Skip to content
DigiCruncher

Why does the same essay get different AI scores on different detectors?

Asked by Ilya V. · AI detectors and false flags1 replyupdated

Short answer

Different vendors, different training sets, different thresholds. I see this every week at the agency. We run the same article through four tools and get four numbers. The trick is not to average them, because they are not measuring the same thing on the same scale.

From the top reply by Colin R. ·
CiteDigiCruncher, "Why does the same essay get different AI scores on different detectors?", https://digicruncher.com/topic/why-does-the-same-essay-get-different-ai-scores-on-different-detectors/, accessed 2026-11-12.
  1. Member ·

    I ran one essay, a 900 word piece I wrote myself for a hobby blog, through three AI detectors in the same afternoon. One said it was almost entirely human. Another called it mixed and highlighted four sentences. The third gave a percentage of AI writing that I would not have believed from a text I wrote start to finish at my kitchen table.

    I understand the rough idea that these are classifiers trained on human and AI text. What I do not understand is why three tools trained for the same job disagree this much on identical input. Is one of them simply bad? Is the disagreement itself telling me something? And if I were a student with a grade at stake, which of the three would I trust, if any?

    I would like answers that distinguish what is known from what is a guess.

    0
    ReplyLink

Plain text or simple Markdown. Links are nofollow. Your email is never shown.

  1. Colin R.Member ·

    Different vendors, different training sets, different thresholds. I see this every week at the agency. We run the same article through four tools and get four numbers. The trick is not to average them, because they are not measuring the same thing on the same scale.