Alves, Justin.
Published in National Security Journal on 20 October 2026.
DOI: 10.36878/nsj20261020.05
Download the full PDF version – New Security Dilemma: Hype, Paranoia, and Vested Interests in the AI Development Race (559 KB)
Rapidly evolving technologies continue to shape the operating environment of
security professionals. This article critically assesses how New Zealand security
professionals respond to the pace and intensity of the race to develop artificial intelligence.
An examination of relevant academic literature, augmented by several
interviews conducted with subject-matter experts, shows that the development
race is driven not only by measurable technological factors, but also by subjective
forces such as hype, fear of missing out, vested commercial interests, and geopolitical
paranoia. This suggests the race to develop artificial intelligence poses a
new serious and complex security dilemma. Challenging conventional thinking
that sees artificial intelligence in world revolutionary terms, this article argues that
New Zealand’s approach aligns with the expert knowledge and is reasonable given
it lacks the economies of scale of larger states. For New Zealand policymakers,
the challenge lies in balancing the need to remain competitive as a sophisticated
adopter while resisting distorted threat perceptions and premature or misdirected
investment driven by hype rather than careful analysis. This content of this article
will help inform those New Zealand policymakers who are obliged to understand
the security dimensions of this development race.
Keywords: Artificial Intelligence, Media Framing, Technology Regulation, Autonomous
Weapons Systems, Geopolitical Rivalry, Deepfakes, Misinformation
Introduction
According to the Organisation for Economic Cooperation and Development, “[a]n [artificial
intelligence (AI)] system is a machine-based system that can, for a given set of
human objectives, make predictions, recommendations, or decisions influencing real
or virtual environments. AI systems are designed to operate with varying levels of autonomy.”
1 There are two main types of AI. The first, and humanity’s current evolution of
AI, is ‘narrow artificial intelligence’ or systems designed to do deliberately constrained
tasks, such as the famous AlphaGo system designed to play the board game Go.2 Narrow
AI are relatively advanced, autonomous systems or machines which act to achieve
set goals. 3 At the other end of the spectrum is the as-yet unbuilt artificial general intelligence
and, evolving from that, artificial super intelligence. This would likely be a system
functionally capable of developing its own solutions to problems, of thinking for itself,
and potentially setting its own tasks autonomously.4 What follows focuses exclusively
on narrow AI because most experts agree that humanity is many decades away from
achieving artificial general intelligence – assuming it is even possible.5
As a rapidly evolving technology, AI has both strengths and weaknesses. Its strengths
include an ability to assimilate vast quantities of data from multiple sources with near
perfect retention, a lack of biological weaknesses such as boredom or the need to sleep,
and a capability to update and upgrade itself at a scale and pace biological beings cannot
match.6 Another strength is its scalability and parallel processing ability.7 AI programmes
can also be copied and mass produced: for example, a single AI trained to a
level where it can defeat “Top Gun” fighter pilots who took years and millions of dollars
to train can be copied and replicated thousands of times at little cost.8 Its weaknesses include
its massive reliance on training data, either (expensively) tagged by humans as in
supervised machine learning or unstructured data as in unsupervised machine learning,
the quality of which in turn shapes its accuracy and behaviour. Other weaknesses
include a lack of reflection (which is a particular issue when ingesting biased data or
interpreting unusual contexts) and, of course, its lack of any corporeal body by which
to physically affect its environment.9
New Zealand’s response to the development of AI is multifaceted. While AI has been a
topic of discussion within the New Zealand business community for some time, with
the AI Forum New Zealand (Te Kāhui Atamai Iahiko o Aotearoa) being founded in
2017, a national strategy for AI was not released until 2025.10 This was relatively slow
uptake compared to, for example, the United Arab Emirates, which established a dedicated
ministerial portfolio for AI in 2017.11 The New Zealand Defence sector’s view of
AI is ambiguous; the Defence Capability Plan (2025) makes no mention of AI at all,
even when describing investments needed in the information domain, but comprehensively
discusses its implications in a subsequent paper which focusses beyond 2035.12
This may reflect a view within the defence sector that AI’s use in the military sphere is
still in its infancy and must build up its credibility with military professionals before it
can perhaps one day be relied upon as a fellow human combatant would. Alternatively
(or as well) it may simply reflect reluctance by Defence to telegraph intent and capabilities
in a public-facing document.13 New Zealand’s National Security Strategy describes
AI as a driver in the context of New Zealand living in an era of disruption, which will
create opportunities but also, potentially, highly disruptive social and economic changes.
