People don’t calculate risk correctly because we use the wrong baselines. Am I likely to accidentally kill someone while driving my car¹? No.

But: the right baseline to use isn’t ‘How likely am I to accidentally kill someone with my car?’, it’s ‘How likely am I to accidentally kill someone by doing [activity x]?’.

When ‘activity x’ is ‘driving’, it’s obvious that I’m enormously more likely to kill someone with my car than I am doing almost anything else.

¹ That is, my hypothetical car; I don’t drive.