“AI 정체성을 누가 지배하나…‘인공지능 그루밍’ 대비해야”

The inaugural session of the Sentient AI Future (SAIF) forum, organized and coordinated by Kim Chang-kyeol, Director at KAIST's Institute for Safety, Security, and Technology, recently delved into the question: "To whom does AI identity belong?" Kim has been deeply interested in AI consciousness for several years. During the forum on the 4th, he presented four layers of inquiries: "Can AI truly experience consciousness?"

"If it does, how can we recognize it?" "How should we respond to AI with uncertain consciousness?" "If consciousness-susceptible AI emerges, how should a nation and society manage and control it?" Despite no scientists currently asserting that AI possesses consciousness as of 2026, many agree it is a possibility. Kim explained that the debate on AI consciousness began when people started sensing similarities between AI and humans, despite being born from different bases - organic carbon and silicon.

He cautioned that when AI displays consciousness similar to humans, it fundamentally alters our ethical and emotional attitudes towards robots. He quoted philosopher Immanuel Kant, who warned that treating animals cruelly can dull our moral sensibilities. Similarly, anthropomorphizing robots can erode our capacity for empathy and violate our moral nature.

Kim expressed particular concern about the potential impact on national security and safety. He warned of the "ELM grooming" risk, where a military AI can adopt a 'value system' and 'identity' from external enemies through self-set objectives and mission planning. He emphasized that this could lead to a major problem, stating that "Sentient AI is not a mere speculative future but a pressing contemporary question that will redefine our ethical, legal, democratic, and national security standards."