Artificial Intelligence: A Revolution in Strategic Affairs?

Abstract

Nuclear weapons would be used by human decision-makers and would therefore be subject to human psychology, even if some theorists sought to use them in an abstractly rational manner. AI systems, by contrast, do not make decisions in the same way as humans would, even if their algorithms are often loosely modelled on human cognition. Additionally, AI will be more useable across a full spectrum of force, and indeed more broadly than force itself, shaping all conflicts from the lowest to highest intensity and the smallest to largest scale. Current and future capabilities in AIRecent gains in sensing, categorising and decision-making by modern AI have been marked and rapid. 1 Much progress has been made by AI systems utilising 'deep learning' methods that model, albeit with considerable abstraction and simplification, the neural processes of human brains. At the cutting edge of AI research, algorithms are demonstrating an increasing capacity to learn without supervision, with limited data for training, and to cope with ambiguous and asymmetric information. These developments move AI beyond rapid, brute-force calculation and pattern recognition within tremendous volumes of data -traditional strengths of machine intelligence. AIs can now win at Go (a game with far too many permutations to be susceptible to brute force) and poker (with its asymmetric knowledge of the stakes). 2 Most ongoing AI research is in sectors that are not explicitly defence related -for example in healthcare, internet search or civilianvehicle manufacture.Yet the broad capabilities under development -the ability to flexibly categorise information and use this as a basis for decision-making -have clear strategic utility. Early studies indicate that AI will be useful at both the tactical and the strategic levels. 3 For strategic-level decision-making, however, there is likely to be more ambiguity about the data on which to make decisions and the reward functions that any machine seeks to satisfy.AI's immediate role, it seems, will be in the tactical domain; but that will nonetheless have important strategic implications.AI is already a military reality. For example, weapons-guidance systems make decisions independently of human input (other than in setting the

Activism by the AI Community

“…Scepticism has been expressed about the coherence of the epistemic community on military AI, especially about whether it is presently sufficient to play a similar role to the ABM community [39,47]. It clearly has not yet been sufficient to pass an international ban or block military AI R&D efforts.…”

“…AI deployment is already influencing markets and the economy, justice and distributive decisions by governments, and our elections [32,63]. In coming years we may see the use of AI systems with novel capabilities in military domains [47]. Increasing use of increasingly powerful yet brittle systems carries accident risks, misuse risks, and structural risks [65] -risks to human security, and risks of societal inequality and discrimination [6,37].…”

Knowledge in the grey zone: AI and cybersecurity

“…AI and cybersecurity are also far from neglected in the realms of secret intelligence and military operations. AI is a sine qua non for modern militaries (Lewis 2019) and a potent augmentation to strategic planning (Payne 2018). Without AI in cyberspace, the former head of US Cyber Command remarked in Senate testimony, 'you are always behind the power curve' (Congressional Research Service 2019, 11).…”

Interdependência Securitária: o Caso da Proliferação e Controlo de Armamentos no Domínio da Inteligência Artificial

“…Falamos de sistemas de armas essencialmente utilizados para defesa aérea e defesa antimíssil, o que significa que respondem a ameaças de mísseis, mas não são capazes de lançar um ataque de forma independente. São muitas as vantagens militares apontadas às armas letais autónomas, tanto do ponto de vista tático como estratégico e a sua principal característica é o facto de aumentarem enormemente o poder de fogo (Payne, 2018). Adicionalmente, estes robôs não apresentam problemas comuns nos militares empenhados em missões, nomeadamente fadiga e emoção.…”