P(doom) Ain't the Question
AI may someday reshape civilization. Today, many of our smartest technologies still can't reliably perform the simplest tasks. That gap has profound implications for military AI.
The AI community has spent the past several years debating P(doom)—the probability that artificial intelligence eventually leads to humanity’s extinction.
It’s an important conversation. If AI one day exceeds human intelligence across every domain, society should think carefully about the risks.
But those conversations often take an unexpected turn.
Someone inevitably brings up the hydraulic liftgate on their SUV that refuses to open half the time, forcing them to wrestle against its own hydraulics.
Someone else complains about AirPods that drain one earbud to charge the other, even when the charging case has plenty of battery remaining.
Another remembers the smart TV that froze during the biggest game of the year.
And, of course, there is always the Blue Screen of Death.
These aren’t catastrophic failures. And none of them are evidence that AI won’t be transformative. They simply remind us that even mature technologies still behave in ways that are unpredictable, frustrating, and occasion…
