Did you know that the World Food Programme is integrating artificial intelligence (AI) into our emergency response efforts and enhancing the efficiency of food delivery to those in need? Read about the SKAI project and how it’s been impacting post-disaster evaluation and humanitarian responses globally.
By Amine Baha and Heath Morrell
Unleashing the Power of AI to Enhance Humanitarian Aid
A lack of on-the-ground information at the start of a humanitarian crisis is a major obstacle to a quick and effective response. Being able to understand the situation on the ground and the best ways to access and assist affected populations is critical. This is particularly difficult in locations where infrastructure and communications networks may be damaged or disrupted.
Developed by the World Food Programme (WFP) in collaboration with Google Research, SKAI is revolutionizing the way we respond to emergencies and deliver food to people in need, faster and more effectively. SKAI has proven to be 13 times faster and 77 percent cheaper in near real-time post-disaster building damage assessment operational situations.
At SKAI’s core lies the power of AI, cutting-edge machine learning algorithms, and vast amounts of satellite imagery, which enable real-time assessment of building damage on a grand scale. SKAI provides insights and actionable intelligence, which empowers organizations to make effective data-driven decisions with unprecedented precision and speed. From rapid building damage assessment to situational awareness and resource allocation, SKAI is redefining the way we respond to crises.
As an open-source tool, SKAI is transforming the landscape of disaster management and making a tangible difference in vulnerable communities worldwide. For example, SKAI has proven that it can undertake three weeks of manual damage assessment work in just 24 hours, targeting and prioritizing areas suffering the greatest impact.
By speeding up post-disaster evaluation and the subsequent delivery of much-needed assistance, SKAI may very well represent the future of disaster response and humanitarian aid. Its benefits can be seen in four key areas:
Machine Learning-based Damage Assessment
Leveraging state-of-the-art machine learning algorithms, SKAI automatically analyzes satellite images to assess the extent of building damage caused by any natural disasters and conflicts, anywhere at scale — context-agnostic technology. This enables rapid and accurate building damage assessment, facilitating prompt response and aid distribution.
Near Real-time Monitoring and Insights
SKAI provides near real-time monitoring capabilities, allowing tracking and analysis of disaster-affected areas by delivering critical metrics on building damage. Organizations, governments and local authorities, can then combine it with population density, infrastructure mapping, socio-economic vulnerability, and food security, enabling efficient resource allocation and targeted interventions.
Enhanced Situational Awareness
By comprehensively analyzing satellite imagery, SKAI detects changes in buildings’ structures, identifies affected regions, and predicts damage outcomes, equipping organizations with valuable insights for proactive disaster response.
Collaborative and Public Good
SKAI serves as a collaborative solution among stakeholders involved in disaster response and humanitarian aid. This promotes effective collaboration and maximizes the impact of response efforts.
Over 2.1 million buildings have been assessed in the near-real time since the open-source release of SKAI in June 2022. The following four country-specific case-studies illustrate the results and impact of SKAI across a range of humanitarian responses.
Türkiye and Syria Earthquake, 2023
Following the devastating earthquake that struck Türkiye and Syria, SKAI proved to be a valuable tool in disaster response. By providing timely and accurate analysis of the affected areas, SKAI played an important role in assessing damage, and assisting with resource allocation. While the exact measurement of its impacts and outcomes was challenging and not extensively verified, there is evidence to suggest that SKAI’s insights positively contributed to the response effort.
Precise and Accurate Assessment: During a month-long deployment, SKAI’s team of just five individuals demonstrated remarkable capabilities. They assessed approximately 600,000 buildings with an accuracy rate exceeding 81%. This assessment helped identify over 28,000 heavily damaged or destroyed structures.
Rapid Coverage of Vast Urban Zones: In less than a week, SKAI’s agile technology covered an extensive area of 530 km2 comprising urban zones with an estimated population of 1.3 million people. The analysis of over 265,000 buildings enabled the identification of over 7,000 structures severely damaged or completely destroyed in three major cities: Kahramanmaras, Iskendirun, and Antakya.
