Scientists put algae to work making fuel. AI keeps watch.

Northeastern University researchers led by Miguel Fuentes-Cabrera are using AI to monitor individual algae cells in biomanufacturing tanks, aiming to improve biofuel production efficiency. The AI analyzes images from the Autonomous Real-Time Microbial Scope (ARTiMiS) to detect cell variations that affect lipid output, addressing the challenge of inconsistent yields when scaling from lab flasks to large tanks.

Scientists put algae to work making fuel. AI reveals which cells are pulling their weight. Not all algae cells behave alike. Miguel Fuentes-Cabrera and his team are using AI to spot the differences and improve biomanufacturing. Algae have a habit of turning up where they’re not wanted. They coat your swimming pool in green gunk, cling to rocks at the beach and even make headlines when they take over prominent reservoirs, such as the White House Reflection Pool https://news.northeastern.edu/2026/06/25/algae-reflecting-pool-explainer/ . But put them to work in the right setting, and algae can be surprisingly useful. In fact, these plant-like organisms may hold the key to sustainable fuels https://news.northeastern.edu/2023/12/12/environmental-data-science-alfond-scholar/ of the future. Algae use sunlight and carbon dioxide to perform photosynthesis, generating the energy and building blocks they need to grow. As part of that process, they naturally make lipids — fats and oils they stash away for energy. Scientists say they can harvest these oils and use them as biofuels — fuels made by living organisms — through a process known as biomanufacturing. Many species of algae ramp up lipid production in times of stress, such as when they are deprived of nitrogen, an important nutrient. By inducing such stressful situations, researchers can get algae to stockpile more lipids. To make large quantities of lipids, scientists need to grow enormous numbers of cells. That means moving production from laboratory flasks into tanks that can hold hundreds or even thousands of gallons of water teeming with algae. Even a smaller 260-gallon tank holds about as much as five bathtubs. Turning algae into living factories that churn out lipids is no easy feat, Northeastern University Khoury College of Computer Sciences professor Miguel Fuentes-Cabrera https://www.khoury.northeastern.edu/people/miguel-fuentes-cabrera/ said. Together with his colleagues, he has been searching for ways to make the process more efficient. Scientists try to maximize output by genetically engineering high-performing strains. They also fine-tune temperature, light, nutrients, acidity and oxygen levels to optimize conditions for growth. However, even genetically identical cells growing in the same environment can veer off in different directions, Fuentes-Cabrera explained. “You assume that all of them behave in the same manner, but they actually don’t,” he told Northeastern Global News. Some may be highly productive, while others might get sluggish due to quirks of their metabolism. Variation adds up to inconsistent output — especially as you scale up production, Fuentes-Cabrera said. A process running “beautifully … in a little flask in your lab” often falters once you move to a huge tank, where tiny differences are magnified. Meanwhile, conventional monitoring treats the entire colony as a single unit, making small glitches hard to spot. Making biomanufacturing more efficient starts with understanding exactly what’s happening inside the tank. But how do you keep tabs on individual algae cells — the source of the trouble? Fuentes-Cabrera and his colleagues are bringing AI on board. Their approach hinges on a tiny apparatus called the Autonomous Real-Time Microbial Scope, or ARTiMiS. It’s an imaging platform continuously photographs cells as they pass through it. AI then scans those images for differences in size, shape and texture, spotting any oddballs in the mix. The AI translates each cell image into a mathematical description of its distinctive features. It can then easily compare thousands of cells to spot ones that stand out from the crowd. Some variation will occur randomly, Fuentes-Cabrera said. However, other differences might arise from environmental conditions, such as temperature, light or nutrient levels. Fuentes-Cabrera referred to these controllable factors as “knobs” that scientists can manipulate to steer the culture toward the outcome they want. Editor’s Picks He explained that once scientists understand how each “knob” affects the cells, they can adjust them — for example, by changing temperature, light or nutrients. Doing so will keep the cells behaving more consistently and optimize production. The proposal earned his team recognition from the Genesis Mission, an initiative launched by the U.S. Department of Energy to accelerate scientific discovery. “It’s an effort by the government to marry AI with science,” Fuentes-Cabrera said. Now, the researchers have to show that their approach actually works. They plan to trigger lipid accumulation in several species of algae. Continuously photographing the cells, they’ll test whether the AI can detect meaningful differences among them. They will eventually compare tanks controlled using information from ARTiMiS and AI with those relying on conventional monitoring methods, like measurements of temperature, acidity and oxygen that track conditions in the tank as a whole without providing a window into individual cells. If the approach succeeds, Fuentes-Cabrera said, the same principle could eventually extend well beyond algae and biofuels to enhance other types of biomanufacturing. Greater efficiency could make lab-grown materials cheaper and expand what companies can make with them, biomanufacturing expert Paul Hill told Northeastern Global News. Hill is the principal and founder of Berkeley BioProcess https://berkeleybioprocess.com/ , a consulting firm that guides organizations from lab phases to commercial production. “With substantial improvements in biomanufacturing science over the past decades, many more and new types of materials can be made,” he said, adding that scientists will be able “to develop new and potentially better products from these natural building blocks.”