Search
Zirui Wang, a researcher at Google DeepMind, declared that the model in which he played a role in training has now advanced beyond his own expertise, to the extent that he has nothing further to impart. Following a seven-month hiatus, Google unveiled Gemini 4 Argon, a model that embodies the principle of Recursive Self-Improvement (RSI). This principle is widely acknowledged as the linchpin for transitioning from Artificial General Intelligence (AGI) to superintelligence, and it remains a fiercely debated issue within the realm of AI safety. Gemini 4 Argon has achieved a comprehensive intelligence index score of 53, matching that of GPT-6 Astra, marking a substantial enhancement over its forerunner. Furthermore, its hallucination rate stands at a mere 15%, rendering it the top-performing model among its high-scoring counterparts.
-
C114 Communication Network -
Communication Home