Providing Valuable Data: SKAI collected additional satellite imagery and released building damage data covering a portion of the impacted area. This data was estimated to affect approximately 5 million people across southeast Türkiye. While it is important to note that the exact impact of this data-driven approach was difficult to measure and not intensively verified, it aimed to empower decision-makers with the necessary information to plan and execute relief efforts, ensuring that assistance reached those who needed it most.
Hurricane Ian, 2022 (United States and Caribbean)
In the aftermath of Hurricane Ian, SKAI helped to effectively prioritize aid delivery and assess the impact on affected populations. Through real-time monitoring and damage assessment, SKAI enabled aid organizations to respond swiftly and effectively, providing critical assistance to those in need.
Enabling Swift and Effective Aid Delivery: With SKAI’s real-time monitoring and damage assessment capabilities, Give Directly was able to respond rapidly to the aftermath of Hurricane Ian, speeding up their disaster responses by six times. In a matter of weeks, US$ 3.3 million in cash assistance was delivered in total to 4,748 low-income families residing in heavily impacted neighborhoods of West Florida. This timely intervention provided essential support to those most affected, helping them rebuild their lives and communities.
Unprecedented Speed and Accuracy in Damage Assessment: In the aftermath of Hurricane Ian, SKAI demonstrated its remarkable capabilities by scanning 410,000 buildings across 3,000 square km within a few days. Through this rapid assessment, SKAI identified 69,000 damaged buildings, achieving an impressive precision and recall rate of approximately 70 percent. These critical insights provided a comprehensive understanding of the impact, allowing Give Directly to open enrollment for cash assistance within less than a week of the hurricane making landfall.
Record-Breaking Assistance Delivery: The combination of SKAI’s technology and Give Directly’s efficient aid distribution systems enabled the record-breaking delivery of $2 million in cash assistance to over 2,900 low-income households residing in heavily impacted areas in a couple of days. This assistance reached those in need weeks earlier than in previous disasters, providing much-needed support for their recovery and ensuring a faster return to normalcy.
Recognized Contributions and Partnerships: Through the partnership between Google and Give Directly, SKAI’s capabilities were harnessed to inform relief aid efforts, demonstrating its effectiveness in supporting data-driven decision-making and enabling swift response measures. SKAI’s impact on operations after Hurricane Ian was also acknowledged in the press and at Google’s AI@ ‘22.
Pakistan Floods, 2022
Amidst the devastating floods in Pakistan, described as “Never seen climate carnage” by UN chief Antonio Guterres, SKAI emerged as a potential game-changing tool in flood response. Its insights facilitated the monitoring of flood-affected regions, identification of the most impacted urban districts and remote villages, and the support of relief assessment efforts conducted by local authorities. The magnitude of SKAI’s impact is reflected in the scanning of over 850,000 households across 8000 km2 of land area, directly benefiting approximately 2.5 million people.
Transforming Flood Response with Timely Insights: In the face of an unprecedented disaster, SKAI’s technology provided situational awareness tools that supported flood response efforts. By accurately monitoring the affected areas, SKAI enabled a proactive approach to relief operations.
Enhancing Relief Assessment and Transparency: The data generated by SKAI’s analysis was shared with the Pakistan authorities, serving as a valuable resource to counter-verify the damage observed during ground surveys. This collaborative approach ensured a comprehensive and transparent process, mitigating risks associated with incomplete assessments. The recognition received from the Khyber Pakhtunkhwa (KPK) provincial government is a testament to the efficacy and reliability of SKAI’s technology in addressing these critical challenges.
Empowering Provincial Disaster Management Authorities: With the sheer scale of the impacted area, manual door-to-door surveys conducted by the provincial government faced significant challenges in terms of time and resource requirements. SKAI’s technology emerged as a game-changer by empowering provincial disaster management authorities in the KPK province. Through the identification of over 60,000 damaged buildings, SKAI achieved an impressive precision rate of 80%-85% and a recall rate of 65%-70%.