14 It also notes the potential for misuse of AI by nation-states and criminal groups,
as well as the difficulty in creating international ethical standards for AI necessary for
establishing guardrails to constrain that misuse.15 The New Zealand Government’s AI
strategy highlights the high capital investment costs of developing AI and the importance
of New Zealand, as a small country, seeking to be a ‘sophisticated adopter’ rather
than a producer of AI.16
To critically assesses how New Zealand security professionals respond to the pace and
intensity of the race to develop artificial intelligence, the article brings together existing
expert knowledge on AI from both relevant academic literature and subject-matter
experts to offer an informed discussion of five themes of contemporary significance:
namely, that artificial intelligence will transform global society; the desire for first-mover
advantage is a key driver in the AI race; the role media plays in sustaining public
interest in the AI race; the lack of regulation resulting in little governmental control
restraining the AI development race; and the development race for AI, resembling an
arms race, may well constitute a new security dilemma. The literature review considered
over 50 published papers with a focus on the competition to develop AI and its
effects on business and geopolitics, including international security. To compliment the
literature review, interviews were undertaken with six experts on AI, computer science,
international security and international relations. These experts were not known to the
author before undertaking this research, did not know each other, and were based variously
in New Zealand, Thailand, Singapore, and China. The interviews were recorded
and transcribed, and written approval of the transcription obtained from each expert
before its use in this research. The interviews are anonymised and referred to as identifiers
I1, I2 etc. The sample size limitation in the use of only six interviewees is recognised;
however, their broad array of backgrounds was designed to partly compensate
for this. Another limitation is the lack of any input from AI companies – efforts to set
up such interviews were not successful. Lastly, as the experts were all citizens of New
Zealand or the United States (albeit some living overseas), there is a risk of associated
bias which must be acknowledged.
By comparing New Zealand’s current response to AI against expert knowledge, the
article also reflects upon what effects, if any, the AI race could have on international
security in the near-term, and how might this affect New Zealand.17 The current accepted
wisdom is that AI will be the key driver of the Fourth Industrial Revolution,
resulting in highly significant (and potentially destabilising) changes across virtually
all sectors, including society, military and economic.18 While this may transpire to be
true, at present the hard evidence to support this narrative remains limited. The key
proposition made by this article is that prosaic, subjective factors such as hype, paranoia
and fear of missing out regarding AI, may be more powerful drivers of the AI race than
objective metrics such as AI algorithm speed and semiconductor capacity. The AI race
may still influence strategic decision-making even in the face of currently limited empirical
evidence of advantage. Over the short to medium term, inter-state paranoia and
fear-of-missing-out regarding AI mean that the AI race will likely impact international
security even if AI products emerge to be less effective than claimed. Nonetheless, states
learning how to differentiate hype from reality and invest in AI appropriately will likely
gain an advantage over those which do not, and this is an important consideration for
smaller states such as New Zealand.
The article argues that New Zealand’s current government strategy regarding AI is
well-considered but entered comparatively late into the race.19 It is notably unquestioning
regarding the presumed benefits of AI for business, despite for instance, one empirical
study in the United States concluding the fiscal benefits of AI to be “zero” for 95%
of businesses that tried to deploy it.20 The race to develop and implement AI – while
simultaneously worried about rivals doing the same but faster – can have a powerful
insecurity multiplier effect and this should be kept in mind when making prioritisation
and investment decisions. New Zealand cannot opt out of the AI race without risk of
compromising future competitiveness in many critical spheres. However, New Zealand
needs to be wary of the “AI hype-cycle” and vested commercial interests of corporations
producing and selling AI. The purpose of this article is not to provide recommendations
or refute the future benefits of AI. It is to increase awareness of the powerful role of hype
and paranoia in elevating the relative importance of AI to security-professional and
government decision-makers, in the hope of contributing towards carefully considered
policy and investment choices.
Justin Alves holds a Master of International Security from the Centre of Defence and Security
Studies, Massey University, New Zealand, which was kindly sponsored by his employer, the Ministry of Business, Innovation and Employment. Correspondence with the author can be addressed to the Editors in Chief j.m.battersby@massey.ac.nz or d.r.rogers@massey.ac.nz