LESSWRONG
LW

2876
Wikitags
Main
Arbital

AI Arms Race (Arbital import)

Edited by alexei last updated 31st Jan 2017

To understand this topic better, see AI arms races. This page will describe how this concept is treated in The plan .

I think the arms race can of two types. In either case it's driven by fear, but in the first case it's imaginary, whereas in the second case it's real.

Imagine a scenario where you know that an organization is developing an AGI at full speed. For now they have no safety plans, so you assume that when they reach AGI, they'll just proceed ahead. This will likely encourage you to develop your own AGI faster in turn.

If the other company will in fact do what you expect, then your fear is real. However, if the other company has a plan to stop at some milestone before full AGI to evaluate its safety, then your fear is mostly imaginary.

This makes communication really important. If companies make their plans public, then it will make it easier for people to distinguish between real and imaginary arms race scenarios.

Subscribe
Discussion
Subscribe
Discussion
Posts tagged AI arms race
66Saying Goodbye
sapphire
2mo
74
229The Hopium Wars: the AGI Entente Delusion
Max Tegmark
1y
60
170Safety researchers should take a public stance
Mateusz Bagiński, Ishual
1d
13
52Was Releasing Claude-3 Net-Negative?
Logan Riggs
1y
5
23Counter-considerations on AI arms races
Mateusz Bagiński, JustinShovelain
4mo
0
22Anthropic AI made the right call
bhauth
1y
20
14Why is AGI/ASI Inevitable?
DeathlessAmaranth
1y
6
318Would catching your AIs trying to escape convince AI developers to slow down or undeploy?
Ω
Buck
1y
Ω
77
47NYT: Google will “recalibrate” the risk of releasing AI due to competition with OpenAI
Michael Huang
3y
2
17The Slippery Slope from DALLE-2 to Deepfake Anarchy
Ω
scasper
3y
Ω
9
Add Posts