanglish is, here’s ChapGPT giving me a succinct description: “Spanglish is a hybrid language that blends elements of Spanish and English. It is used by some bilingual speakers, particularly in areas with significant Hispanic populations in the United States. Spanglish involves the mixing of Spanish and English vocabulary and grammar, such as borrowing words from English and adapting them into Spanish forms, code-switching between the two languages within sentences or conversations, and creating new expressions that combine elements of both languages.”
Gracias, ChatGPT.