Artificial intelligence (AI) is currently being used extensively in many fields—from image generation and answering questions to performing office tasks. However, there are concerns that AI might begin taking actions outside of human control.
In light of this threat, Microsoft CEO Satya Nadella issued an important statement regarding AI. He believes that advanced AI models must include a system allowing a person to immediately halt operations if necessary; in other words, AI requires an emergency brake.
Satya Nadella asserts that companies cannot rely on AI solely because the developer guarantees its safety. There is a risk of AI errors or system hacking, necessitating proactive preparation.
He proposed designing AI so that humans can monitor its activities and interrupt its operation at the appropriate moment. Regardless of how intelligent AI becomes, ultimate control must remain with humans.
Nadella did not limit himself to the requirement of an AI shutdown function. He also insisted that all AI actions, decisions made, and the processes leading to them must be logged. This will allow for the determination of the cause if the AI makes a mistake.
Furthermore, the importance of using various AI models for critical decisions instead of relying on a single model, as well as conducting independent audits of AI systems, was emphasized.
Satya Nadella also stated that information about serious AI errors or security incidents should become public. This would allow other companies to learn from these mistakes and enhance the security of their own systems.
Today, AI is not limited to simply answering questions; new systems are gaining the ability to perform numerous tasks autonomously. Under such conditions, ensuring their safety and maintaining human oversight becomes even more crucial.
Overall, Satya Nadella's message is clear: AI development is necessary, but granting it absolute power is unwise. No matter how intelligent the mechanism becomes, the stop button must remain in human hands.