Why can GPT-3 be over 15x more expensive for certain languages?

The article discusses the cost differences in using the GPT-3 model across various languages. The author points out that the computational demands and resource requirements necessary for processing language can significantly vary based on complexity and popularity. As a result, the cost for certain languages may be considerably higher, impacting the accessibility and usage of these models in different regions. Calculations indicate that while AI technology is becoming more widespread, its application in less-popular languages may be restricted due to these increased costs. The author concludes by contemplating how these differences might affect the future of language technology and what measures could be taken to make artificial intelligence more accessible for all languages.