Hooking the future: Scientists deploys AI to track hilsa in Bengal—but fishers fear it may net more than it should

In a move that could reshape traditional fishing practices, scientists in India have rolled out an AI-powered forecasting system to help fishers locate hilsa shoals in the Bay of Bengal, Times of India reported, marking a first-of-its-kind initiative that promises bigger catches but is already stirring concerns over sustainability.

Developed by the Indian National Centre for Ocean Information Services under the Ministry of Earth Sciences, the Hilsa Fishery Advisory Service uses a machine learning model—XGBoost—to predict where the prized fish is most likely to be found. The system draws on historical data collected between 2012 and 2016 and has been validated with more recent datasets from 2021 to 2023. According to INCOIS, the service is already operational.

At its core, the model decodes the complex environmental signals that influence hilsa movement—tracking variables such as water temperature, salinity, current speed and flow direction. “Hilsa doesn’t follow simple patterns. Traditional methods struggle to handle these complexities, but machine learning can recognise such changing relationships much better,” said Sandip Giri, the scientist leading the project.


The output is translated into digital maps that highlight hilsa-rich zones, offering fishers a data-backed guide to improve their catch efficiency. The advisories are designed with safeguards, including compliance with seasonal fishing bans and a 5-km coastal buffer zone to prevent overexploitation near shorelines.

To fine-tune the model, researchers deployed GPS-enabled boats in the Hooghly estuary, logging granular data on catch volumes, fishing duration and location using government-approved gill nets. The results, Giri noted, have shown “promising prediction accuracy”. The underlying research has been published in the journal Fisheries Oceanography.

Efforts are also underway to take the technology to the grassroots. Vidyasagar University has been conducting training and awareness programmes to familiarise small-scale fishers with the advisory system and related mobile applications. “The model factors in biogeochemical parameters such as water salinity, wind direction and speed, and temperature. We are currently testing its accuracy,” said Sourav Maity, a scientist at the university’s Coastal Observatory and Outreach Centre and a member of the project team.

Yet, beyond the laboratories and pilot projects, awareness remains limited—and scepticism is growing.

“I do not know about anything like this,” said Shyam Sundar Das, secretary of the Digha Fishermen and Fish Traders Union, reflecting the information gap at the ground level.

Others are more cautious about the implications. “We are not aware of any such move. But if it happens, I doubt how sustainable it will be to locate fish using AI, bypassing traditional knowledge. Targeting specific concentrations of hilsa could trigger overfishing, disrupt the ecological balance, and ultimately cause severe stock depletion,” warned Debashis Shyamal, general secretary of the Dakshinbanga Matsyajibi Forum.

Researchers, however, say outreach efforts are ongoing and will intensify. “We have already conducted awareness camps for fishermen in Digha and Kakdwip. And with the April-June hilsa fishing ban now in effect, we will organise more such camps, including in Fraserganj and Namkhana,” Maity said.

As India blends artificial intelligence with age-old fishing practices, the hilsa project sits at a delicate intersection—where technology promises efficiency, but sustainability may ultimately decide its success.