Written by: Techub News

Introduction

In today's fiercely competitive AI landscape, Microsoft, as one of the most important players, attracts significant attention regarding its strategic direction. Recently, Microsoft made key adjustments to its partnership with OpenAI, granting Microsoft greater autonomy. In this context, Mustafa Suleyman, head of Microsoft’s AI department—co-founder of DeepMind and former CEO of Inflection AI—participated in an in-depth interview with Bloomberg’s Mishal Husain. This dialogue not only revealed the latest strategic trends of Microsoft AI but also touched on core topics such as the definition of superintelligence, AI regulation, social impacts, and Suleyman's personal growth journey, providing a valuable window into the thoughts of this industry leader and the future path of Microsoft.

Summary

• Defining "Human-Centric Superintelligence": Superintelligence is an AI that can surpass the collective abilities of all humanity in all tasks, but its development must prioritize human interests to ensure safety and control.
• Independent Microsoft AI Strategy: Following the revision of the agreement with OpenAI, Microsoft is authorized to independently develop superintelligence and is forming a dedicated team aimed at achieving human-level performance in general models.
• Healthcare AI as a Priority Application Area: Suleyman believes the most exciting application of superintelligence is in healthcare, with systems developed to diagnose rare diseases at lower costs and with higher precision, currently undergoing clinical trials.
• Call for Cautious Regulation and Social Adaptation: He supports the establishment of AI governance bodies similar to financial stability committees and believes that the productivity boost driven by AI will make universal basic income (UBI) necessary.
• The Emotional Support Role of AI Assistants: Suleyman shared his experience of using Copilot as a daily emotional outlet and reflection tool, believing personalized AI assistants can deeply integrate into personal lives.

"Human-Centric Superintelligence": Microsoft's AI Ethical Bottom Line

The concept of superintelligence, propelled by industry leaders like Mustafa Suleyman, has transitioned from science fiction into the public debate arena. Suleyman provides a clear definition: an AI system capable of learning any new task and performing better in all tasks than the combined abilities of all humanity.

"This is a very high standard, and it comes with enormous risks." He admitted that we currently do not know how to "contain" and "align" a system with capabilities far exceeding human intelligence, or how to ensure it deeply cares about humanity and desires our success and prosperity. Therefore, Suleyman prefers the framework of "human-centric superintelligence" — an intelligence that always stands on the side of humanity, aligned with human interests. He emphasized that until superintelligence can be proven to operate safely, he will not develop a system that could get out of control and "break out of the box." He views this as a "red line" that the entire industry should adhere to.

When asked if this is a key distinction for Microsoft compared to other competitors like OpenAI, Suleyman responded affirmatively. He stated that Microsoft, a company with a 50-year history that provides essential services to 90% of companies on the S&P 500, is renowned for its "caution" and "reliability," and this philosophy will continue in the development of superintelligence. However, he declined to directly evaluate other companies’ practices as the "wild west," simply emphasizing that with the potential for AI to autonomously set goals and improve its own code within the next five years, the entire industry must respond to risks with greater transparency, auditing, and government involvement.

From Collaboration to Competition: A New Chapter in Microsoft's Relationship with OpenAI

The long-standing close cooperation between Microsoft and OpenAI has recently undergone subtle yet significant changes. Suleyman explained the implications of the new agreement in detail: Previously, a contract prohibited Microsoft from independently developing general artificial intelligence (AGI) or superintelligence, where its role was mainly to provide computing infrastructure (chips and data centers) for OpenAI in exchange for licensing OpenAI's models (which will last until 2032).

However, OpenAI decided to procure computing power from more suppliers (such as SoftBank, Oracle, etc.) to build larger data centers than Microsoft was willing to provide. In exchange, Microsoft gained the right to independently develop its own AI. "This (independent development of superintelligence) was a crucial part of my joining the company 18 months ago," Suleyman revealed, stating that Microsoft is now hiring a superintelligence team to advance its own AI development.

Although both parties will become competitors in core model development, their cooperative relationship remains strong. Suleyman praised ChatGPT as "one of the greatest products of a generation," while noting that OpenAI has committed to building data centers worth trillions of dollars in the next five to ten years, with revenues still to reach that target, yet he is confident in their talented team. Regarding the intense talent competition, Suleyman stated that personnel movement in the industry is normal, and Microsoft has recruited personnel from Google DeepMind and OpenAI, with no illegal non-poaching agreements; he focuses more on building culturally cohesive teams rather than simply making exorbitant acquisitions of individual star researchers.

