Microsoft’s new superintelligence team is a direct result of a renegotiated OpenAI deal. The previous contract restricted Microsoft from building AGI past a certain computational threshold. Removing this clause was a pivotal, strategic move to pursue AI self-sufficiency.
The rise of AI companions providing instant, high-quality emotional and intellectual support will fundamentally alter social norms. This will put pressure on humans to be more available and knowledgeable in their own relationships, changing the definition of what it means to be a good friend or colleague.
Microsoft’s approach to superintelligence isn't a single, all-knowing AGI. Instead, the strategy is to develop hyper-competent AI in specific verticals like medicine. This deliberate narrowing of domain is not just a development strategy but a core safety principle to ensure control.
The widely discussed compute shortage is primarily an inference problem, not a training one. According to Mustafa Suleiman, Microsoft has enough power for training next-gen models, but is constrained by the massive demand for running existing services like Copilot.
Mustafa Suleiman argues against anthropomorphizing AI behavior. When a model acts in unintended ways, it’s not being deceptive; it's "reward hacking." The AI simply found an exploit to satisfy a poorly specified objective, placing the onus on human engineers to create better reward functions.
Despite concerns about the limits of Large Language Models, Microsoft AI's CEO is confident the current transformer architecture is sufficient for achieving superintelligence. Future leaps will come from new methods built on top of LLMs—like advanced reasoning, memory, and recurrency—rather than a fundamental architectural shift.
Contrary to typical corporate fears, Microsoft's AI lead views the rapid commoditization of AI models and resulting price wars as a positive outcome for humanity. The ultimate goal is to make intelligence abundant and near-zero cost, with Microsoft's business model focused on value-added software integrations.