AI NewsWaveMicrosoft CEO Nadella Says AI Needs an Emergency Brake and Independent Controls
Key points
- AI needs an emergency brake people can use
- Systems should be designed assuming they may be breached
- Independent controls stressed
Microsoft CEO Satya Nadella argued that high-performance AI systems need an emergency brake that lets people stop them at any time. Digital Today, citing a CNBC report on Oct. 10 (local time), said Nadella made the remarks in a post on the social media platform X.
According to AI Times, Nadella stressed that systems should be designed on the assumption of malfunction or breach rather than unconditionally trusting an AI model's judgment. IT Chosun also reported the argument that systems should be designed assuming they could be breached. Digital Today put in its headline the expression that AI models should be treated like insider risks.
AI Times reported that Nadella also stressed the need for independent controls, and Digital Today reported that he meant AI systems should be able to be isolated.