Difference between revisions of "Importance of knowing about AI takeoff"
| Line 3: | Line 3: | ||
* Weight on [[AI prepping]]: In a FOOM scenario it is unlikely that there are any selfish actions that benefit individuals, so one should shift those resources into either altruistic actions (like helping with alignment) or short-term consumption. In contrast, with more continuous takeoff, AI prepping becomes relatively more important. | * Weight on [[AI prepping]]: In a FOOM scenario it is unlikely that there are any selfish actions that benefit individuals, so one should shift those resources into either altruistic actions (like helping with alignment) or short-term consumption. In contrast, with more continuous takeoff, AI prepping becomes relatively more important. | ||
* Working to mitigate specific threats: e.g. if one expects AI-powered propaganda to be a thing during AI takeoff, then it makes sense to spend time thinking about it now. | * Working to mitigate specific threats: e.g. if one expects AI-powered propaganda to be a thing during AI takeoff, then it makes sense to spend time thinking about it now. | ||
| − | * Under some scenarios we should expect to see [[warning shots]]/early misalignment, whereas in others we should expect a [[treacherous turn]]. This has consequences for how careful we should be about alignment/how much of an "on paper understanding" we should have about alignment. | + | * Under some scenarios we should expect to see [[warning shots]]/early misalignment, whereas in others we should expect a [[treacherous turn]]. This has consequences for how careful we should be about alignment/how much of an "on-paper understanding" we should have about alignment. |
==See also== | ==See also== | ||
Revision as of 01:42, 5 March 2021
Importance of knowing about AI takeoff is about the "so what?" of knowing which AI takeoff scenario will happen. How will our actions change if we expect a FOOM or a continuous takeoff or some other scenario?
- Weight on AI prepping: In a FOOM scenario it is unlikely that there are any selfish actions that benefit individuals, so one should shift those resources into either altruistic actions (like helping with alignment) or short-term consumption. In contrast, with more continuous takeoff, AI prepping becomes relatively more important.
- Working to mitigate specific threats: e.g. if one expects AI-powered propaganda to be a thing during AI takeoff, then it makes sense to spend time thinking about it now.
- Under some scenarios we should expect to see warning shots/early misalignment, whereas in others we should expect a treacherous turn. This has consequences for how careful we should be about alignment/how much of an "on-paper understanding" we should have about alignment.
