For more than 16 years, Indonesian fisherman Dedi Arianto relied largely on his instinct as he ventured out into the Java Sea every day on his boat. But he could never be certain about the size of his catch. Then two years ago, officials from the National Research and Innovation Agency (BRIN) visited his village and demonstrated an AI-based mobile application they said would take the guesswork out of fishing.

The NN Marlin application can locate the densest fishing grounds based on the sea surface temperature and the level of chlorophyll. The app uses machine learning to crunch satellite data and remote sensing to identify schools of fish within a certain radius. “Now, catching fish is much more certain and efficient,” 42-year-old Dedi told Rest of World. “We are guaranteed to get fish every day, and the use of fuel is also more efficient.”

Fisheries contribute nearly 3% of Indonesia’s GDP and support millions of livelihoods. So the technology is invaluable to fishermen like Dedi. It is one of the many AI-based tools that BRIN, a government agency set up by President Joko Widodo in 2019, has built for everyday use. The agency leads the implementation of the national policy on AI, which has five priority areas, including education, health care, and food security. 

Unlike wealthier nations where startups and big tech firms have taken the lead on AI, in Indonesia BRIN plays this role, collaborating with local and international tech and research organizations on AI. It has so far built AI-based applications to study climate change, predict forest fires, monitor mangrove deforestation, enable more sustainable mining, and improve disaster management, among other uses. While investments and adoption of AI in Indonesia have lagged behind neighboring countries, such as Malaysia and Singapore, BRIN is helping bridge the gap, Pratama Persadha, chairman of the think tank Communication and Information System Security Research Center, told Rest of World.

“Indonesia has extraordinary cultural, geographic, and economic diversity, so with BRIN leading on AI policy, applications can be tailored to the unique needs and specific challenges of Indonesia,” he said. In addition, BRIN will “strengthen national research and innovation capacity by ensuring that AI development involves local scientists and researchers … and that national interests are maintained, because big tech companies that lead AI development are often focused on commercial gain.”

AI could contribute nearly $1 trillion to Southeast Asia’s GDP by 2030 and boost Indonesia’s GDP by about 12%, according to a study by the consulting firm Kearney. AI is key to Indonesia’s goal to transform into a high-income economy by 2045, and achieve “more inclusive and sustainable economic growth,” Teguh Prayogo, a researcher at BRIN, told Rest of World.

The NN Marlin app, for example, is used by hundreds of fishermen across the archipelago, said Teguh, who helped develop the app. Each fisherman pays 1.2 million rupiah ($73) for a tablet bundled with an internet data subscription. BRIN sends information on potential fishing zones up to six times a day, and it has other uses, too, he said.

“Authorities also use it to detect illegal fishing … and monitor the activities of foreign fishing vessels,” he said. Indonesia, the world’s largest archipelago, loses at least $3 billion annually to illegal fishing activities, authorities estimate. It is also entangled in territorial disputes in the South China Sea. BRIN is figuring out ways to detect illegal and suspicious activities more easily using AI, Teguh said.

BRIN is also using machine learning to help law enforcement agencies. In May, the North Sumatra police said they had used remote sensing technology in an application developed by BRIN to locate illegal marijuana fields. The technology can distinguish marijuana plants even in the middle of a forest by the quality of light reflected off its leaves, BRIN researcher Dedi Irawadi told Rest of World. “This algorithm can even be used to determine areas that have the potential to be planted with marijuana,” he said.

“Using AI makes our operations more certain, as, previously, informants would sometimes leak the information beforehand.”

The app has helped the police enormously, as they previously had to depend on unreliable informants, Inspector Jaya Syah Putra told Rest of World. “Using AI makes our operations more certain, as, previously, informants would sometimes leak the information beforehand, and the operations would fail.”

With the largest economy in Southeast Asia, Indonesia has recently drawn major commitments from big tech firms. Earlier this year, chipmaker Nvidia and Indonesian telecom firm PT Indosat Ooredo Hutchison announced $200 million in investment for an AI center in Central Java. Microsoft has pledged $1.7 billion in AI and cloud infrastructure, along with training for 840,000 people, while Google Cloud said it would back development of generative AI solutions for health care in the country. 

For its part, the government is offering incentives to firms and dual citizenship to professionals to retain talent. It is also looking to use AI to preserve its culture and identity. Indosat is one of several firms building multilingual large language models (LLMs) trained on Bahasa Indonesia and other regional languages to support chatbots and other AI tools for businesses. A government agency is building an AI system with OpenAI that will “align with the nation’s values.”

While the country is ramping up adoption of AI, there are concerns, including safeguarding the large amounts of data being collected to train AI models. In June, a major ransomware attack affected more than 160 government agencies, disrupting services, including immigration and operations at major airports. It was the latest in a series of cyber attacks on Indonesian companies and government agencies.

Get The Global, our (free) newsletterSign up for our weekly newsletter and we’ll send you our latest stories, dispatches from our staff, what we’re reading, and more. A world of tech, right in your inbox.

The vast amounts of data collected for training AI increases vulnerability to such attacks, said Persadha. The government must implement the Personal Data Protection Law, which regulates how data is collected, stored, and used, and “security must be considered from the design and development stage,” he said.

For fisherman Dedi, the security of data is not as big a concern as its accuracy. With dwindling fish stocks due to climate change impacts and territorial disputes, his livelihood is increasingly dependent on the app’s ability to locate fishing grounds, he said. But the fee is too high for some fishermen, and the app does not provide information on the types of fish, he said.

“For example, we would like information on red snapper or white snapper because of their high value. But the app does not give those details,” he said. “What is certain with this app is that we are confident that we never go home without a decent catch.”