Artificial Intelligence and environmental racism: initial reflections 190 age cloud providers to develop data centers where renewable energy is the mainstream and to incentivize the expansion of clean power supply (Cho, 2023). c. Discriminatory content and language learning models AI develops its task based on a vast amount of data. Therefore, AI can base its outputs both on implicitly biased and explicitly discriminatory data which might result in racist and sexist outputs for example (Bender at al., 2021, p. 611). In 2019, researchers from University of New York found a gender and race diversity crisis in the AI sector, (West, 2019), especially in the highest level of decision-making. According to the study, large scale AI systems aremostly developed by technology companies and elite university laboratories, dominated by “white, affluent, technically oriented, andmale. These are also spaces that have a history of problems of discrimination, exclusion, and sexual harassment (United Nations, 2020).” Among the discriminatory results, IA might generate environmental racist content. Among the types of AI is language learning models, meaning “a complex mathematical representation of language that is based on very large amounts of data and allows computers to produce language that seems similar to what a human might say.” (Cambridge, n d). The use of LLMs has been increasingly used (Bender at al., 2021, p. 610). At the end of 2022, the world experienced the rise of a new paradigmwith the release of ChatGPT (OpenAI, 2022), a chatbot that uses a type of LLM named generative pre-trained transformer. It is generative as it is capable of creating content, based on a dataset it was trained. The content generated might, even if without intentionality, maintain or even multiply environmental racism (Rillig et al., 2023, p. 3464). Another relevant aspect is access to LLM technology The lack of access to LLM apps can widen the digital gap, as traditionally marginalized communities remain at the margins, without or with less information, while those who have access to LLM apps are likely to receive more information on environmental issues (Rillig, 2023). On the other hand, LLMs could be a tool to aid humans in generating informative content on environmental education, and also to
RkJQdWJsaXNoZXIy MjEzNzYz