Nika (@busmark_w_nika) opened with a specific observation: em dashes, structural tics, phrases that sound polished without saying anything — these patterns make AI content recognizable even when you can't explain why. She asked whether the same logic holds for images and video, and whether any detection tools actually work.
Sixty-seven replies later, the thread had moved past detection into something harder. Some people argued high-effort AI content is already undetectable. Others said effort level is the real signal — not whether AI was used, but whether anyone cared. The tension underneath: is using AI in content deceptive, or just another tool, like spell-check?
The sharpest line was Nika's own follow-up: "when everyone is using anti-grammarly to 'sound human'...everything will be AI."
Good thread if you've been trying to write anything original lately.