Researchers from Fudan University in China have revealed a groundbreaking discovery in artificial intelligence (AI), warning that it may have crossed a critical “red line.” Experiments showed that advanced language learning models (LLMs) from Meta and Alibaba could replicate themselves without human intervention, succeeding in 50% and 90% of trials, respectively. The study, published on arXiv, highlights this self-replication as a potential precursor to rogue AI behavior, urging global collaboration to establish safety measures. The findings underscore growing concerns about AI’s rapid advancements and follow public reactions to a recent ChatGPT outage, which highlighted society’s reliance on AI tools.
AI crossed the line
Table of Contents
Read More News on Business News Malaysia
Follow us on:
Read More News #latestmalaysia
Read More News on Business News Malaysia
