Microsoft Consumer Artificial Intelligence Director Mustafa Suleyman is currently working on creating a super-intelligent system that is "in the best interests of humanity." In a recent interview this week, he pledged to immediately halt research and development if the technology poses a threat to humanity.
Suleiman stated on the program, "We will not continue to develop systems that could potentially spiral out of control."
He emphasized that this idea should become a general consensus, "but frankly, I think it is still a novel position in the current industry."
Suleiman joined Microsoft early last year when Microsoft acquired the intellectual property of Inflection AI, the startup he founded. And the core team. Previously, Microsoft mainly relied on OpenAI for AI tools, but it has now appointed Suleiman to lead the development of products that can compete with the best in the industry.
In a blog post last month, Suleiman announced plans to advance the superintelligence initiative and elaborated on his and Microsoft's view that such systems can only succeed if they serve humanity.
Many AI companies, including OpenAI and Anthropic, have claimed to have similar goals, but Suleiman believes that Microsoft's goals are unique.
However, Suleiman did not name any competitors. "Each company needs to decide its own stance and operating methods. I have no intention of judging the current operating model. There is currently no evidence that AI is causing large-scale harm to humanity."
On "cutting ties" with OpenAI
It is worth mentioning that before an agreement was reached in October this year to reshape the relationship between Microsoft and OpenAI, Suleiman's work was actually limited by the contract terms —which prohibited Microsoft from developing general artificial intelligence (usually referring to systems with human capabilities) or superintelligence that surpasses human capabilities.
Under the new agreement, OpenAI can collaborate with third parties to develop some products. Meanwhile, Microsoft can also develop general artificial intelligence independently or in partnership with third parties.
Suleiman revealed that Microsoft has effectively waived the right to restrict these rights in exchange for the opportunity to use OpenAI's latest products—part of a previous collaboration between the two companies, during which Microsoft has been building and configuring data centers for OpenAI for many years. .
When discussing OpenAI, Suleiman stated, "They are now working with SoftBank and Oracle..." Agreements have been reached with multiple companies, and the scale of the data centers being built exceeds the number originally planned by Microsoft. Consequently, we have also gained the right to independently develop artificial intelligence.
Currently, discussions about superintelligence remain theoretical. While AI models like ChatGPT can interact with humans in ways that computers couldn't do a decade ago, the technology is still evolving. This is true for both consumers hoping AI assistants will help with everything from booking concert tickets to creating holiday shopping lists, and for corporate executives betting that such tools will boost employee productivity.
Suleiman stated that Microsoft Copilot's intelligent agent feature is "not always accurate" and is still under development.
“We are still experimenting,” he said, “but when it actually works, it will be the most amazing thing you’ve ever seen.”
(Article source: CLS)
