Microsoft CEO Satya Nadella stated that the development of superintelligence must prioritize human control and service to humanity. He argued that technology development is meaningless if it cannot be kept under human oversight. Nadella emphasized that the industry should focus on accelerating the adoption of AI while ensuring safety and control, rather than simply slowing down progress.
The company plans to release a code of conduct for its proprietary MAI models. Unlike traditional internal policies, Microsoft intends to open this document for public consultation and external review. This approach aims to incorporate broader societal perspectives into the development trajectory of the models, moving away from unilateral corporate decision-making.
Nadella stressed that businesses deploying AI should maintain full control over their proprietary knowledge assets. He advocated for organizations to build independent learning systems that do not rely on a single model provider. This strategy allows companies to embed their own knowledge into controlled model parameters, ensuring the autonomy of their data and intellectual property.
Regarding safety, Microsoft supports research into model alignment and the implementation of embedded evaluators to monitor AI behavior continuously. Nadella noted that alignment with human goals must be a core design requirement from the outset. He also called for global, cross-industry, and academic participation in AI governance to prevent control by a few entities.
