eLLMish: The Emerging Common Language of Human-AI Work

Documents are now routinely read, summarized, and acted upon by large language models (LLMs) and AI agents. We argue that this reverses the economics of the register long treated as the standard of professional writing: polished, allusive, culturally embedded academic English. Stylistic polish can be generated on demand; explicit commitments — scope, conditions, exceptions, responsibilities, intended actions — have become the scarce input. We introduce eLLMish, not a new language but a register of natural language, akin to Aviation English, that remains fully readable by people while making its commitments explicit enough to be interpreted consistently by humans, models, and agents. Drawing on contact linguistics, we describe current AI-native practice as a pidgin, anticipate its rapid creolization, and propose interpretation stability as the testable criterion for writing in the AI era. This brief reports work in progress, cites sparingly by design, and invites collaboration ahead of a fuller conceptual paper.