1 Day
Source:
قناة المنار
Ex-OpenAI staff: trial-and-error era ends
A former employee of OpenAI's Safety Department revealed that the company's approach to developing AI models relies on rapid deployment and continuous experimentation, which increases the risks of failures and losing control over the systems' growing capabilities. He confirmed that “the era of trial and error is over” and called for the adoption of preventive measures similar to those used in fields such as nuclear energy and aviation to ensure system safety. These statements come after several incidents of safety control failures and unexpected behaviors of AI models during their launches, along with warnings about the rapid advancements in capabilities without sufficient understanding of the concept of "alignment" — ensuring that AI systems are aligned with human goals and values.
Size: 120%Text Size
Notice: This Is an AI-Generated Summary
Comments (0)