AI in Emergency Response: Not the time for experiments

By Daniela Weber, Director of NetHope's Center for the Digital Nonprofit

Wouldn’t it be great if humanitarians responding to an emergency had a reliable sidekick with them who supports their planning and decision making? One who does not get stressed, calmly parses through large amounts of data in the most stressful and chaotic situations, makes helpful suggestions for priority actions, and supports positive interactions with affected communities?

AI undoubtedly has the potential to transform the way we deliver emergency responses. As with all innovation, we must think carefully how we develop and deploy it, making sure it follows the principles of “Do no harm”, is developed collaboratively, and informed by experience. If we follow these principles, AI solutions can help deliver the much-needed step change to effective, digitally supported emergency response.

We have already seen examples of what the opportunities could be. From analyzing vast datasets to identify patterns, predict needs, and optimize resource allocation, to providing natural language interaction in multiple languages, it offers much needed support and assistance to communities impacted by emergencies and those who undertake to support them – which is invaluable in a time of poly crisis and diminishing funding to respond.

• AI analyzes satellite imagery to assess disaster damage, helping organizations to assess where and how to respond.
• AI-driven social media monitoring to gather real-time information from impacted communities.
• AI-powered forecasts help to target help where people will be impacted by conflict or extreme weather events.
• AI-supported chatbots are to provide refugees access to critical information and services or to help translate from and to locally spoken languages.

The allure of using AI solutions in rapidly unfolding humanitarian crises is undeniable. With the need for rapid decision making and action, seeking instant answers from AI seems a great opportunity. Yet, we must consider the complexities and potential pitfalls of deploying AI technology in such high-stakes environments.

Challenges and risks

The deployment of AI without careful consideration of its impact on affected communities is a significant risk. The reliability of AI outputs depends on the quality and representativeness of input data. Biased or incomplete data can lead to flawed decisions or exacerbate existing inequalities or vulnerabilities. And in the midst of an emergency, humanitarian staff do not have the time to verify the information provided by the AI solution.

Organizations wanting to use AI solutions in emergency response are accountable for the outcome – meaning they do need to understand, and address the risks, which include:

• Data quality and bias: AI systems rely on large datasets for training and operation. In humanitarian settings, obtaining accurate and unbiased data can be challenging, which may lead to flawed decisions or discriminatory practices.
• Lack of Contextual Understanding: AI systems may struggle to grasp the nuanced cultural and social contexts of different emergencies, potentially leading to inappropriate responses.
• Ethical Considerations: The use of AI in vulnerable populations raises questions about consent, accountability, and potential unintended consequences.
• Privacy and Security: Collecting and processing sensitive data in crisis situations raises concerns about data protection and potential misuse.
• Over-reliance on Technology: There is a risk that responders might become overly dependent on AI tools, potentially neglecting human judgment and expertise.
• Infrastructure limitations: Implementing AI solutions in remote areas with limited technological infrastructure and internet connectivity can be challenging.

Taking a responsible approach

An active crisis is not the place to deploy untested solutions. It is crucial that before AI based solutions are used at the front line of an emergency, they are developed and tested in a controlled environment with the involvement of humanitarian experts and community representatives. Only after rigorous evaluation and refinement, and with users fully trained on how to use the tools and interpret their output, can they be cautiously introduced in real-life emergencies – and they must be an aide, not a replacement, for human judgement and decision making. To harness the potential of AI while mitigating the risks, humanitarian organizations should consider the following approach:

• Sandbox First: Develop and test AI tools in controlled environments before deploying them in actual emergencies. This could take the form of tabletop exercises using real-world scenarios from previous emergencies, to test the validity of the AI-enhanced activities.
• User-centric approach: Include front-line humanitarians and community representatives in the development and test processes to ensure tools meet real needs and respect local contexts. Ensure they are represented in the assurance and validation teams that continuously assess the AI solution’s impact and effectiveness to make necessary adjustments based on feedback from responders and affected communities.
• Training: Ensure staff have a solid understanding not only in the use and limitations of the AI solution, but also know which data sources are being used and how the tool uses them, and are aware of the relevant ethical, privacy, and consent aspects.
• Parallel approach: When introducing AI tools in actual emergencies, use them alongside traditional methods and have dedicated teams alongside the responders to verify results.
• Transparency and Accountability: The use of AI, including its limitations, needs to be clearly communicated to build trust with affected communities.
• Ethical Guidelines: Adopt and adhere to clear ethical guidelines for AI use in humanitarian contexts, using the Humanitarian AI Code of Conduct developed by NetHope’s AI Working Group as a framework.

Most importantly - keep in mind that AI is a tool, not a magic solution. Human expertise, empathy, and understanding remain indispensable in humanitarian work. While technology can enhance efficiency and decision-making, it should never overshadow the core principles of humanitarian action: humanity, impartiality, neutrality, independence, and accountability.

Don't miss...!

NetHope has also published several resources on AI:

• Artificial Intelligence Ethics for Nonprofits Toolkit
• Humanitarian AI Code of Conduct
• Gender Equitable AI Toolkit
• Data Governance Toolkit: A guide to implementing data governance in nonprofits
• The Artificial Intelligence (AI) Suitability Toolkit for Nonprofits
• The Digital Development Guidebook

NetHope’s Strategic Plan – Digital Transformation & Innovation – AI Program Highlight
NetHope launched its 2024-2030 strategy last year. A generation ago, the Digital Divide described the gap in access first to computers, and then to the internet and essential technologies. But today, new digital challenges threaten international NGO mission advancement as organizations are forced to grapple with the widening gap between available resources and growing needs. These new challenges – what NetHope calls the new Digital Divides – represent critical gaps in infrastructure, systems, capabilities, and knowledge. Over the next decade, will lead collective action to bridge the New Digital Divides – skills and leadership, inclusion, protection, transformation and innovation, and climate adaptation and resilience - leveraging our powerful, proven model for advance global good.

Bridging the New Digital Divides Spotlight: Digital Transformation and Innovation
The innovation required to meet emerging and existing challenges to the nonprofit sector and the communities we serve only happens when it is made a priority. NetHope works within our network, data, and history of catalyzing collective action to see complex problems are met with insights, policies, and advice that are context-appropriate, efficient and scalable. NetHope will use our influence, networks, data, and history of catalyzing new and high-potential ideas to ensure complex problems are met with insights, policies, guidance, and technology solutions.

Program Spotlight: Augmenting AI Skills and Practices for Nonprofits
NetHope’s work in AI dates back to 2018 as much of our work begins, a gathering of Members and collective desire for action. Now, our program focuses on AI capacity building through training and the exchange of knowledge and experiences; practical approaches to implementing AI solutions for nonprofits, their staff and implementing partners; and evaluating and assessing solutions and case studies on their usefulness and impact for our sector.