“AI winter now” means the squo does not upend nuclear stability
Geist & Lohn 18, Edward Geist: Policy researcher at the RAND Corporation. Formerly a MacArthur Nuclear Security Fellow at Stanford's Center for International Security and Cooperation and a Stanton Nuclear Security Fellow at RAND, Geist received his Ph.D. in history from the University of North Carolina in 2013. Andrew J. Lohn: Senior Fellow at Georgetown's Center for Security and Emerging Technology (CSET), where he works on the CyberAI Project (How Might Artificial Intelligence Affect the Risk of Nuclear War? RAND Corporation, Available Online at: https://www.rand.org/pubs/perspectives/PE296.html)
The two extreme cases have only limited relevance for the future of nuclear warfare. Stalling development (also referred to as AI winter) would result in only minor changes from the current nuclear-security environment. With superintelligence, AI would render the world unrecognizable and either save or destroy humanity in the process. The other two cases, in which AI progresses substantially and enables many new capabilities while still remaining fallible and inferior to humans in at least some respects, seem to have more support from the expert community, although experts disagree about the national security implications of such capabilities. Some fall in the category of “Complacents”: These tend to believe that producing an AI capable of performing the types of tasks that would destabilize the nuclear balance is sufficiently difficult that it is unlikely to be achieved. “Alarmists” hold the opposite view, that an AI could be capable of certain tasks but should not be included in any aspect of nuclear war. A third group, “Subversionists,” focus on an adversary’s ability to alter, mislead, divert, or otherwise trick the AI, which could prove either stabilizing or destabilizing.