Microsoft Abandons Artificial General Intelligence: Will Develop Humanist Superintelligence!

Metaverse Planet November 10, 2025Last Updated: November 10, 2025

2 minutes read

Microsoft has announced its new artificial intelligence vision, which it calls “humanist superintelligence.” The vision, announced by Microsoft AI CEO Mustafa Suleyman, rejects artificial general intelligence (AGI), which aims to surpass human intelligence. The goal is to develop a superintelligence that remains under human control, specializes in specific areas, and is focused on solving humanity’s problems.

Artificial intelligence technologies, which are beginning to show up in almost every aspect of our lives today, have brought with them major controversies. In this context, the thing almost everyone fears most is “artificial general intelligence” (AGI). That is, a system close to or even more advanced than human intelligence. As giants like OpenAI, Google, and Anthropic rapidly advance in this field, debates like “what if it brings about the end of humanity?” or “what if it solves all our problems?” are on everyone’s mind.

However, Microsoft has emerged with a new vision that completely rejects the AGI race. Microsoft AI CEO Mustafa Suleyman announced the company’s new goal in an article he personally penned: Humanist Superintelligence (HSI). So, why is this new plan from Microsoft so different from the superintelligence goals we know? What is Humanist Superintelligence (HSI)? How is it different from AGI?

According to Microsoft, the real question the industry should ask, instead of the “AGI race,” is this: “What kind of artificial intelligence does the world want?”. The company’s answer to this question is HSI. This approach does not aim for a superintelligence that knows everything, can do every task, and can make decisions on its own. In fact, HSI is designed as a problem-focused system that specializes in specific areas and works to serve humans.

Suleyman describes this approach as a vision that will “keep humanity at the top of the food chain.” In other words, the goal is not to create an “entity” that can surpass humans, but to build a human-centric technology that enhances humanity’s collective intelligence and creativity. According to Suleyman, the order is: human-centricity first, then technological acceleration.

Contents show

So, why did Microsoft pull out of the AGI race?

One of the most striking parts of Microsoft’s announcement is the security concerns about AGI and superintelligence. Suleyman seems quite honest on this topic: “How will we control or align a system designed to be smarter than us?”. Microsoft’s AI boss says that neither AI developers nor security researchers currently have a “reassuring” answer to this question.

This is where HSI aims to remove this “uncontrollable risk” factor from the equation. Microsoft plans to minimize these alignment and security problems by narrowing the focus to more specific domains. For example, creating a “medical superintelligence” that can only make medical diagnoses, instead of an all-knowing system, is a much less risky and more manageable goal.

Where will humanist superintelligence be used?

Microsoft AI has identified three inspiring areas where HSI can improve our lives. The first goal is a perfect and inexpensive AI assistant for everyone. The purpose of this assistant is to alleviate people’s “daily mental load” and help them learn.

The second, and perhaps most ambitious, goal is the medical superintelligence mentioned earlier. Suleyman states that we will see expert-level performance in this field within the next few years. He even says that Microsoft’s system, MAI-DxO, achieved 85% success in diagnostic cases where even expert doctors barely exceeded a 20% success rate. The third goal is clean energy. Microsoft predicts that with the help of AI, cheap and abundant renewable energy will be achieved before 2040.

So, what do you think of this new vision from Microsoft? Is a human-focused superintelligence goal more logical than the AGI race? We look forward to your comments!