The Indonesian government is looking to tap into artificial intelligence (AI) to improve its social protection (Perlinsos) programme, verify identities and assess eligibility of applicants for social assistance.
However, it recognises that AI is not a silver bullet. Even the most advanced systems cannot produce good decisions unless they are supported by accurate, up-to-date, and interoperable data.
“One of the most important foundations is high-quality data. How we can make use of data held across different agencies for the benefit of public services,” says Indonesia’s Digital Transformation Acceleration Committee (KPTDP)’s member, Rahmat Danu Andika, to GovInsider.
To address data quality concern, the government is first focusing on enabling interoperability to connect multiple data sources across ministries, agencies, and other institutions.
According to Andika, this approach aims to accelerate the verification process for social assistance services and improve the accuracy of eligibility assessment.
By drawing on data from multiple agencies, the government can tap into AI to analyse more comprehensive datasets to support decision-making in the delivery of social assistance, he adds.
Ahead of the nationwide rollout of the Perlinsos programme later this year, Andika shares how the government is building the foundations for cross-government data interoperability to support the use of AI.
AI strengthens social assistance verification
AI is already being deployed at the registration stage of the Perlinsos Portal.
“By using liveness detection supported by AI models, we ensure that the person registering on the portal is genuinely the individual entitled to receive assistance,” he says.
This helps reduce the risk of identity fraud and manipulation during the social assistance application process.
According to Andika, this is an early example of how AI can strengthen public services without completely replacing human decision-making.
The potential of AI extends beyond identity verification.
As data becomes increasingly interoperable across agencies, AI can also support other functions, including assessing an individual’s eligibility for social assistance.
“Once this data becomes interoperable, AI layers built on top of it can be used for a wide range of purposes,” he notes.
Piloting cross-agency data interoperability
Andika highlights one of the biggest challenges in delivering social assistance is ensuring that recipient’s data accurately reflects their current circumstances.
To address this challenge, the government has begun connecting data across ministries, agencies and other institutions, including the State Electricity Company (PLN), BPJS Kesehatan (Indonesia’s national health insurance programme) and Indonesia Statistics Bureau (BPS).
“This is the first time we have used data interoperability to assess whether someone is eligible for a social assistance programme,” says Andika.
The data interoperability pilot was first launched in Banyuwangi before being expanded to more than 40 regencies and cities across Indonesia.
According to Andika, combining information from multiple sources gives the government a more comprehensive picture of people’s circumstances. It also enables AI to generate more accurate recommendations than relying on a single data source.
Interoperability has also transformed the way beneficiary’s data is updated.
Previously, citizens had to submit appeals through the Ministry of Social Affairs and local governments. Now, they can do so much more quickly through a system that is directly integrated with BPS.
“The impact is that people can enjoy significantly greater transparency and convenience than before, when we often relied on data that was already three or four years out of date.”
Ensuring no one is left behind
Even so, Andika acknowledges that the new system still has its limitations.
People’s condition can change much faster than government datasets are updated. As a result, we can have a situation where someone, who has become eligible for assistance, does not get it as the changed status is not yet reflected in the system, while others who may no longer qualify are shown eligible, he notes.
For that reason, mechanisms to continuously update data, as well as to allow citizens to submit appeals manually, remain an essential part of the process.
“When someone is excluded from the list of social assistance recipients, there must be a mechanism for them to understand why and to correct the data being used,” he says.
At the same time, the government must ensure that digitalisation does not create new barriers to accessing public services. Not everyone has access to digital devices, digital literacy, or reliable internet connectivity.
As a result, AI-powered public services must continue to take into account communities with limited digital access.
The true measure of success is whether AI helps the government better understand people’s circumstances and ensures that assistance reaches those who genuinely need it, he concludes.
