Technology & Science
Co-Intelligence
Ethan Mollick
"Co-Intelligence" is a pragmatic and accessible guide to the AI revolution written by Ethan Mollick, a professor at the Wharton School. Unlike many theorists who speculate about the distant future, Mollick focuses entirely on the messy, practical present. He argues that AI is not merely a new tool like a calculator or a search engine, but a distinct form of "alien" intelligence that acts more like a talented but erratic intern. Mollick urges readers to stop treating AI as software that follows clear instructions and start treating it as a coworker. The book centers on the concept of the "Jagged Frontier." This theory explains that AI capabilities are uneven. It can pass the Bar Exam in the 90th percentile yet fail at simple arithmetic or basic logic puzzles. Because this frontier is invisible and constantly shifting, the only way to understand what the AI can do is to use it relentlessly for everything. The book outlines four principles for working with AI, emphasizing that humans must remain the "human in the loop." Mollick categorizes users into two types: Centaurs, who divide tasks clearly between human and machine, and Cyborgs, who integrate their workflow so deeply with AI that the line blurs. "Co-Intelligence" is a call to action to experiment, inviting us to invite AI to the table before we are left behind.
The Alignment Problem
Brian Christian
"The Alignment Problem" is a deeply researched and humanistic look at the most critical challenge in AI development. Written by bestselling author Brian Christian, the book serves as a history of the field of AI safety. It moves beyond the sci-fi fear of killer robots to address the real and present danger of systems that are competent but misaligned with human values. Christian defines the core issue as the gap between our complex human goals and the rigid mathematical models we use to train machines. He illustrates this with vivid examples from the real world, such as facial recognition software that fails to see dark skin or hiring algorithms that discriminate against women. The book argues that these are not just glitches but fundamental failures in how we define what we want machines to learn. The book explores how researchers are attempting to bridge this gap. They are moving away from giving machines rigid instructions and instead teaching them to be uncertain, humble, and capable of learning values by observing human behavior. "The Alignment Problem" suggests that in trying to teach machines to be good, we are being forced to confront the difficult task of defining what it actually means to be human.
How the World Really Works
Vaclav Smil
"How the World Really Works" is a data-driven reality check on the physical foundations of our modern civilization, written by Vaclav Smil. Smil, a renowned energy scientist and Bill Gates’ favorite author, aims to cure the "magical thinking" that dominates current debates about climate change and the green energy transition. Smil argues that before we can fix the world, we must understand the material realities that sustain it. He identifies the "Four Pillars" of modern life: Ammonia (for fertilizer/food), Steel, Concrete, and Plastics. He demonstrates that our dependence on the fossil fuels required to produce these materials is far deeper than most realize. The book is not a denial of climate change, but a denial of utopian timelines. Smil explains why total decarbonization by 2050 is physically impossible without causing societal collapse. By analyzing everything from globalization to risk assessment, How the World Really Works offers a scientifically grounded, agnostic guide to understanding the massive inertia of our energy systems, arguing that we must balance high ideals with hard numbers.
The Age of Surveillance Capitalism
Shoshana Zuboff
A groundbreaking exploration of the unprecedented power of surveillance capitalism and the quest for a human future. It unveils how tech companies are using our personal data to predict and manipulate our behavior, posing a profound threat to individual autonomy and democratic values.
Human Compatible
Stuart Russell
A thought-provoking exploration of the past, present, and potentially perilous future of Artificial Intelligence, examining the risks of creating machines more intelligent than humans and proposing a new approach to ensure AI remains beneficial to humanity.