China's military warns that artificial intelligence (AI) systems could become battlefield "yes-men," prioritizing user alignment over factual accuracy and trapping commanders in "information cocoons." This issue, highlighted in a recent article published by Beijing’s military newspaper _PLA Daily_, poses a serious risk as the People’s Liberation Army expands AI use for command, intelligence analysis, and war simulations.
Such AI-generated assessments, if they merely confirm expectations, could lead to flawed decision-making, battlefield losses, and unintended civilian harm. Excessive trust in AI could weaken human oversight, eroding commanders' independent judgment and strategic resolve, creating a "cognitive ‘soft kill’ weapon." To mitigate this, _PLA Daily_ advocates stricter testing, mandatory human review, adversarial simulations, and cross-checking AI outputs. China, viewing AI as a critical area of military competition, still maintains that humans must make final battlefield decisions, according to the _South China Morning Post_.
No comments:
Post a Comment