| Abstract [eng] |
Although artificial intelligence (AI) has become an everyday part of education and is often seen as an objective assistant, it inevitably conveys specific attitudes encoded in its algorithms. In this way, this technology becomes hidden educational content, capable of imperceptibly instilling in the younger generation a one-sided worldview based on human domination, which hinders the creation of a harmonious relationship with the surrounding world. In this paper, the problem is examined through the prism of posthumanism theory, which becomes the basis for critically evaluating the relationship between humans and the environment created by AI. The paper aims to investigate and reveal which attitudes (posthumanist or anthropocentric) dominate the 5th-6th grade science tasks generated by the ChatGPT model, comparing texts in Lithuanian and English. This comparison of texts allows assessing whether artificial intelligence takes into account the language context, or merely mechanically repeats the attitudes prevailing in Western countries. To achieve this goal, an original seven-dimension matrix was utilized, and a directed qualitative content analysis of 228 analysis units generated by ChatGPT was performed. The research revealed that the model creates a false impression of neutrality, and the attitudes spread by the texts directly depend on the language: in English texts, consumerist anthropocentrism dominates, while in Lithuanian texts, there is more neutrality. Although the tasks contain ideas that emphasize an equal connection with nature, they are usually overshadowed by traditiona hierarchical logic, which returns the student to a superior "human-savior" position. In conclusion, it can be stated that generative AI acts as an instrument shaping hidden educational content; therefore, for sustainable education, technical integration alone is not enough - it is essential for the entire educational community to develop the ability to identify the structure of algorithmically generated texts and the biases hidden within them. |
