The Machined Humanitarian: Optimizing Crisis Decision-Making With Artificial Intelligence

Imagine widespread ethnic unrest has broken out in an austere foreign location where you are managing a humanitarian relief organization. The violence is quickly spreading to many parts of the city and involves an amalgam of armed factions, criminal gangs, and government forces. The scope and scale of humanitarian needs are high and growing by the hour. Hundreds of thousands of people have fled their homes, markets are shuttered, and local hospitals are quickly overwhelmed. Furthermore, the country's social and ethical norms are rapidly eroding. Mass looting and communal violence ensues, and media outlets are espousing inflammatory hate speech towards minority ethnic groups and the international community, including your aid organization. Indeed, the situation is highly volatile and urgent decisions are necessary to shape a workable relief response and protect your staff.

Now imagine that one of your colleagues can capture and make sense of the crisis in real-time. This teammate continuously monitors the entire open-source media environment and flags events pertinent to your security and operations. Critical information such as an imminent threat to your staff are immediately delivered to your smartphone or desktop. As the crisis unfolds, this teammate also captures relevant macro trends to help you anticipate and prepare for new destabilizing dynamics, including a mass atrocity or a famine. The staff members' support enables your team to navigate extreme levels of urgency, uncertainty, and complexity, which optimizes your relief response and ultimately saves thousands of lives. This teammate, however, is not human. It is an artificially intelligent platform called Project Arrow.

At the Intersection of Human and Machine Intelligence

Humanitarians plus artificial intelligence (AI) can make better crisis decisions than either alone. Indeed, during the past decade, AI innovations have made human-machine collaboration a reality and are enabling leaders to make faster, more informed, and effective decisions than ever before. Such human-machine teaming is called intelligence augmentation, which injects a mix of robust data sets, analytics, and AI into organizational decision-making processes.

Intelligence augmentation is well-suited for supporting operational decisions during a crisis. On the one hand, AI outperforms humans in analytic approaches to decision-making and can collect, process, and synthesize information at speeds far beyond the cognitive capabilities of humans. AI can also increasingly help leaders overcome their cognitive biases, and deliver personalized, evidence-based rational recommendations in real-time.

On the other hand, humans have distinct advantages over machines with intuitive approaches to decision-making. Human intuition, emotional intelligence, and creativity are powerful tools for managing a crisis response and placing problems into a meaningful context. Sometimes pure instinct takes over, and emergency action requires creative approaches where there is no past precedent or data to guide the AI system. With intelligence augmentation, then, it's all about the synthesis of human and machine intelligence for optimal crisis decision-making.

Understanding Crisis Decision-Making

Humanitarian emergencies, like the fictitious crisis above, are highly volatile, adaptable, and multilayered — and the stakes for humanitarians are incredibly high. Indeed, humanitarian assistance is one of a few professions where a single decision straddles the fine line between saving lives and human catastrophe. Not surprisingly, emergencies are described as "natural laboratories of human error," and relief efforts often result in unintended consequences for the very people and societies organizations are dedicated to serving.

Humanitarians must continuously grasp changes in the operating environment, decide what resources are needed, when and how they will be delivered, and of course, to whom. These decisions are often laden with extreme levels of urgency, uncertainty, and complexity, and require a holistic understanding of multiple stakeholder equities. There is often limited information, finite resources, and high expectations to deliver aid to those most in need. Many aid organizations have emergency action plans, humanitarian guidelines, and communications procedures activated during a crisis. However, mistakes are commonplace and therefore require greater decision-support resources.

Project Arrow - How Machines Can Help

AI technology such as VoxCroft's Project Arrow exemplifies the intelligence augmentation approach to crisis decision-making. Project Arrow was designed to codify the decision support function of an elite risk analytics unit, providing timely and accurate insights for optimal business decisions and policy action. Project Arrow autonomously collects, analyzes, and provides critical information in near real-time, helping organizations overcome the challenges of urgency, complexity, and uncertainty. In essence, Project Arrow is an organization's eyes and ears on events and emerging trends impacting your business — all through a highly intuitive software application.

Project Arrow's real-time monitoring and alert functionality is exceptionally reliable at mitigating the liability of urgency and uncertainty during a crisis. Project Arrow captures, synthesizes, and displays the most relevant information much faster than a human being. Thus, it enables organizations to quickly grasp what is happening and efficiently respond to protect assets and save lives. Project Arrow's algorithm sifts through millions of data points from hundreds of data sources and displays what is most relevant for the end-user. The information is presented in either a customized online database or text message and email alert. For example, humanitarian organizations can set up a Project Arrow instance to monitor and track security threats to aid personnel, implementing partners, and aid deliveries, and receive a text message alert when especially threatening incidents are discovered within Arrow's data streams. Arrow also reports on dynamics and factors that pose an indirect security risk to humanitarian activities, such as roadblocks, looting, and political protests. Indeed, if time is the most precious commodity in a crisis, then Project Arrow's processing and analytical speed is a critical force-multiplying capability to humanitarian response.

In addition to real-time monitoring and alerts, Arrow synthesizes multiple open data sources for early-warning trend detection, helping humanitarian leaders overcome complexity in their decision-making and anticipate changes in the crisis. For example, Arrow can autonomously identify indicators of food shortages and dynamically assess the risk of famine or communal violence — collecting data and matching keywords associated with academic and expert early warning frameworks. This task has long been carried out by humans and requires days, if not weeks of research to aggregate the data and make an assessment of an area's vulnerability to a food crisis. Arrow's AI can do this in seconds.

Additionally, humanitarians can better grasp and anticipate humanitarian needs with Arrow's high quality data sets and opinion mining capabilities. Arrow draws on large volumes of public data derived from traditional and social media sources. High quality data is vital to AI. If the data is well structured and classified into accurate categories, then the machine performs its task with utmost accuracy. AI algorithms, furthermore, struggle to extract nuance, context, and public sentiment. To overcome this limitation, Arrow leverages data filtered through tailored algorithms and a human crowd. The crowd captures public sentiment and identifies issues of highest concern.

In the fictitious scenario at the start of this commentary, for example, Project Arrow's data provider leverages a virtual human crowd with cultural vernacular language expertise to assess public sentiment and perceptions toward the aid community, and identify the issues and needs of highest concern. The human-verified data and Project Arrow's algorithm, therefore, can detect and alert to rising negative public sentiment toward particular actors in the crisis and the critical themes in social media discourse. This means that humanitarian leaders will have a much clearer sense of what the people want and need during a crisis, and how various stakeholders will react to certain humanitarian activities.

Time to explore automation

AI serves organizations best when it complements, rather than replaces human decision-makers. Humanitarian leaders should continue to hold agency over crucial decisions and leverage the wisdom, experience, and creativity of their staff. But they also need access to advanced technological resources to navigate extreme levels of uncertainty and make sense of a chaotic environment. Project Arrow showcases what is possible through human-machine teaming in a crisis. Humanitarian organizations, both big and small, should explore the decision support advantages of automation, and integrate technical solutions in a manner that works for their teams, and ultimately, their missions.