(maybe I was a little bit off, it's more likely that the risk of failure is 1/1000, so sum is 1/1.000.000)
but the point is, it's more risky to drive back home after the weekend than to recreational cave dive, especially with dive fatigue!
In particular, with something like cave diving and mountain climbing the "bad" outcome is going to be death of the individual or their friends and it will likely be the result of a series of bad decisions under stress (each of which makes the situation more dire) rather than an exceedingly rare multiple systems failure.
I think that's the same issue with risk assessment and mixing up probability and impact: A problem when driving home is not a 100% certain death sentence, and it is not something you could not mitigate. You can drive slower, try to avoid dangerous streets etc. And even in the event of an accident you have systems that might save you (airbag etc).