/AI3h ago

Study Finds Copy-Paste Blocks and Bot Checks Cut LLM Use in Surveys to 4.7%

--0--
Comments
xuan (ɕɥɛn / sh-yen)#745
Original post
xuan (ɕɥɛn / sh-yen)@xuanalogue#745inAI

In a recent study, we also gave clear instructions that LLM use was disallowed, and added a scary pop-up message upon any attempt to paste or drag-drop text reminding participants that LLM use might result in their response being rejected.

xuan (ɕɥɛn / sh-yen)@xuanalogue

For people running Prolific + Qualtrics studies, a few methods to prevent LLM responses: - Disabling copy-paste w Javascript deters most attempts at LLM-use - Turn on Prolific's authenticity checks for LLM + bot use on Qualtrics

2:24 PM · Jun 2, 2026 · 213 Views
Sentiment
Sentiment building, check back later.
Cluster Engagement
-
Views
-
Comments
-
Reposts
-
Bookmarks
Expand data
Posts from X
Most Activity
Most ActivityTimeline
VIEWS212BOOKMARKS1LIKES2REPLIES1
xuan (ɕɥɛn / sh-yen)@xuanalogue

At least according to Prolific's authenticity checks, our resulting responses had very little signs of LLM or bot use, despite our study being the kind of tedious problem-solving task that participants might be tempted to use LLMs for.

xuan (ɕɥɛn / sh-yen)@xuanalogue

In a recent study, we also gave clear instructions that LLM use was disallowed, and added a scary pop-up message upon any attempt to paste or drag-drop text reminding participants that LLM use might result in their response being rejected.

3hViews 212Likes 2Bookmarks 1