Coming soon · Book Profile
The Alignment Problem
A journey through the burgeoning field of AI alignment, exploring how we can ensure machine learning systems understand and act according to human values, from present-day ethical dilemmas to future existential risks.
A profile of this book is on the way.
What it’s about
As machine learning systems become more powerful and pervasive, controlling everything from bail decisions and hiring processes to autonomous vehicles and potentially superintelligent AI, a critical new question emerges: how do we ensure these systems do what we want? Brian Christian's *The Alignment Problem* is a comprehensive exploration of this central challenge of the 21st century. Through stories of biased algorithms, misaligned game-playing agents, and the cutting-edge research attempting to solve these problems, the book reveals the technical, ethical, and philosophical complexities of teaching machines our values. It's a journey into the heart of AI safety, showing how our attempts to align machines with human intentions serve as a revelatory mirror, forcing us to understand our own values with unprecedented clarity.
The through-line
- Who it’s for
- The reader is a thoughtful individual—a technologist, policymaker, business leader, or concerned citizen—who sees the growing power of AI and machine learning and wants to ensure this technology is developed and deployed safely and ethically for the benefit of humanity.
- The problem
- Machine learning systems are being deployed at scale with flawed objectives and biased data, leading to unfair outcomes, unintended consequences, and the risk of catastrophic failures as systems become more powerful. They feel a mix of excitement and anxiety about the future of AI. They are worried about being complicit in building or using harmful systems, and feel overwhelmed by the complexity of aligning powerful technology with subtle human values.
- The plan
- Understand the roots of the alignment problem through present-day challenges in supervised learning (Representation, Fairness, Transparency).
- Grasp the dynamics of autonomous agents through the lens of reinforcement learning (Reinforcement, Shaping, Curiosity).
- Explore the frontier of AI safety research and the most promising techniques for aligning advanced systems with human norms (Imitation, Inference, Uncertainty).
- The payoff
- The reader becomes an informed and effective advocate for AI safety and ethics in their field. · They can confidently build, deploy, or regulate machine learning systems with an eye toward human values. · They contribute to a future where powerful AI is aligned with humanity's best interests, avoiding dystopian outcomes and unlocking immense potential.
See our guide
Additional reading
- Perceptrons · Marvin Minsky and Seymour Papert
This 1969 book's mathematical proof of the limitations of early neural networks is credited with launching the first 'AI winter,' making it a crucial historical text that the modern deep learning movement had to overcome.
- The Organization of Behavior · Donald Hebb
Published in 1949, this book introduced the theory of Hebbian learning ('neurons that fire together, wire together'), which provided a core biological inspiration for Geoff Hinton and the entire connectionist approach to AI.
- On Intelligence · Jeff Hawkins
This book's thesis that the brain's neocortex operates on a single master algorithm directly inspired Andrew Ng and shaped his successful pitch to Larry Page to create the Google Brain lab.
- Superintelligence: Paths, Dangers, Strategies · Nick Bostrom
This philosophical book, heavily promoted by Elon Musk, framed the debate around the potential existential risks of AGI and became a foundational text for the AI safety movement and organizations like OpenAI.
- Gödel, Escher, Bach: An Eternal Golden Braid · Douglas Hofstadter
It exposed the author to the idea that the mind could be understood in discrete, mathematical terms and introduced her to the philosophical implications of computation.
- The Emperor's New Mind · Roger Penrose
Along with Hofstadter's book, it challenged the author with its rich connections between different fields and its rigorous, scientific approach to understanding intelligence and the mind.
- What Is Life? · Erwin Schrödinger
This book by a famous physicist turning his attention to biology sparked the author's shift from physics toward the life sciences and the mystery of the mind.
- WordNet · George Armitage Miller (and team)
This lexical database project provided the author with the conceptual map and ontology that became the structural foundation for ImageNet, revealing a path to organizing the visual world at a massive scale.
- Superintelligence · Nick Bostrom
The book's exploration of AI's future became a mainstream success and a topic of discussion in the author's 'AI Salon,' highlighting the growing societal and philosophical questions surrounding the field.
- Clinical Versus Statistical Prediction · Paul Meehl
A foundational 1954 book that demonstrated through numerous studies that simple statistical formulas consistently outperform the intuitive judgments of human experts, providing an early rationale for algorithmic decision-making.