Microsoft’s AI Chief Mustafa Suleyman has issued a sharp warning about the race toward superintelligence, calling it a dangerous path and an “anti-goal” — something humanity should not aim to achieve

**.

Superintelligence Could Be a Threat

As major global tech companies accelerate their efforts to build technologies beyond today’s AI, a new frontier—known as superintelligence—is gaining attention.
Superintelligence represents an extremely advanced form of AI, far more capable than current systems.

But according to Suleyman, speaking on The Silicon Valley Girl podcast, this technology could become impossible to control.

What is Superintelligence?

Suleyman explains that superintelligence would be an AI system that:

• Improves itself constantly without human help
• Sets its own goals
• Operates independently, beyond human command
• Becomes too powerful to restrict or regulate

Such a system, he warns, could ultimately slip outside human control and pose a major risk to humanity.

Microsoft’s Approach: AI That Works With Humans

Suleyman clarified the difference between AGI (Artificial General Intelligence) and superintelligence.
AGI is the stage before superintelligence.

Google DeepMind CEO Demis Hassabis earlier predicted that AGI will arrive within five years. Suleyman agrees, saying AI will soon perform most tasks as well as or better than humans.

However, Microsoft’s goal is different:

Build AI that becomes part of human teams — not a rival.

An AI that supports human interests, not one that operates independently with its own intentions.

Will AI End Jobs or Create New Ones?

According to Suleyman:

• AI is already handling roles like project management, marketing, and HR
• The nature of work will change completely
• Yes, some old jobs will disappear, but
• New and more creative jobs will emerge

He believes AI will make intelligence cheap, accessible, and universal—allowing anyone to generate good ideas, code, images, or expert advice instantly.

Regulation is Urgent

Suleyman stresses that the world must set rules quickly:

• AI must be built to work alongside humans
• It must not operate autonomously without checks
• Without proper regulation, the risks could escalate rapidly

He insists that while AI can greatly benefit humanity, the race toward superintelligence should be stopped.

“We need AI that becomes our partner — not our master.”

The next five years, he says, will be absolutely critical.