Free beta access. 50 free AI tokens each week. Optional top-ups available. See what is included
Opening Word document…

AT Nuclear Instability

While countries will adopt autonomous systems, most countries will avoid using Destabilizing AI. Nations are Cautious about Nuclear Systems, and will only utilize autonomous AI if they enhance stability.

Horowitz, 2019 – prof of Political Science at Univ of Pennsylvania [Michael with Paul Scharre, and Alexander Velez-Green “A Stable Nuclear Future? The Impact of Autonomous Systems and Artificial Intelligence” December arXiv:1912.05291 https://arxiv.org/abs/1912.05291 Acc 12/27/20 TA]

What might lead to variation in how countries make choices about the relative utility of autonomous systems? The answer could depend on how secure they feel about their non-autonomous nuclear systems. States that feel extremely secure in their second strike capabilities at present may see fewer advantages to automation. In that case, the advantages greater speed and precision but might not appear worth the potential risk of accidents. Instead, states like the United States would likely prefer to use existing non-autonomous systems for nuclear command and control and delivery. In contrast, countries whose nuclear arsenals are more insecure may be more accepting of risk and may find the perceived advantages of automation more valuable. If a country thinks that its nuclear command and control might be at risk of severe degradation or destruction, it might be more likely to automate early warning to increase its response speed, deploy autonomous nuclear delivery platforms with higher endurance, automate new aspects of target selection for nuclear delivery vehicles, or shift towards more automated nuclear launch postures. All may not happen in unison, of course, but as a general relationship, countries whose arsenals are more insecure may be more willing to take risks to better enhance their arsenal’s survivability.

AI enhances nuclear deterrence because they provide the Speed necessary to handle the information overload in a crisis, and the Resilience to enhance deterrence.

Horowitz, 2019 – prof of Political Science at Univ of Pennsylvania [Michael with Paul Scharre, and Alexander Velez-Green “A Stable Nuclear Future? The Impact of Autonomous Systems and Artificial Intelligence” December arXiv:1912.05291 https://arxiv.org/abs/1912.05291 Acc 12/27/20 TA]

A critical question is thus how militaries will employ advances in AI to influence their early warning and NC2 systems. There may be many places where militaries could employ new forms of autonomous systems to bolster the reliability and effectiveness of early warning and NC2. Human-machine teaming could help offset automation bias and thus enable the use of more autonomous systems. More advanced automation in nuclear early warning systems could allow greater situational awareness, reducing the risk of false alarms. It could also play a valuable role in helping human decision-makers process large amounts of information quickly. In this regard, automated data processing may play a critical role in helping human nuclear early warning operators to identify threats – and false cues – in an increasingly data-saturated and complex strategic environment. Increased automation in NC2 could also help to reduce the risk of accidents or unauthorized use. And an expanded role for automation in communications could help ensure that command-and-control signals reach their targets quickly and uncorrupted in highly contested electromagnetic environments. Automation could also be used to enhance defenses – physical or cyber – against attacks on nuclear early warning, command-and-control, delivery, and support systems, thereby enhancing deterrence and fortifying stability. It could also be used to bolster the resilience of vulnerable NC2 networks. For instance, long-endurance uninhabited aircraft that act as pseudo-satellites (“pseudo-lites”) to create an airborne communications network could increase NC2 resilience by providing additional redundant communications pathways in the event of satellite disruption. Automation could even enable autonomously self-healing networks – in physical or cyberspace – in response to jamming or kinetic attacks against command-and-control nodes, thereby sustaining situational awareness and command and control and enhancing deterrence.

Autonomous AI systems are essential to Counter cyberwarfare – only Defensive LAWs can react to hacking and accidents.

Messinger 2015 - Research Associate at the Center on Law and Security [Eric January 15, Just Security “Is it Possible to Ban Autonomous Weapons in Cyberwar?” https://www.justsecurity.org/19119/ban-autonomous-weapons-cyberwar/ Acc 12/27/20 TA]

