AI Will Kill Us All: Why Artificial Intelligence Scares and What the Real Risks Actually Are

Artificial intelligence is going to take your job, destroy democracy, manipulate your children, design bioweapons, escape from its servers, outsmart every government on Earth, and eventually convert the planet into something computationally efficient. At least, that is one version of the story. AI has become the subject of extraordinary claims. Some are based on genuine technical progress. Others rest on assumptions about intelligence, power, autonomy, economics, and human behavior that become considerably less obvious once examined closely. AI Will Kill Us All looks at how the modern AI catastrophe narrative was built, why it is so persuasive, and what happens when dramatic thought experiments collide with the inconvenient details of the physical world. Along the way, the book examines superintelligence, intelligence explosions, fast takeoff, alignment, hallucinations, benchmarks, automation, surveillance, deepfakes, cybercrime, autonomous weapons, economic concentration, algorithmic decision-making, and the increasingly popular practice of assigning precise numerical probabilities to events that have never happened. It also asks a simpler question that tends to receive less attention: what can today's AI systems actually do, and what still requires money, infrastructure, institutions, energy, factories, supply chains, legal authority, human cooperation, and access to something other than a chat window?