What are the best examples of significant harm caused by "infohazards"? (don't say Roko)
I tend to internally roll my eyes when I hear this concern, often from EA-land, and I'd like to inoculate myself against that if appropriate.
Conversation
I think it depends a lot what “information” and “hazard” mean. A few logical thought experiments can shatter the idea that individuality is real, which can destroy some people (eg Heidegger). Molecules are information too, and they can do the same. People who do high-dose..
3
7
Replying to
Thinking in terms of Bostrom's def'n—i.e. true information which harms the listener. Those both qualify, though I notice my eyes do still roll a bit. I guess it's because the propensity for hazard remains quite low (assuming true information about molecular risk is offered).
Replying to
maybe you mean more like nuclear physics equations? Once we figured out the bomb was buildable it was only a matter of time. There’s similar things in AI imo, but I won’t point out publicly what they are
7
There's two different concepts here - knowledge which, being spread, harms the world (nuclear weapons), and knowledge which harms the listener. The latter is much more rare, especially if you exclude "misleading but true" statements.
3
Maybe hazard doesn’t necessarily result in harm. Hazardous information might demand strenuous coping and readjustment, stressing your cognitive immune system—possibly making you stronger…
1
5
Consider information about how to use darknets. Almost every parent would classify this as hazardous information. If it’s real, truthful disclosure might be better than suppression. But it’s a different type of challenge which most parents aren’t equipped for.
1
2
Show replies