What happened
Microsoft established the MAI Superintelligence Team to develop 'humanist superintelligence' (HSI), a problem-oriented AI designed for breakthroughs in medicine and materials science. This initiative focuses on creating specialist models with superhuman performance within defined limits, prioritising human control and benefits. Unlike general AI, HSI aims to minimise existential risks by ensuring safety and alignment with human values, with applications such as early disease detection, molecular design, and energy storage. The development occurs amidst increasing regulatory scrutiny of AI in healthcare.
Why it matters
The development of problem-oriented HSI models for medicine and materials science introduces a significant oversight burden for compliance and legal teams. This initiative raises due diligence requirements for demonstrating adherence to evolving regulatory frameworks concerning AI safety, transparency, and accountability in healthcare. The focus on superhuman performance within defined limits, while aiming for human control, creates a control gap in verifying alignment with human values and ensuring the absence of unintended consequences, increasing exposure for platform operators and IT security to novel operational risks.