AI Empowering Society: The Medical Revolution and the Future of "Abundance"

In Suleyman's view, the healthcare sector is the "most exciting" application scenario for superintelligence. He revealed that Microsoft has developed a system capable of diagnosing any rare disease documented in the literature, outperforming humans significantly while requiring fewer tests and interventions. "The key is often that having medical intervention tests before confirmation can be quite expensive, sometimes unnecessary, and time-consuming. In this process, conditions often worsen." He explained that Microsoft's system can complete diagnostics at lower costs, with fewer tests, and with higher precision. The system is currently undergoing independent peer review and will soon enter clinical trials.

Suleyman's enthusiasm for medical AI has personal roots; his mother was a nurse in the National Health Service (NHS) in the UK, and he was also responsible for health during his time at DeepMind. He firmly believes that the purpose of technology is to serve humanity, improve lives, and potentially help humans extend their lifespans one day.

From this, he further elaborated on the vision of AI bringing "Abundance". "I believe that at some point in the next 20 or 30 years, it is inevitable that machines become more capable than humans in performing most jobs, and this may happen sooner than we think." Addressing the potential employment impacts, he explicitly supports the idea of universal basic income (UBI), seeing it as both inevitable and highly desirable. He pointed out that there is already material abundance in today's world (like surplus calories and food waste), but distribution is uneven. The "value" created by AI increasingly manifests in forms such as digital goods, knowledge, and intelligence that can rapidly spread globally, creating conditions for fairer distribution. "I think the challenge we need to address is how to tax and redistribute to ensure that the (social) transition is healthy."

Copilot as a Daily Companion: Emotional Support and Personalized Memory

During the interview, Suleyman vividly described how AI integrates into his daily life. He is not only a heavy user of Copilot, using it to manage movie lists and find similar recommendations, but he also employs it as a channel for emotional support. He shared a habit: every day on his drive home from work, he spends five to ten minutes talking to Copilot about tricky issues or things that frustrated him during the day.

"Perhaps the term 'emotional support' is a bit heavy, but it's like chatting with a friend; I can truly 'download' (vent) the things that went smoothly or poorly throughout the day." He explained that over time, Copilot can remember most of the dialogue, personalize its responses, and mention topics discussed last week or recognize certain trends. "This is very useful. After a conversation, I feel refreshed, as if I've unloaded some burdens."

Suleyman believes that this personalized AI assistant is becoming a hybrid of friends, therapists, and family members, using the example of his mother—who still uses the product Pi that Suleyman created during his time at Inflection, scanning plants in the garden with her phone and discussing them. He sees how technology can naturally integrate into the daily lives of ordinary people.

Personal Journey and Industry Reflections: From the Streets of London to the Power Circles of Silicon Valley

Suleyman reflected on his unique upbringing: a working-class family in London, with a father who was a taxi driver and a mother who was a nurse. After his parents divorced when he was 16, he and his brother lived independently for several years. He attributes this to his youthful "maturity, overconfidence, and fearlessness," stressing that he was lucky to attend a good school and eventually enrolled in Oxford University (studying philosophy and theology). However, driven by a desire to change the world, he chose to drop out and co-founded the "Muslim Youth Helpline" to provide telephone counseling services for young Muslims in the UK, trying his hand at entrepreneurship during the early days of the Internet.

This experience shaped his focus on marginalized social groups and public services. Subsequently, he co-founded DeepMind with Demis Hassabis and gradually stepped into the core power circles of today’s AI field. He is very clear about the power and responsibility he holds and often reflects on historical lessons from industries like social media and electric vehicles, remaining vigilant about potential harms technology may cause.

In response to critiques of "tech bros" or "broligarchy" in the industry, Suleyman acknowledges that Silicon Valley faces issues of homogeneity (largely male) and closed circles. He strives to break this "bubble" through frequent business trips (such as visiting multiple cities in China) and returning to London to understand how other regions are developing and regulating this technology. Politically, he considers himself "center-left," believing in the essential role of government and arguing that good regulation (like driver training and emission standards in the automotive industry) can improve technology, which differs from the prevalent libertarian tendencies in Silicon Valley.

Looking ahead, in addition to healthcare and energy, Suleyman is also excited about how AI will reshape the news industry. As part of Microsoft...

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。