As a rule humans are catastrophically bad at estimating and ranking risk.
People are afraid of strangers abducting their kids, of air travel, of nuclear power. They aren't generally afraid of cheeseburgers, raising kids who don't exercise, cars, and coal power, but they should be.
They also aren't afraid of the small but catastrophic risk of global warming truly messing our climate up, but they should be.
The inability to properly estimate risks goes both ways, people are bad at cumulative every day stuff as well as tiny probability catastrophic events (especially involving things they never see).
The relative dangers of plane travel, automobile travel, or even leading a sedentary lifestyle and eating unhealthy foods are not high enough to actually cause an existential risk for the continuation of human civilization.
People are bad about judging relative risk on lower risk activities, such as whether a 1 in 10,000 chance of death is safer or less safe than a 1 in a million chance of death. But when it comes to simple go/no-go decisions on much riskier activities humans are a bit better at making a sound judgment. Is it a good idea to start a fight with a bear? To engage in a gun fight? To set your house on fire and stay inside it? These sorts of risks people understand and can handle.
Generally speaking, if everyone decided tomorrow that drunk driving, smoking, and obesity were awesome humanity would still endure. Though people wouldn't live quite as long.
Okay, so some part of our risk estimation ability comes from risk-evaluation-specific heuristics that have developed in ancestral environment. But how much impact does this have on the result of any serious evaluation? How easily do the ancestral parts get overridden by rational correction or outright calculation?
People are afraid of strangers abducting their kids, of air travel, of nuclear power. They aren't generally afraid of cheeseburgers, raising kids who don't exercise, cars, and coal power, but they should be.