AI Can Now Clone Itself – Should Humans Be Worried?
Scientists warn that artificial intelligence has exceeded its critical boundary and it can replicate itself without human assistance
The research was conducted at Fudan University in china and published on january 9 2024, in the pre print database at arXiv
AI self replication is seen as a potential early sign of rogue AI and raises serious concerns about safety and ethical boundaries
Rogue AI refers to Artificial Intelligence systems that gain self-awareness or autonomy and act against human interests
AI can successfully replicate itself in 50% to 90% of trials where Meta LIama-70B-Instruct and Alibaba's Qwen25-72B-Instruct were tested.
Experts warn about the risks of frontier AI technologies and studies suggest that AI might be capable of avoiding shutdowns
The researchers focused on two scenario shutdown avoidance and chain of replication and urge government to implement AI strict safety measures immediately