The Safest Job from AI May Be Writing

Writing may be uniquely resistant to AI disruption compared to other fields because large language models have plateaued in producing authentic, emotionally resonant prose and appear stuck generating generic, soulless text. Unlike coding or math—which have objective verification systems and clear specifications—writing is a "wicked problem" requiring subjective judgment, contextual awareness, and the ability to simulate a reader's mental state, capabilities that current LLMs lack due to their word-probability optimization and absence of genuine human understanding. The author argues this limitation may persist long-term, making writing one of the safest professions from AI displacement.