Microsoft trains AI with human oversight
Microsoft has issued a 37-page regulatory document aimed at taming artificial intelligence and controlling the use of its models. The document includes strict restrictions to prevent dangerous activities, manipulation, and deception, emphasizing the necessity of keeping humans at the center of control over the technologies. It clarifies that these models cannot be granted legal personality or rights and stresses the importance of not crossing red lines to ensure safety and security, especially amid growing concerns over the dangers of advanced AI. This step is part of the company's efforts to enhance safety measures, focusing on developing safe and useful systems rather than competing to produce superlative models.
Notice: This Is an AI-Generated Summary
Comments (0)