More Than Ten Percent
An Anthropic alignment lead put a number on it: greater than ten percent that AI ends humanity within a decade. A colleague had just quit, saying the industry is gambling with everyone's life.
AI Ethics
Accountability, bias, manipulation, and the human consequences of systems designed without asking who gets harmed. Not philosophy but consequence.
An Anthropic alignment lead put a number on it: greater than ten percent that AI ends humanity within a decade. A colleague had just quit, saying the industry is gambling with everyone's life.
AI insurance models show 17% pricing disparities correlated with racial composition of zip codes. No human decided this. The algorithm inherited decades of redlining from historical training data. Nobody knows who is legally responsible.
Since 2016, Pokemon Go players have mapped the physical world for Niantic without knowing it. Thirty billion images later, those scans are training Coco Robotics delivery robots. The game was always the mechanism.
AI companions promise to solve loneliness. They may be better at creating it.