Why human touch remains key to crisis response in the age of AI

When disasters strike, speed and accuracy in understanding the damage and needs can make a real difference for affected populations. Technological advances are helping make strides in the efficiency of humanitarian response — but ground truthing, the process of verifying digital insights through local knowledge, remains indispensable.

New technology is ushering in a shift in the way humanitarian responses take place. Digital tools such as satellite imagery, remote sensing, and artificial intelligence can now deliver critical information in hours. These insights would previously take weeks if not months to trickle in. This means help reaches communities faster and recovery needs are identified sooner.   

While technology helps accelerate and scale up response, one important lesson is also emerging: Human touch has never been more critical.

Community leaders, builders, and assessors with local knowledge hold the key to bringing data to life and driving recovery that is resilient and inclusive. We need both digital insights and local knowledge to work together.   

Local knowledge in action 

The United Nations Development Programme’s RAPIDA, or Rapid Digital Assessment, leverages a wide range of digital tools to provide key impact numbers on damage and affected areas within 72 hours of a crisis. Paired with ground truthing, RAPIDA ensures recovery efforts are targeted to rebuild homes, restore livelihoods, and strengthen communities.

In the aftermath of the August 2025 earthquake in Afghanistan, satellite imagery and remote-sensing analysis helped identify the most severely hit areas of Kunar, Nangarhar, and Laghman provinces. In mountainous and remote regions where access is limited, these tools provided an important layer of data on priority areas for debris removal, infrastructure repair, and response support. But then came the hard part: reaching the communities and verifying this information.

This requires a different set of abilities. Since the quake, staff on the ground have been working with locals to understand their needs and provide critical supplies before the harsh winter sets in.  

Similarly, in Myanmar, when a 7.7 magnitude earthquake struck in early 2025, AI-based models helped identify where damage and debris were most severe, including roads blocked by rubble, cutting off communities’ access to essential services. But it was local community workers who brought their knowledge into use while clearing the way: “We knew which paths mattered most for the school, the clinics, and the market,” a volunteer from a faith-based organization told us. AI pointed to the blockages and local knowledge prioritized solutions that put communities back on their feet fastest. 

In Colombia, remote sensing revealed the extent of flooding across Chocó and La Guajira after Tropical Storm Rafael in late 2024, highlighting the hardest-hit zones. But reaching communities required working with the locals and gaining their trust. Assessors from the Wayuu indigenous group worked alongside community leaders, leading door-to-door assessments and navigating inundated areas in four-wheel drive vehicles. “People opened up because they knew we were one of them,” a local assessor explained. Technology traced the waters’ reach, while community voices ensured no one was left behind. 

As disasters intensify, the future of crisis response lies in their intersection: digital tools and local knowledge working hand in hand.  

Why balance matters  

Without ground truthing, response risks being fast but blind. Satellite imagery may show floodwater receding, but not whether walls are structurally compromised, what is damaged inside homes, or how exactly crops and agricultural yields have been affected. That is where local assessors and experts play a critical role in deeper verification. 

And it’s not just the immediate response. Human network and local knowledge go a long way in ensuring the recovery and reconstruction efforts are in harmony with the environment and resilient to threats that may not be apparent in the initial assessments.

For instance, in Afghanistan, after the 2022 and 2023 earthquakes, UNDP’s response focused on vernacular architecture that leverages local materials and building techniques that have been passed down over centuries. Among other things, these structures are highly sustainable and better suited to the local climate. Coupled with modern scientific inputs, a new set of guidelines was prepared to support the reconstruction process. 

Unfortunately, there is often a tendency to swing too far in one direction: either over-relying on digital tools or leaning exclusively on traditional assessments. Technology enhances efficiency, scale, and accuracy. But if pursued at the expense of local participation and knowledge, it risks undermining the very goals it aims to achieve. Striking a balance is essential.  

From experience, three lessons are emerging:  

1. Complement, don’t replace: Digital tools work best when paired with local knowledge. Neither is sufficient alone; together, they provide a fuller, more reliable picture.  

2. Recognize and invest in local strengths : In some cases, local assessments can be faster and more precise, especially where language, culture, or context-specific nuances are critical.  

3. Build smarter tech and partnerships: AI models must be trained in local languages and contexts, and partnerships with the private sector are essential for the computing power and expertise needed to innovate. 

Humanitarian and development groups must resist the temptation to swing between extremes. Instead, we must integrate digital innovation with local knowledge, invest in better AI models, and strengthen partnerships with the private sector. Only by doing so can we ensure that responses are both efficient and grounded in the realities of the communities we serve.  

Editor’s note: The views expressed in this article are those of the authors and do not necessarily reflect the positions of their affiliated organizations.