Here's how AI could kill us all (if the worst fears come true)

# Why This Matters This article illustrates a widening gap between public AI discourse and actual technological reality—one with genuine consequences for policy and investment. When researchers assign extinction probabilities spanning nearly 100 percentage points (10% to 99.99%) without scientific grounding, it signals that "existential risk from AI" functions more as a philosophical framework than an empirically testable claim. This matters because governments and corporations are making billion-dollar decisions based partly on these narratives: the EU's AI Act, for instance, was shaped by catastrophic risk arguments, yet dedicates minimal resources to the concrete harms AI systems cause today—discrimination, labor displacement, and misinformation. By elevating speculative doomsday scenarios to headline prominence, coverage can inadvertently legitimize unfalsifiable claims while crowding out scrutiny of demonstrable problems that regulation could actually address now. The real-world effect is a misdirection of limited regulatory attention toward hypothetical future scenarios while present-day inequities baked into deployed AI systems continue largely unchecked.