08/31/2026
In April 2023, Tucker Carlson sat down with Elon Musk and asked him, on camera, how worried he actually was about artificial intelligence destroying civilization. That conversation is where our poll "AI and the end of civilization" started. We built it to test whether the fear in that interview was a fact people were reacting to, or a mirror of something else entirely.
Two years later, the conversation hasn't slowed down. Dr. Roman Yampolskiy (the researcher who coined the term "AI Safety") has become one of the most-watched voices on the topic, reaching 20 million viewers on The Diary Of A CEO alone, and, in March 2026, taking the stage at Stanford University. Whatever your position on AI risk, the audience for this question has never been bigger.
So we pulled our data. Up to 908 responses, up to 77 countries, two years of tracking the same statement Musk raised in that interview: "AI is dangerous to human civilization."
The answer isn't yes or no. It's seven different answers, and none of them are really about AI.
🔴 One value system says "not dangerous" because caution itself feels like a threat to be overcome.
🔵 Some value systems want the state to step in - not because they fear the technology itself, but because of what happens when nothing is accountable.
🟠 Others reject the danger narrative entirely - not because they dismiss risk, but because fear without a plan reads as something else.
💎 One value system won't even answer the question as asked, and lands on "not dangerous" too, for the opposite reason: not defiance, but transcendence of the question itself.
If a question this widely watched is landing so differently depending on who's hearing it, the real question isn't "is AI dangerous." It's: whose value system is doing the listening?
Full breakdown of all seven positions in SDTEST ® Insights #112.
Data from up to 908 responses across up to 77 countries reveals that the AI-and-civilization debate isn't really a debate about AI.
