Humans or AI? For Fisheries Data, The Future is Both

In recent years, the international discourse surrounding Artificial Intelligence (AI) has nearly reached its boiling point.

Questions have been raised on generative AI art and its issues of legality, theft, and negatively impacts on artists. Schools have voiced their concern on students ‘cheating’ their way through education using AI. Most recently, in the ongoing Hollywood strike, writers and actors are protesting for fair wages, better residuals from streaming, and, among other things, against the use of AI for creating scripts and actors’ likeness.

These discourses often pit humans and AI against each other. One extreme side claims that AI is the future, and the humans who fail to adapt are ‘losers’ in the game. The other thinks that AI will advance capitalist greed, creating a future where art loses its humanity and workers lose their jobs in favor of corporate profit.

As AI continues to disrupt various fields, the debate on whether humans and AI can coexist does not seem to reach its conclusion just yet.

So, what about the fisheries sector?

How Data Deficiency Hurts Fisheries Management

The best fisheries management effort is one that champions the interests of both the people and the environment. This is achieved, firstly, by knowing the said interests, which can be seen reflected through accurate and representative data.?

This is the main point of the science-policy nexus—how to best ‘marry’ the two very different worlds of science and politics. By crafting environmental policies based not only on stakeholder interests but also on the recommendations from scientific data, sustainable governance is believed to be achievable.

Before that, we need good data.

Currently, fisheries data in Indonesia is most-commonly collected manually or semi-manually through field surveys (e.g. port sampling), logbook catch reports, and academic research.?

Judging from their effectiveness, these methods tend to be costly, prone to limitations, and time-consuming. Furthermore, Indonesia is the largest archipelagic country in the world, with over 17.000 islands and more than 2 million fishers, making data collection incredibly difficult due to the sheer extent of it.

The collected data, though it provides valuable insight, are often still not fully representative because of the constraints on scale, accuracy, and timeliness. Additionally, data is scattered across regions and institutions. The data’s unstandardized nature also makes it a challenge for stakeholders to compile and analyze.

It is not difficult to see how this problem domino-effects its way into fisheries management. Unrepresentative data can give birth to unsuitable policies that potentially stall the realization of sustainable development, even harming the fish stock, ecosystems, and the communities that depend on them. In the worst case scenario, the programs may not even reach their intended targets, resulting in a futile use of budget and resources.

So, what is this ‘Humans or AI’ clickbaity title about?

The need for science-based ocean governance creates a push for better data. The word ‘better’ not only refers to quality, but also quantity. And the data is not limited only to field ground truth. Thanks to the success of the internet, the world has now entered the era of Big Data, focusing on the massive flow of information in the digital form. The fisheries sector is no exception.

To answer the aforementioned challenges and limitations, the Fisheries Resource Center of Indonesia (FRCI) has executed two different initiatives for alternative fisheries data collection. Interestingly enough, these programs just happen to sit on the polar opposite of the ‘Human vs AI’ spectrum.

Field Data Collection Using Citizen Science

The term ‘citizen science’ was coined separately by Alan Irwin, a British sociologist, and Rick Bonney, an American ornithologist, in the mid 1990’s. Now, the term is most commonly understood as the participation of the general public in scientific research—a democratization of research, if you will.

Citizen science and collaborative actions are considered a viable option to address the issues of field data availability, both in quality and quantity, to inform fisheries management.

Launched in 2019, our citizen science initiative is called IKAN, which is an abbreviation of “Inisiatif Kolaborasi Pendataan Perikanan” in Bahasa Indonesia or the “Fisheries Data Collaboration Initiative” in English. This initiative was born through a collaboration between FRCI and the Department of Fisheries Resource Utilization, IPB University.

As the name implies, IKAN is a fisheries data collection method that involves multiple parties. This citizen science program is implemented through a mobile application platform, while still prioritizing data quality according to the data collection standards set by the Ministry of Marine Affairs and Fisheries. The IKAN application has also been registered with patent number 000439998.

To participate, users only need to register themselves, then the app is ready to go. The inputted data include trips, vessel, operation, and catch. Users can also upload photos of their catch, as well as the species and length.

So far, IKAN has been tried out in multiple fish landing spots in Indonesia. The collected data covers various fishery commodities, namely demersal fish, reef fish, and Elasmobranch (sharks and rays).

Throughout its initial run, IKAN has shown promising results. Other than providing an alternative scheme that is both cheaper and wider in scope than the conventional field data collection, IKAN is participatory in nature, meaning it can accommodate inputs in large quantities. At the same time, it can also be used for campaigning for sustainable fisheries management.

IKAN would not be possible without humans, since at its heart, citizen science is about people.

IKAN was created as an effort to build a stronger and wider network with various individuals, groups, organizations, and institutions to realize sustainability and justice for Indonesian fisheries. Thus, anyone who is interested in undertaking data collection is welcome to join.

Currently, the most active users of IKAN are ship captains and our field enumerators, but training programs are underway for students and the Marine and Fisheries Agency officials.

Of course, there are challenges in implementing fisheries citizen science initiatives. For IKAN’s case, community and stakeholders participation is still limited due to difficulties in dissemination. Also, we notice the need for capacity building for IKAN contributors to standardize data input.

In the future, we plan to implement IKAN in more locations. This can be achieved through cooperating with multiple parties from various walks of life—fishers, communities, students, academics, researchers, and the private sector.

Internet Big Data Collection Using Crowd Data Crawling (CDC)

Crowd Data Crawling (CDC) is a method we recently developed with the purpose of crawling and scraping big data from all across the world wide web.?

The CDC is powered by python and Auto-GPT—sometimes lovingly referred to as ‘ChatGPT on steroids’—to monitor and record fisheries data from the nooks and crannies of the? internet with pre-defined keyword and data types.

As an AI, Auto-GPT can work autonomously, performing tasks without the need of constant human input. The AI can also break down tasks by itself, making the workflow more streamlined.

When run, CDC has the capability to collect, classify, and index real-time, unlimited data from various websites. Not only that, CDC can also explore and process the data and run calculations all the way to data visualization and future predictions.

In a preliminary CDC implementation, we have tasked the system to explore the trade of tuna products in selected Indonesian marketplaces. Using relevant keywords (i.e. tuna filets, fresh tuna, tuna loin, canned tuna), the CDC managed to determine the price range and average price of various tuna products.?

We also tested the CDC to monitor the real-time tuna trade in selected stores. This resulted in a time series analysis of the number of tuna sold, of which product, and at what time of day (morning, noon, night).

As proven by the CDC, AI can automate tedious labor by collecting and processing fisheries big data in a fraction of time.?

In the future, there is massive potential in using AI in various niches of fisheries, such as processing weather and oceanographic data up to predicting future fish stocks.

In Conclusion…

Long-term environmental sustainability can only be achieved through good fisheries management. For that, we need science-based decisions supported by good data.

Here, collaboration is key. Fisheries data collection and analysis are the most effective when they involve all the resources at our disposal, including relevant stakeholders and available technology to complement each other.

When all is said and done, the one most important question left is whether humans and the environment can coexist.

Hopefully, the answer is yes.

Acknowledgement: CDC is funded by Kedaireka matching fund from the Directorate General of Higher Education, Ministry of Education and Culture (2023)

Authors: Agavia Kori Rahayu , Mohamad Natsir