This is my note of book The coming wave, written by Mustafa Suleyman, newly appointed CEO of Microsoft AI.

Content

The key idea of the book can be summarized as:

Revolutionary technology always proliferates. Containment of AI seems impossible. However, uncontrolled AI may spell the end of human civilization. Thus, we must take actions to contain AI.

The first section of the book lays the foundation that revolutionary technology diffuses. It will eventually permeate society and bring signifant change. Examples include the steam engine, internet, smartphones, and now, AI.

The second section explains why AI is not containable.

First, because of its characteristics:

  • Asymmetric power: 100x productivity boost, redefining what is useful and powerful, and redistribution of power.
  • Evolves fast, permeating everywhere. Think Internet and social media.
  • Multi-purpose and general-use. Think electricity and smart phone.
  • Only partially understood by even scientists and can potentially put people “out of loop” (aka lose people control).

Second, because of the irresistible incentive:

  • National strategy, intensified due to geopolitical competition, especially between the US and China.
  • Huge profits, pursued by companies and individuals.
  • Individual (entrepreneurs and academies) drive to make a mark in this historic moment.
  • The open nature of academia and preference of open software will further accelerate rapid evolution.

The logic flows well to this point.

The third section of this book is less coherent, in my opinion. It begins with the author’s disappointing experiences in nonprofit and government jobs. Then, it delves into how nation-states are fragile and unstable due to the erosion of trust, polarization, pandemics, and war. This fragility will be further amplified by the forthcoming wave of AI, citing examples such as AI-powered ransomware, robots armed with guns, deepfakes, and AI Machiavellis.

The book ends with ten recommendations for the containment of AI. Regardless of what the recommendations are, up to this point, I’m convinced that the regulation of AI is necessary and vital.

Style

Overall, I quite enjoyed the book.

The prose style is fluid, concise, straight to the point, and with little pretense. The writing is engaging and conversational. The author shows dexterity in explaining things and concepts. He likes to use analogies with known events or objects1. It is also evident that the author is quite versatile in history.

I have noticed lots of people talking about the risk of AI, including Elon Musk and Geoffrey Hinton. But no one is as eloquent as Mustafa is in this book. Perhaps that is one of the reasons he can be CEO of Microsoft AI, despite lacking of “strong technical background”.

  1. For example, a car seat belt is a type of safety mechanism to reduce the consequences of a car accident. Hence,it makes sense to have a safety mechanism to reduce the potential harmful impact of AI. This seemingly native analogy makes concrete what safety means for AI and why it is necessary. ↩