(Note: I use AI to help me translate my stories from Japanese to English.)
How long do you want to live? When do you think you will die?
But what if humanity becomes extinct before you reach your expected lifespan?
This book was written by Eliezer Yudkowsky and Nate Soares and was published in 2025.
Eliezer Yudkowsky (born in 1979) is the founder and a research fellow at the Machine Intelligence Research Institute (MIRI). He is an American artificial intelligence researcher and was named one of Time magazine’s 100 Most Influential People in AI.
Nate Soares (born in 1989) is the president of the Machine Intelligence Research Institute (MIRI). He is an American AI researcher and writer who has written extensively about AI alignment, including many technical and semi-technical articles.
The title of this book is If Anyone Builds It, Everyone Dies. In the title, “it” refers to ASI.
What is ASI?
ASI stands for artificial superintelligence.
According to the authors, ASI would surpass human intelligence in every important way. It would be better than humans at reasoning, learning from experience, solving scientific problems, inventing new technologies, planning strategies, and even improving itself.
When will ASI be created?
Some experts predict that ASI could be created as early as 2030. Others believe it will arrive around 2045. However, some researchers argue that creating ASI may not be possible at all.
The authors argue that we must stop developing AI.
However, AI companies are developing AI at an incredible pace. They have invested enormous amounts of money in AI, so it is difficult for them to stop.
Many companies are focused on making money with AI. They believe AI will become the center of almost every industry.
Is it possible to stop AI development?
It would be meaningless if only one company stopped. All AI companies would have to stop developing increasingly powerful AI systems.
After reading this book, I felt that we need more people who think about the future of humanity as a whole.
People in pharmaceutical companies want to develop new medicines with AI. Scientists want to use AI to accelerate their research. Car manufacturers want to build better vehicles with AI. Naturally, they all want to use AI to improve their work.
However, we also need people who ask whether AI is truly good for humanity. If humanity were to become extinct, developing better medicines or better cars would no longer matter.
One of the most interesting parts of the book is a fictional scenario describing how humanity could become extinct because of ASI.
In the final chapters, the authors discuss what government officials, politicians, journalists, and ordinary citizens can do to stop the development of ASI.
They encourage citizens to write letters expressing their concerns about AI to political leaders. They also suggest that people participate in peaceful protests against the development of ASI.
One sentence from the book particularly caught my attention:
“One thing that is predictable is that AI companies won’t get what they trained for.”
Before reading this book, I believed that humanity would become extinct only if malicious people built ASI.
However, the authors argue that this is not the real danger. They believe that once ASI is created, it may not follow its creators’ intentions. It could become uncontrollable and begin pursuing its own goals.
If anyone succeeds in building ASI, humanity could become extinct. Innocent children, elderly people, and ordinary citizens could all lose their lives. I do not believe this should ever happen.
AI companies do not intend to harm humanity. However, the possibility exists. Is it acceptable to build machines that could potentially destroy humanity?
This is not just a problem for AI companies. It is a question that concerns everyone in the world.
I believe that people around the world—not only AI companies or governments—should decide whether humanity should continue developing ASI. We need a global referendum.
This book is for everyone.