As global geopolitics grow more complex, foreign affairs policymakers and analysts need to anticipate and predict conflict and crises more effectively, as well as devise better ways of estimating the impact of decisions and actions. But how? For example, today they need to assess the interrelationship of record inflation, a global heatwave, growing food shortages, the largest war in Europe since World War II, and a pandemic without seeming end.
One approach is the use of data-driven computer simulations, which have greatly advanced in the past decade, to combine the power of data and technology with domain knowledge.
The Department of State has placed emphasis recently on advancing its use of data. In 2021, it unveiled its first-ever Enterprise Data Strategy, with the second identified goal being to “accelerate decisions through analytics.”
Simulations are a tool that can help meet that goal. They use machine learning and artificial intelligence (AI) to apply behavioral science-based algorithms to data from trusted sources. These systems can rapidly process massive data sets, enabling sophisticated understanding of relevant social, economic, and demographic conditions affecting critical stakeholders, groups, and coalitions. They incorporate both quantitative and qualitative inputs to produce valuable insights.
Simulations were the recent topic of conversation at an event hosted by the Atlantic Council’s GeoTech Center. At the event, attendees discussed how – through the use of simulations – people, processes, and technology can combine to produce better evidence-based decision-making in diplomacy. The best insights are useless, however, if policymakers don’t trust them. When faced with technology whose behaviors are not fully transparent or explainable, skepticism is a natural response.
<<< Start >>>
<<< End >>>
However, computer simulations are designed to enable analysts and policymakers to predict and understand the outcomes of real-world issues by testing various scenarios, transparently evaluating the data, and modeling assumptions that drive the simulation. Simulations also allow analysts and policymakers to harness computing capabilities to uncover complex patterns in group or human behaviors that would be difficult to understand or detect through human judgment alone. This ultimately leads to better judgement and discovery of solutions to complex geopolitical problems.
When the stakes are high – as they always are in diplomacy – trust is paramount. Fortunately, computer simulations offer not only speed and scalability, but also unprecedented transparency and accountability to build that trust.
Overcoming traditional limitations
Foreign affairs has long relied on evidence-based decision-making – pulling data from myriad sources and stakeholders to understand the causal mechanisms driving situations and to define trustworthy, actionable recommendations on paths forward.
When exploring probable impacts and outcomes, policymakers and analysts often use traditional qualitative tactics to support decision-making, such as consulting subject matter experts and conducting tabletop exercises.
These tactics offer valuable insight that technology cannot always replicate. Consider - during the 2020 U.S. presidential election, aggregate human judgement correlated from prediction markets outperformed polls and most complex statistical models derived from them. Despite significant advances in AI and machine learning, humans can understand and analyze social trends in ways machines are not yet – and may never be – capable of.
Yet, human judgement can fall prey to unknown biases and assumptions. For example, a study found evidence that the length of sentences a set of judges gave to convicted defendants was influenced by the wins and losses of the local football team. It’s simply impossible for a person to truly identify and explain every single assumption and bias that may be unconsciously impacting a decision.
To compensate for potential bias, policymakers may use statistical models, which use data to forecast broad geopolitical events. Though they provide speed and scale, these models are highly influenced by input data, often yielding little or no explanatory insight.
<<< Start >>>
What’s key to improving geopolitical decision-making is a tool and process that both generates meaningful, relevant insight and inspires trust in stakeholders to actually use it.
<<< End >>>
Best of both worlds
Simulations combine the strengths of experts, tabletop exercises, and statistical models while overcoming their core weaknesses. They take data, which may include quantitative information on qualitative behaviors, as inputs and test the potential outcomes of different decisions. This capability helps decision-makers select optimal paths in terms of policy, negotiations, action, timing, and other factors.
With simulations, we can apply scientific rigor to what would otherwise be anecdotal data, erasing the false dichotomy between the quantitative and the qualitative. These systems answer context-based questions in a replicable and falsifiable form, enabling us to generate more meaningful insights.
Simulations are not inscrutable. Rather, the application of simulations allows analysts and policymakers to transparently reveal the data inputs, initial conditions, and assumptions of the formal theory that led to a certain decision. This provides a level of explainability essential to the diplomatic process.
Speaking the same language
To underscore just how critical this explainability is, here’s an example. A few years ago, I was working with a client to help model high-level negotiations for a major regional conflict. This model not only showed the trade-offs between the various parties’ positions, but it also revealed a previously undiscovered mechanism to achieve a cease-fire.
When I presented this result to the client, they asked how I got it. I said we ran a computer simulation that used an AI decision engine.
The client was more than a little doubtful: “You expect me to listen to a computer on an issue pertaining to conflict?” They didn’t use the model. Those words stuck with me.
<<< Start >>>
Decision-makers expect and deserve more justification than simply, “The AI engine said so.” Simulations combine human and machine insight to provide that needed transparency.
<<< End >>>
Where to begin
To successfully adopt and scale simulations for decision-making, technologists and data scientists must work together with policymakers and domain knowledge experts to set a strong foundation.
It’s important to answer the following questions to ensure the insights generated by a simulation are relevant and explainable:
- Scientific inquiry – How do we collect observations and test and falsify hypotheses?
- Social science theory – What does existing peer-reviewed literature say that can help target and explain our efforts?
- Historical context – How can we understand the assumptions underlying our models and the problems we are seeking to solve?
- Math – How can we present data transparently and replicate it for others to test, falsify, and validate?
- Data and technology – How do we determine the right tools to share, analyze, and synthesize data at scale?
The more data-driven decision-making is embraced, the more we can help prevent, mitigate, or terminate conflict around the world. Studies have found that in many cases of conflict, both sides would have chosen not to fight if they had known in advance the ultimate cost and potential means of resolution.
In recent years, data-driven decision-making has contributed to monumental achievements in other fields, such as vaccine development and reducing world hunger. Today’s simulation technologies offer similar opportunities for foreign affairs decision-makers to make compelling cases for peace and progress.