Tweet
I am curious how much processing detail a browser AI detector should expose to a normal PC user.
Someone may paste text or choose a PDF, DOCX, TXT, or Markdown file and assume the whole check happens on the computer because the interface runs in a browser. In practice, a site might extract document text locally and still send that text to a hosted analysis service. That distinction matters for privacy, performance, troubleshooting, and user expectations.
The clearest design I can think of would explain:
Disclosure: I work on a small browser detector/reporting workflow, but I am deliberately leaving out its name and URL. I am interested in software transparency and practical PC-user expectations, not advertising a service.
Would you put the local-versus-hosted explanation beside the file picker, beside the result, or in a separate privacy panel? What information would you want before selecting a document?
Someone may paste text or choose a PDF, DOCX, TXT, or Markdown file and assume the whole check happens on the computer because the interface runs in a browser. In practice, a site might extract document text locally and still send that text to a hosted analysis service. That distinction matters for privacy, performance, troubleshooting, and user expectations.
The clearest design I can think of would explain:
- which file types are read in the browser and which are uploaded
- whether the extracted text leaves the device
- what happens when the hosted provider times out or is unavailable
- whether the report is temporary or retained
- what request and file limits apply
- that highlighted sentences and probability scores can be wrong
- that the output is not proof of who wrote the text or whether misconduct occurred
Disclosure: I work on a small browser detector/reporting workflow, but I am deliberately leaving out its name and URL. I am interested in software transparency and practical PC-user expectations, not advertising a service.
Would you put the local-versus-hosted explanation beside the file picker, beside the result, or in a separate privacy panel? What information would you want before selecting a document?