Part 5/8:
However, the recent report from researchers at the Center for AI Safety introduces a more sobering perspective: their new AI assessment, dubbed "Humanity's Last Exam," reveals that the best-performing AI still struggles to meet even 10% efficiency in answering questions across a variety of scientific disciplines.
The Secrecy of the Meeting
The choice for OpenAI to conduct a closed-door meeting with the government rather than making a public announcement raises immediate questions about the implications of these advancements. It suggests that there may be sensitive user data involved, potentially not in the best interest of public disclosure.