Nevertheless it probably is much easier (in terms of computational resources needed) to interpret pseudographical picture than to interpret a real graphical pic

Nevertheless it probably is much easier (in terms of computational resources needed) to interpret pseudographical picture than to interpret a real graphical picture. This said we probably can put an LLM which would input a pseudographics-rich terminal UI and describe what's there in a meaningful and concise way. I' mot a huge fan of putting an AI into everything but this seems like a natural way to really help disabled people.