371
Productivity & Workflow357
Software Development253
Automation & Workflow230
Marketing & Growth213
AI Infrastructure & MLOps197
Photography & Imaging170
Writing & Content Creation197
Design & Creative176
Data & Analytics162
Customer Support161
Education & Learning150
Sales & Outreach132
Voice & Speech131
Operations & Admin118
Marcus Olang says writing that looks "AI-like" can reflect Kenyan education and workplace norms, not proof a text was made by a machine.
In short: As people try to spot AI-written text, some are treating certain writing styles as “machine-made”, and a Kenyan writer says that can wrongly target human writers.
Kenyan writer Marcus Olang has published an essay titled “I’m Kenyan. I Don’t Write Like ChatGPT. ChatGPT Writes Like Me.” He argues that the traits many readers now see as clues of AI, like formal transitions, neat lists, and very orderly arguments, are also common in Kenyan classrooms and workplaces.
Olang’s main point is simple. AI systems learn from huge amounts of human writing. So if you were trained to write in a formal, structured way, your work can look similar to what an AI produces, because the AI learned patterns from people who write that way.
This debate is showing up in book and literary publishing. In 2026, some authors faced public suspicion about AI use based partly on detector scores. A review connected to the Caine Prize and its winner Simbarashe Kundizeza concluded there was no evidence of AI use, and the prize stood. NPR has also reported that different AI detectors can disagree on the same text and can falsely flag human writing.
Olang is not saying AI detectors are always useless. He is warning that style is not proof of who wrote something. As schools, publishers, and employers keep using detectors, watch for clearer guidelines and appeals processes, especially for writers whose “formal” style is simply how they were taught to communicate (like being told there is one correct way to format an argument).
Source: NYTimes