Hacker News Reader

> risk of extinction from AI

That's a pretty strong statement. Extinction of humanity, no less. I don't get why so many experts (lots of them aren't crackpots) signed this.

View on HN

Given risk is probability * damage, and the damage is enormous, the risk can be high even if the probability is fairly low.

This fallacy is also known as Pascal's wager: https://en.wikipedia.org/wiki/Pascal%27s_wager

But this argument got billions of people to believe in the concept of hell so i expect it to work again to convince people to believe in AI doomsday.

I agree it's a fallacy when the probability is like 10^-10 but in this case I believe that the probability is more like 1%, in which case the argument is sound. I'm not trying to make a pascal's wager argument.

Correction: you perceive it to be a fallacy when others assign high probabilities to things you believe are low probability. Unfortunately, this cuts both ways. Many people believe your 1% estimate is unreasonably high. Are you therefore promoting a fallacy?

Too many ridiculous arguments can be justified on the backs of probability estimates pulled from nether regions.

Maybe seen some things we don't yet know exist?

Reply on news.ycombinator.com