Of course, those familiar with the debate over AWS in kinetic warfare have already heard arguments about potential upsides for efficacy. Yet the nature of the cyber battleground, and especially cyber defense, will provide strong incentives to employ autonomous offensive cyber systems. The cyber theater consists in whole or in part of computerized systems, where the speed of movement is not constrained by the physical limitations of feet and engines and rockets, and where the scope and scale of combat may proceed beyond the ability of human observers to comprehend in real-time. As Dorothy Denning argues, “[a]t the speed of cyber, placing humans in the loop at every step is neither practical nor desirable.” As a result, in direct analogy to defense systems such as anti-missile systems, [m]ost anti-malware and intrusion prevention systems have both manual and automated components. Humans determine what goes into the signature database, and they install and configure the security software. The processes of signature distribution, malicious code and packet detection, and initial response are automated, but humans may be involved in determining the final response. Effective cyber defenses, in short, will have to rely upon automatic routines. Further, to date, the technological development and practical adoption of autonomous weapons systems for defense has progressed further than that of autonomous offensive systems.

AI enhances the security of a stable second strike, which is key to deterrence.

Horowitz, 2019 – prof of Political Science at Univ of Pennsylvania [Michael with Paul Scharre, and Alexander Velez-Green “A Stable Nuclear Future? The Impact of Autonomous Systems and Artificial Intelligence” December arXiv:1912.05291 https://arxiv.org/abs/1912.05291 Acc 12/27/20 TA]

Uninhabited nuclear launch platforms may be seen to offer some strategic benefits to nuclear-armed states. Nuclear-armed UAVs, for instance, could be kept aloft for far longer than is possible with human pilots, decreasing fear of a disarming first strike. B-2 bomber pilots, for instance, have flown a maximum of 40-hour missions.88 By contrast, refuelable UAVs could stay aloft for several days, limited only by engine lubricants and other reliability factors. Uninhabited aircraft have already conducted 80-hour flights.89 The maximum endurance record for a refuelable aircraft is 64 days.90 The ability to keep nuclear bombers in the air for longer periods of time might offer policymakers new tools for managing escalation. Long-endurance nuclear-armed UAVs could provide policymakers with additional options for nuclear signaling, since they could be kept on-station longer than would otherwise be possible. Likewise, if they are sufficiently survivable against adversary countermeasures, nuclear-armed UAVs might improve a state’s ability to deliver nuclear weapons in a timely manner since they could be kept aloft closer to potential targets longer than their manned counterparts. For some less powerful nuclear-armed states, UAVs may also be seen as a lower-cost, longer-range alternative to human-inhabited nuclear bombers. Lower-cost systems are unlikely to be as survivable as their more expensive counterparts, however, thus limiting their utility. Nuclear delivery vehicles that leverage AI for certain functions may also be seen to provide strategic benefits. For instance, the Status-6’s notional AI-enabled counter-ASW capabilities may help to improve Russian leaders’ confidence in their secure second strike regardless of advances in U.S. missile defenses by convincing them that their nuclear-armed torpedoes will always be able to reach their targets. This might constitute an improvement to U.S.-Russian nuclear stability.91 But any such reassurance will be limited by the fact that, while torpedoes may pose a threat to coastal targets, they cannot strike inland strategic targets, such as enemy leadership redoubts, command centers, strategic forces, critical infrastructure, or population centers.92 As a result, even if automation does improve Status-6 survivability, it would constitute only a marginal improvement to the overall viability of Russia’s nuclear deterrent, since from Moscow’s perspective, U.S. missile defenses and strike capabilities could still prevent it from using missiles to hold the full range of necessary targets at risk.93

Specifically, a fear of accidents will limit nuclear LAWs to only positive ones.

Horowitz, 2019 – prof of Political Science at Univ of Pennsylvania [Michael with Paul Scharre, and Alexander Velez-Green “A Stable Nuclear Future? The Impact of Autonomous Systems and Artificial Intelligence” December arXiv:1912.05291 https://arxiv.org/abs/1912.05291 Acc 12/27/20 TA]

Many of these ways that autonomous systems could increase the resiliency and accuracy of NC2 are speculative, however. Existing automation, as the Petrov incident shows, already creates the risk of automation bias. Knowledge of this will probably make most nuclear-armed states unlikely to further automate the early warning or command-and-control processes, with two exceptions: first, in situations where human-machine teaming might be further integrated to mitigate potential false alarms; second, in situations where a state fears for its secure second strike, and believes that further automation would reinforce deterrence of a potential aggressor. It is also possible, though less likely, that more automation could occur via a highly risk-tolerant nuclear-armed state that believes automated NC2 protocols would improve its ability to manage escalation.

Sign in to recut, update, or save this card.