Tech-specific experts agree – there’s systematic bias toward alarmism.

Recent commentary has sounded the alarm on the ability of so-called “emerging” technologies to tilt the balance of terror and complicate strategic stability calculations. Indeed, from the longbow and gunpowder to the tank and nuclear weapons, technological innovation has revolutionized and redrawn the borders of the battlefield. The same will be true of the range of new capabilities on the horizon, not least those in the space and cyber domains. But how are these new technologies distinguishable from that which came before them—in qualitative and quantitative terms? And how can we measure the impact of these technologies in a way that is not alarmist, but rather, allows us to systematically evaluate them on their potential for disruption?

Comparing the impacts of heterogenous technologies side by side (especially across two domains) can be challenging; it is the policy equivalent of comparing apples with oranges. This becomes especially apparent when we consider the numerous and wide-ranging ways in which emerging technologies have the potential to complicate strategic stability. They could, for example, provide new ways to use or stop the use of nuclear weapons (e.g., AI for missile defense could significantly change the deterrence calculation). They might blur the boundary between nuclear and non-nuclear infrastructure (e.g., the dual-use nature of satellites might increase the potential for misidentification or unintended escalation). They could create new vulnerabilities within existing systems (e.g., new technologies might enable cyberattacks on civilian infrastructure). And more fundamentally, they might even change the game (i.e., the technology, and its proliferation to new actors, could make attribution more difficult).

However, it is crucial to note that we have always had to contend with new technologies, and it is too easy to forget what was considered “emerging” and when. As a result, it is important to frame a discussion about the impacts of emerging technologies upon strategic stability as exhibiting elements of both continuity and change. Continuity does not mean that nothing changes, but rather, it helps to sober overblown claims that every emerging technology seemingly represents a revolution in military affairs.

With that in mind, perhaps it is possible to develop a measured and systematic means of forecasting the actual impact of individual technological capabilities.

STREAM

The Systematic Technology Reconnaissance, Evaluation, and Adoption Methodology (STREAM), developed by RAND, facilitates an assessment of the potential relative impacts of new technologies. The STREAM approach was designed to assess current and future technologies according to a range of impact and implementation criteria. STREAM culminates in a scoring exercise, wherein subject-matter experts score technologies in the cyber and space domains on the impact that they might have on strategic stability, as well as any barriers to their implementation, out to a ten-year horizon.

The findings presented in the following section are based on research that focused on the UK and its allies. They operate on the assumption that there are five major factors that dictate whether a given technology will impact strategic stability within the next ten years:

Current technological maturity

Rate of development

Capability increase

Ease of countering

Crisis instability

While this is by no means a definitive list, together these metrics produce a model that is not too heavily reliant on any single definition of strategic stability and that is modular (i.e., it can be arranged in different ways to reflect different definitions for strategic stability).

Research Findings

The STREAM scoring exercise generated a substantial amount of qualitative and quantitative data. In the interest of brevity, I will focus here on the most high-level findings.

Figure 1 combines the current technological maturity with the rate of development, which facilitates a better understanding of where the technologies are right now and how quickly they might reach maturity (if they are not already at a high level of maturity). The rate of development for each technology is deduced from the extent to which barriers to implementation differ for the UK and its allies versus for the UK’s adversaries. A higher degree of dissimilarity would translate into a higher likelihood of asymmetric capabilities in the medium to long term.

The technologies highlighted in red in Figure 1 are at a moderate maturity level, with lower ethical barriers for the UK’s adversaries—relative to the UK and its allies—and have around the same practical barriers for both parties. These include rendezvous and proximity operations; AI-trained offensive cyberattacks; AI for intelligence, surveillance and reconnaissance; and direct ascent anti-satellite weapons.

