Post by Samrat Chowdhery

Decades-old research papers are failing AI detection tests. This exposes a major myth we have about generative writing. Some professionals are more impacted by AI writing than others—journalists, researchers and lawyers—as LLMs have gorged on reams of digitised content from their high-authority fields, which is now, perversely, causing their work to be disproportionately flagged as AI-generated. This has given rise to a troubling trend. I see writers intentionally abandoning the clear sentence structures they spent decades mastering. People are second-guessing their own voice. Some professionals are literally introducing typos, awkward phrasing and grammatical inconsistencies into their copy just to prove human authorship. The key point is that these professions developed their specific formats for a reason: they are simply the best ways to communicate complex ideas. Think of the em dash. Journalists rely on it to insert supplemental facts or to add another dimension to a sentence that already contains a comma-separated list. It is prescribed AP Style, used for generations. Now journalists and copyeditors are fleeing from it to appease an algorithm. The same logic runs through other fields. Researchers are trained to hedge with specific formulations: "the findings suggest”, "further research is warranted”, “results should be interpreted with caution”. Law has its IRAC structure, its Latin maxims, its precisely nested conditionals. These are functional conventions that made their way into training data and now return to haunt their practitioners. Which brings us to a massive misconception. When a detector flags your work, it does not mean you are writing like a machine. It means the machine is writing like you. We therefore need to reframe how we view LLM output. It is not AI writing. It is collective writing. The algorithm is simply holding up a mirror to the collective consciousness, thoughts and established frameworks of human professionals. Do not abandon the gold standards of professional communication just because a machine finally learned how to mimic them. #AIWriting #ContentStrategy #Journalism #Research #LegalTech #FutureOfWork