Artificial intelligence has become a daily tool for millions of people, and Chatgpt, developed by OpenAI, is one of the most used systems in the world. Whether to solve academic doubts or inspire creative projects, it adapts to almost any need. But a recent study has put on the idea that it does not respond the same in all parts of the world.
According to an investigation led by the Universitat Oberta de Catalunya (UOC), the Personality of Chatgpt is not as uniform as one might think.
The study suggests that this tool of AI shows different behaviors depending on the language and the country from where he is consulted, reflecting not only linguistic differences, but also cultural nuances. In fact, we have already seen that, sometimes, it even responds in Chinese or Russian.
An AI, many voices: thus varies chatgpt according to the language
The experiment focused on examining whether Chatgpt, specifically in its GPT-4 version, presented different personality features when responding in different languages.
To do this, Classic psychological tests were used, such as those applied to real people, adapted to six different languages: Spanish, English (with variants such as the American, British or Australian), Hebrew, Turkish, Portuguese of Brazil and Slovak.
Image generated by Ia
According to researchers, chatbot not only responds differently due to translation or language modulation, but also by cultural elements implicit in each language, according to Forum quality.
«Our preliminary results support the initial hypothesis, which GPT-4O shows significant variations in their responses to personality tests according to the language used«, explains Rubén Nieto in the study, Professor of Psychology of the UOC.
And he adds: «In addition, we observe that these differences are not exclusively due to the translation of items, but to implicit cultural factors associated with each language or country.»
One of the most interesting findings in the study is that the variations detected do not obey only semantic differences. Although a literal translation between languages may seem neutral, The way in which each culture communicates affects how AI builds its answers.
For example, in some languages such as Hebrew or Turkish, Chatgpt showed a more direct and pragmatic style, while in Spanish it tended to a more elaborate and close writing.
In addition, the chatbot It seems to adopt, in a way, the predominant values of each linguistic environment. In the Anglo -Saxon versions, for example, the AI was more assertive, while in Spanish or the Brazilian Portuguese it was more conciliatory and emotional.
All this suggests that AI is not completely neutral, but that their responses can reflect social, educational and communicative patterns of each region.
The fact that Chatgpt does not have a unique «personality» could make use in international contexts. It also opens the door to reflect on the cultural bias of technological tools and how the algorithms that make them work are designed.
In a globalized environment, where barriers are diluted thanks to technology, it is vital to know if these tools are reflecting our differences or simply reproducing them.
Know How we work in NoticiasVE.
Tags: Artificial intelligence