Durban Region Floods, 2022, (South Africa)
After the catastrophic Durban flash floods in April 2022, SKAI provided insights on the affected areas and infrastructure damage and played a role in facilitating post-disaster assessment with the South African WFP team.
Assessment of Flood Impact: After the South African floods occurred, the WFP Country Office and the provincial government sought support in assessing the extent of the damage. SKAI analyzed nine areas within the region, covering over 250,000 residential buildings and infrastructures. With an exceptional precision rate of 88% and a recall rate of 73%, SKAI successfully identified over 7,768 buildings that were damaged, providing valuable information for recovery efforts.
Empowering Decision-Making and Collaboration: The per-building and per-infrastructure assessment maps generated by SKAI were shared with the WFP South African Country team. These maps became tools for communication and validation, enabling the identification of impacted wards and quantifying the severity of the impact. SKAI’s data empowered local authorities to better understand and classify the effects of infrastructural damage, fostering effective collaboration and informed decision-making.
The success of SKAI across this range of disasters reinforces its position as a transformative tool in disaster response and recovery. By harnessing the power of artificial intelligence, SKAI exemplifies the potential of technology to make a tangible and life-saving impact in the most challenging situations, transforming disaster response into a proactive and efficient endeavor.
SKAI’s the limit?
As well as being used directly by humanitarian organisations, there is immense potential for SKAI to be integrated within the systems of local and national authorities to enhance and reinforce their suite emergency response tools. WFP is open to collaborating with local and national authorities to adapt SKAI models to the needs and challenges of specific regions. For example, by leveraging national satellite capacities and data sources, SKAI assessments can be tailored to the local context, increasing accuracy and relevance. Through knowledge exchange and capacity building, the SKAI project team can also provide training to local staff, empowering them to utilize SKAI effectively and independently.
SKAI will also be the core technology of the upcoming Disha platform, which aims to leverage AI and data innovation for sustainable development and humanitarian action. Led by UN Global Pulse, Disha brings together a coalition of partners including: Google.org; the Patrick J. McGovern Foundation; the Jain Family Institute; McKinsey & Company; the World Food Programme (WFP); the United Nations Development Programme (UNDP) and; the United Nations Satellite Centre (UNOSAT). By integrating SKAI’s powerful capabilities into Disha, its reach and impact will be expanded, driving greater efficiency, and fostering collaboration among diverse stakeholders across disaster response, and beyond.
For a deeper understanding of SKAI’s technical details, methodologies, and code, explore our GitHub repository. Learn more by engaging with the SKAI community, contributing to ongoing projects, and helping to shape the future of AI-driven disaster response.
Developing the Future of Disaster Response
WFP is open to exploring partnerships and opportunities for collaborations that support ongoing development of projects, such as SKAI. Our goal is to further leverage the capabilities of SKAI to ensure deployment of the most effective damage assessment tool for disaster-prone regions. We remain committed to assisting in the development of more resilient communities and driving effective humanitarian action in the face of adversity.
If you are an investor or an organization passionate about leveraging AI for disaster response and humanitarian aid, you can contact the WFP Innovation Accelerator at email@example.com and join our mission to end hunger by 2030.
The WFP Innovation Accelerator sources, supports and scales high-potential solutions to end hunger worldwide. We provide WFP colleagues, entrepreneurs, start-ups, companies, and non-governmental organizations with access to funding, mentorship, hands-on support, and WFP’s global operations.
This story was originally prepared in advance of the AI for Good Summit, where WFP participated in the session Disrupting hunger with AI: AHEAD and HungerMapLIVE and shared insights on SKAI at a booth co-hosted with the United Nations Satellite Center (UNOSAT).
Find out more about us: http://innovation.wfp.org
Subscribe to our e-newsletter