The Irla Foundation , the Coppieters Foundation and the Accent Obert organization organized, at the beginning of February, the conference 'Language and Artificial Intelligence: challenges and opportunities', where the study 'Artificial intelligence in the future of non-hegemonic European languages' by Albert Cuesta was presented. Currently, in Europe, there are twenty-four recognized official languages, but more than sixty regional or indigenous minority languages that are spoken by about forty million people. The meeting focused on reflecting on the opportunities, challenges and preservation strategies that open up in the era of artificial intelligence. "We live in a time when artificial intelligence is gaining a lot of prominence and, of course, it is not neutral. We live in a time when this tool is taking on a lot of prominence in our lives, in our daily lives. Therefore, we cannot shy away from the role that we want to give it, that we must give it, and we link it to languages, because, in the end, language is the instrument that allows us to exist as a society, as a community," says Raül Romeva of the Irla Foundation.
The conference 'Language and Artificial Intelligence' warned of the vulnerability of languages with few resources in the era of artificial intelligence, which is a reflection and amplification of deep-rooted sociocultural and historical inequalities. "Artificial intelligence encompasses a whole series of realities that, if they are not defined, if they are not explained, if they are not used, do not exist," adds Romeva. "With the study, we have detected four major blocks of opportunities: the preservation and documentation of a lot of oral tradition, the ease of creating personalized educational applications and services adapted to citizens, translation and accessibility, and content creation," comments Albert Cuesta, author of the study. Also, within the framework of the meeting, some examples were mentioned where artificial intelligence is put at the service of non-hegemonic European languages: the Catalan experience with the AINA Program, the Icelandic experience with OpenAI, the Greenlandic experience with the artificial intelligence translation tool for the newspaper Sermitsiaq, the Basque experience with the kAIxo chatbot, the Irish experience with the E-STÓR project and the Welsh experience with Macsen.
In essence, the conference 'Language and Artificial Intelligence: Challenges and Opportunities' highlighted that machine translation, voice recognition and text generation tools can revitalize languages with few speakers, while chatbots and virtual assistants facilitate interactive learning. The meeting also warned about how the lack of data on non-hegemonic languages leads to algorithmic biases, grammatical errors, hallucinations and a linguistic homogenization that perpetuates the dominance of English. "We want to accompany this process throughout so that it moves in a desirable direction," Romeva points out.