The machines are already improving themselves. You just have not been told how fast.

In the spring of 2026, Claude began writing between seventy and ninety percent of the code used to train its own next version. AlphaEvolve spent over a year optimizing the training process for the very models that powered it, recovering enough wasted compute across Google's global fleet to power a small country's worth of servers. OpenAI's Codex debugged the training pipeline that produced it. And a 630-line Python script showed that anyone with a single GPU could run autonomous AI research experiments overnight.

"AI Builds Itself" is the first comprehensive survey of recursive self-improvement as it exists right now. Written by H. Peter Alesso, an author with deep expertise in AI systems, web intelligence, and emerging technology, this book maps the full landscape: the organizations driving the science, the technical architectures that make self-improvement possible, the billion-dollar startups founded by the architects of AlphaGo and GPT-4, and the safety researchers warning that the feedback loops powering progress are the same loops that could outrun human oversight.

This book is for technology professionals, AI researchers, policymakers, investors, and informed general readers who want to understand the most consequential technological development of the decade, not through speculation or hype, but through documented evidence, cited research, and careful analysis of what the leading minds in the field are actually building.

You will learn how AlphaEvolve discovers algorithms that outperform fifty-six-year-old mathematical records. How the Darwin Godel Machine rewrites its own source code and evaluates whether the rewritten version is better at rewriting itself. How Karpathy's AutoResearch framework turns plain-language instructions into an overnight autonomous research laboratory. How Constitutional AI attempts to keep self-improving systems aligned with human values. And why the AI 2027 forecast, graded at sixty-five percent of its predicted pace, still projects transformative change within the next eighteen months.

One appendix with a working implementation you can run yourself. A comprehensive glossary and bibliography. No hype. No science fiction. Just the clearest account available of what is happening inside the world's most advanced AI laboratories and what it means for everyone outside them.
Read more

The machines are already improving themselves. You just have not been told how fast.

In the spring of 2026, Claude began writing between seventy and ninety percent of the code used to train its own next version. AlphaEvolve spent over a year optimizing the training process for the very models that powered it, recovering enough wasted compute across Google's global fleet to power a small country's worth of servers. OpenAI's Codex debugged the training pipeline that produced it. And a 630-line Python script showed that anyone with a single GPU could run autonomous AI research experiments overnight.

"AI Builds Itself" is the first comprehensive survey of recursive self-improvement as it exists right now. Written by H. Peter Alesso, an author with deep expertise in AI systems, web intelligence, and emerging technology, this book maps the full landscape: the organizations driving the science, the technical architectures that make self-improvement possible, the billion-dollar startups founded by the architects of AlphaGo and GPT-4, and the safety researchers warning that the feedback loops powering progress are the same loops that could outrun human oversight.

This book is for technology professionals, AI researchers, policymakers, investors, and informed general readers who want to understand the most consequential technological development of the decade, not through speculation or hype, but through documented evidence, cited research, and careful analysis of what the leading minds in the field are actually building.

You will learn how AlphaEvolve discovers algorithms that outperform fifty-six-year-old mathematical records. How the Darwin Godel Machine rewrites its own source code and evaluates whether the rewritten version is better at rewriting itself. How Karpathy's AutoResearch framework turns plain-language instructions into an overnight autonomous research laboratory. How Constitutional AI attempts to keep self-improving systems aligned with human values. And why the AI 2027 forecast, graded at sixty-five percent of its predicted pace, still projects transformative change within the next eighteen months.

One appendix with a working implementation you can run yourself. A comprehensive glossary and bibliography. No hype. No science fiction. Just the clearest account available of what is happening inside the world's most advanced AI laboratories and what it means for everyone outside them.
Read more