By Akshaiyaa V S
As plastic and other forms of wastes continue to penetrate nearly all the worlds water bodies, it will not be surprising if nearly half of the worlds population would have to live in water-stressed areas by 2025.
Planktons are the natural sensors of changes in the health of water bodies and play a major contribution in controlling the quality of air and water on earth. They also are the foundation of the oceanic food chain. However, studying planktons is still a challenge as it requires the collection of adequate samples and shipping them to the laboratory without damage after they are dead. Also, plankton populations have fallen by over 40 per cent since 1950 due to rise in global temperatures.
To circumvent these issues, IBM has introduced innovative Artificial Intelligence (AI) powered robotic microscopes networked in the cloud and deployed around the world to monitor planktons. By observing the behaviour of these planktons, the microscopic robots would help provide a better insight into the water and oxygen level in the water bodies.
How does the microscope work?
An autonomous robot microscope is placed inside the water body, helping to create 3D models of the swimming planktons. The microscope makes use of twin LED lights to capture the planktons shadows (called the shadowgraph) as they move about, on an imaging chip. This technology is the same as that of the cameras used in smartphones without the need for focusing on the image. These robots are the worlds smallest computers working with blockchain smaller than a grain of salt. Nevertheless, this robot has the computing power of the x86 chip from 1990. That might not amount to much as of now, but it also packs a thousand transistors which would help it to monitor and communicate on the datasets provided.
Planktons are extremely sensitive to even slight changes in water and once a digital sample of the planktons is created, the device would monitor their health from their size, shape and behaviour. This would indirectly monitor the health of the aquatic life and help monitor threats like oil spills and red tides.
For instance, a zooplankton would normally move in random directions to avoid being caught by predators. But when it gets affected by toxins in water, the zooplankton starts moving in straight lines, exploding the algae and sucking out all the oxygen out of the water while simultaneously releasing unhealthy amounts of toxins which would affect other organisms while decaying.
Similarly, a single-celled organism called a Stentor is usually trumpet-shaped, but it becomes a ball in high sugar concentrations. Thus, with the help of AI, studying the behaviour of such organisms would help study aquatic health.
IBM is not the first company to create such robots. The Flowcytobot is a similar robot created by the McLane Research Laboratories. This robot sticks itself to piers and sucks out a sample of water which then passes through a laser. The planktons in the water reflect light, creating an image which when studied alerts scientists to outbreak of toxins. The only disadvantage of the Flowcytobot is that it is immobile, but there is ongoing research to make it mobile without having to compromise on the functionalities of the robot.
Purging AI of bias
The problem with AI is that the performance of such machines is solely based on the type of data fed into them. It is important to ensure that human biases do not affect the algorithms fed into the systems. Researchers have developed a methodology to reduce any bias that may be present in the data sets but it might take another year or two to develop a proper sample testing system which would provide continuous monitoring of the microorganisms.
Identifying bias is essential to building trust between the AI systems and humans. When AI find and point out human bias in decision making, they eventually lead us to adopt more impartial views towards analyzing data sets. IBM’s goal is to create systems which can tackle bias even when training data is not available.
A global centre where such digital images can be archived and stored would help researchers work with the datasets more efficiently. Once studies with the local water bodies are successful, researchers plan to move on to oceans and subsequently to the Mariana trench, helping us trek further on the path to finding a solution to the problem of polluted water supplies, plaguing the entire planet.
Akshaiyaa V S is an analyst at Qrius