The technologies highlighted in amber in Figure 1, meanwhile, paint two distinct pictures. On the one hand, cyberattacks on biotechnology is technologically immature, but it has the greatest disparity in ethical barriers between the UK and its allies versus the UK’s adversaries, meaning that its development could happen quite quickly. For this reason, it is potentially destabilizing. On the other hand, unmanned aerial vehicles are very mature technologically, but the lower ethical barriers for the UK’s adversaries relative to the UK and its allies might mean that the UK’s adversaries are less constrained in their deployment of this technology. Thus, in combining technological maturity and rate of development, we can see that these technologies constitute different kinds of threats to strategic stability.

Figure 1: Prioritizing technologies using technological maturity and rate of development metrics

Figure 2 combines capability increase, ease of countering, and crisis instability. Combining capability increase and ease of countering illustrates how technologies that could significantly increase capabilities (defined as the single largest increase in offensive capability, defensive capability, or ability to reduce or enhance the reliable functioning of C4I—command, control, communications, computers, and intelligence) and are also difficult to counter could pose a significant threat to strategic stability. Furthermore, the addition of the crisis stability variable indicates which technologies have the highest potential to trigger an escalatory response, possibly even one that results in a nuclear first strike.

The technologies highlighted in red in Figure 2 (hypersonic missiles and AI-trained offensive cyberattacks) have the potential to significantly increase capabilities, are difficult to counter, and are highly likely to accelerate crisis instability. On the other hand, the technologies highlighted in amber (quantum computing, quantum cryptography, AI-powered autonomous response to cybersecurity threats, AI for missile defense, and direct ascent anti-satellite weapons) will have a similar impact on capabilities and are similarly difficult to counter, but they are slightly less likely to accelerate crisis instability.

Figure 2: Prioritizing technologies using increase in capability, difficulty in countering, and acceleration of crisis instability metrics

This study used a simple addition method in order to pull all the quantitative results together. A more advanced model (such as one that uses machine learning to cluster technologies) could capture these results in a more nuanced way and include more factors (such as the likelihood of the technology being produced in the private sector, for example, or the ability to attribute the use of this technology to a specific actor). Notwithstanding the various ways that this study could be further developed, this independent study was intended to be a proof of concept that would form the foundation for future research.

Figure 3 shows how the addition method identified the five technologies with the highest potential to disrupt strategic stability. These are: AI-trained offensive cyberattacks, hypersonic missiles, rendezvous and proximity operations, direct ascent anti-satellite weapons, and satellite jamming systems. The caveat for these findings is that the five aforementioned metrics (current technological maturity, rate of development, capability increase, ease of countering, and crisis instability) combine to produce a specific model of strategic stability that is not universal. Using the same data set, but different metrics for determining whether a given technology will impact strategic stability within the next ten years, could yield a different ranking altogether. This should not be seen as a shortcoming of the research, but rather, as an intentional characteristic.

In using these five metrics, the research is not too heavily reliant on any single definition of strategic stability, which is a contested concept—Russia and China have their own definitions, for example, which differ from ours—and one that is used without clear meaning. As a result, the data set generated by this research can be arranged in different ways to reflect different definitions for strategic stability. The intent here is establishing a model within which the appropriate metrics can be identified and weighted with greater precision and awareness of the context.

Figure 3: Ranking the technologies

Controlling for Bias

Another important caveat of this research is that just because a method is systematic does not mean that it is necessarily free of bias. For example, research participants of a specific background might respond differently to the STREAM scoring exercise. However, measures were taken to minimize bias to the greatest extent possible. When shortlisting technologies, I used a literature review and key informant interviews with a different group of experts from those who responded to the STREAM scoring exercise. In the scoring exercise, experts were asked about their certainty levels and to provide sources to justify their responses.

In spite of this, one of the major qualitative findings was that even experts are susceptible to hype in their assessments. This is evident when examining the scores for hypersonic missiles. Experts were given the opportunity to abstain from scoring any technology if they were not familiar with it. However, all experts (n=10) felt that they were in a position to assess hypersonic missiles—thereby indicating some literacy with it—which might stem from the fact that hypersonic missiles are arguably the most widely covered and publicized technology included in this study. It is possible that a cohort of missile technologists might be more skeptical of whether hypersonic missiles really are a game-changing technology (i.e., how different they are from ICBMs).