The Dashboard Delusion: What Years on the Ground Have Taught Me About AI and Disaster Response in Indonesia

I have stood in places where the data said one thing and the truth said another. In Palu, after the earthquake, a national dashboard marked a health facility as "operational." It wasn't. The staff had fled days earlier, their homes destroyed, and the building sat empty under a banner of green on someone's screen in Jakarta. Meanwhile, a facility flagged "severely damaged" was already treating patients again — because the community had quietly rebuilt it themselves, with no system anywhere to record that fact.

That gap — between what the screen shows and what is actually happening — is the story of 2026.

A Year of Digital Ambition

This year, Indonesia has rolled out an extraordinary wave of digital ambition. SIKELIM, the real-time dashboard mapping disaster-affected health facilities, has expanded nationwide. BMKG and the Ministry of Health have launched an AI system that predicts dengue outbreak zones a full week in advance, already running in Jakarta, Bali, and Yogyakarta. The state has earmarked up to 60 trillion rupiah for emergency disaster response. And just two weeks ago, the WHO released a new discussion paper on using AI for "evidence-informed" health policy.

On paper, this is a triumph. In the field, it is something else entirely.

What No Satellite Can See

When floods and landslides tore through Aceh and North Sumatra earlier this year, killing more than a thousand people and displacing hundreds of thousands, the technology arrived fast. AI-powered satellite mapping. Automated damage assessments. Crowdsourced disaster data platforms. All of it sophisticated. All of it, on paper, "evidence-based."

But here is what no satellite can see: who actually has the authority to redirect a health worker, activate a backup facility, or decide which village gets aid first. In the chaos after a disaster, that decision is never made by a dashboard. It is made through negotiation with local leaders, pressure from district officials, and logistics realities that exist nowhere in a database. AI can show you where the damage is. It cannot tell you who has the power to act on it — and in a crisis, that question matters more than any map.

The Last-Mile Trust Gap

I saw this play out again just last week, when flooding hit Medan, displacing thousands of families. While global reports celebrated Asia-Pacific's "digital health scale-up," the actual data collection at the evacuation posts was handwritten notes, WhatsApp messages, and Excel sheets typed on phones with patchy signal. The AI tools being celebrated in policy circles require clean, structured, real-time data. That data simply does not exist where it's needed most — not because the technology is lacking, but because nobody designed the system for the people who would actually be entering it: exhausted, frightened staff with twenty more urgent things to do.

The dengue prediction system is, in many ways, the clearest example of what I call the "last-mile trust gap." The AI model is accurate. It can tell you that cases will spike in a specific district next week. But that warning has to travel from a provincial dashboard into the hands of a posyandu worker, a local NU or Muhammadiyah volunteer, a Red Cross team — people the community already trusts, who can mobilize fogging crews and prepare hospital beds within hours, not days. Right now, that signal often gets stuck. It reaches the screen. It doesn't reach the street.

Asking the Wrong Question

None of this means Indonesia should slow down its digital ambitions. The investment, the political will, the technical talent — all of it is real, and all of it matters. But the question driving every one of these systems is the wrong one. The question has been: how do we get better data, faster? The question should be: who, on the ground, receives this signal — and do they have the authority and the trust to act on it within hours?

Indonesia already has the answer sitting in plain sight. We have a vast, trusted network of community health workers, faith-based organizations, and local volunteers who have been doing this work — often without any technology at all — for decades. The smartest investment we can make isn't a more powerful algorithm. It's building the bridge between what the algorithm knows and what the person standing in the floodwater can actually do about it.

Until that bridge is built, every dashboard we launch will be a more beautiful picture of the same unsolved problem.

Sources Referenced