Real or Rendered

In 1917, two young cousins in Yorkshire faked fairy photographs, and Sir Arthur Conan Doyle staked his reputation on their authenticity. The camera had not lied; nobody had questioned what it could not see. A century later, readers face the same blind spot, guessing whether a photo, an article, or a recording is genuine.
Most advice for spotting synthetic media focuses on details that vanish with the next update: six-fingered hands, mismatched earrings, a background that swirls into nonsense. Real or Rendered argues that chasing these glitches is a losing strategy, since the models producing them improve faster than anyone can catalog the mistakes.
The book follows a forensics researchers shift from counting visual errors to the mathematics underneath an image, and a 2017 Stanford study where fact-checkers outperformed PhD historians simply by leaving a website to search for it elsewhere. A teachers staged photo of a nonexistent "Patagonian Rock Hound" fools a classroom in under a minute, and OpenAIs own AI-text detector was right barely a quarter of the time before being shut down.
Discover why the real defense against synthetic media was never a sharper eye, but a habit of reading sideways instead of staring harder.