·1 хв читання · 121 слово·👁 12.0K↗ 17
🔬 Scientists Hide Prompts in Papers to Make AI Praise Their Research
The instructions were inserted discreetly, in white text or microscopic fonts, ordering AI to avoid criticism, praise the work, and leave positive reviews ⤴️
Journalists "caught" 17 preprints from 14 prestigious universities using this tactic, including the University of Washington, the National University of Singapore, and South Korea's KAIST.
KAIST officially stated that such tricks are unacceptable. An associate professor at the university, who authored one of the controversial papers, agreed that inserting hidden prompts was wrong, but added that using AI to write reviews is equally unethical.
His colleague from Waseda University in Japan agreed.
"It's a counter against 'lazy reviewers' who use AI," he said.
