Artificial intelligence (AI) is gaining wide acceptance from global seafood-harvesting operations, but its adoption among seafood processors remains low, leaving these businesses with missed opportunities to make their processes more efficient and turn greater profits.
According to London, U.K.-based nonprofit Planet Tracker, the average pre-tax earnings margin for publicly listed seafood-processing companies is 3.4 percent – far below what financial experts consider a healthy benchmark of 10 percent.
Eric Enno Tamm, the CEO and co-founder of Vancouver, Canada-based seafood software company ThisFish, said he believes there are many opportunities for seafood-processing firms to adopt AI and improve upon that low figure – but the problem has been getting companies to embrace its potential.
“There is a lack of understanding of what AI is, what it can do, and what sort of problems it can solve in seafood manufacturing,” Tamm said. “AI needs to be trained on digital data that is clean, comprehensive, and connected. Even seafood companies that have digital systems often find that their data is error-prone, patchy, and disconnected by being stored in different formats in different software. AI needs good data to be accurate and useful.”
To ensure that AI adoption rates rise, ThisFish is exploring ways to effectively implement the technology in the sector.
For example, in collaboration with a Thai tuna cannery, ThisFish used AI to study raw material variables; such as fish species, size, and harvest methods; and process variables such as cold storage duration, thawing time, and temperature; to determine their impacts on yield.
The analysis determined that the primary variables affecting yields were fish size, species, time in cold storage, cooking temperature, and cooking time.
“Most production and cold storage managers know that the longer you store tuna, the quality will deteriorate,” Tamm said. “However, we were the first to actually quantify how cold storage duration impacts the yields, which means you can calculate the financial loss from storage. This insight can help procurement managers when they buy tuna, applying a discount on raw material that has been stored longer.”
With AI at a seafood processor’s disposal, Tamm said these businesses can make better-informed decisions and ensure that they are maximizing their potential.
“A yield-prediction model can help improve certainty in production,” Tamm said. “Yields can fluctuate a lot depending on fish size, cold storage duration, species, quality, and the production process. AI can help provide insights on why yields might be high or low. These insights can then be used in production planning.”
Thailand is not the only spot ThisFish has carried out AI cost-saving collaborations.
In 2023 and 2024, ThisFish partnered with two tuna canneries in Manta, Ecuador, to develop a drain weight prediction model.
Tuna canneries must ensure cans meet minimum drained weight requirements, which concern the weight of meat remaining after liquid is removed. Due to numerous factors affecting this weight, canneries often use basic drained weight tables that are prone to errors, leading to intentional overfilling to meet standards.
Preliminary results from ThisFish’s study suggest that the canneries had potential savings of up to 3 grams per can if measured properly.
Saving just 1 gram of skipjack per can, according to ThisFish, could reduce raw material costs by up to USD 500,000 (EUR 476,000) each year, totaling about USD 75 million to USD 225 million (EUR 71.5 million to EUR 214 million) each year globally for canneries.
AI can also help inspect products for quality at a fraction of the cost, according to Tamm.
For instance, Tamm said a tuna cannery producing 1 million cans a day will only be able to inspect a few hundred or thousand manually. AI would be able to inspect all of those 1 million cans effectively and take pictures for historical record.
“There is a lot of visual inspection in seafood processing, so we've deployed smart cameras to count salmon fillets and detect defects and classify their color,” Tamm said. “Seafood processors recognize that there is a lot of manual visual inspection, and it can be difficult to train workers. AI can perform these tasks exponentially better.”
In general, AI makes the seafood industry more predictable and profitable, according to Tamm, and there is growing market demand for responsible, traceable seafood in which transparency has become essential.
In this environment, Tamm advised companies to embrace digitization, prioritize efficiency, and invest in learning to remain competitive.
