AI Can Help Find Plastic Pollution—But Can It Help Us Act?
Plastic pollution is easy to see on a beach—but much harder to measure, understand and prevent.
Artificial intelligence, or AI, can be used as a tool to collect and analyse information on plastic pollution. This can help guide positive action.'
Not all AI is the same
“AI” is often discussed as though it is one single technology. That can be misleading.
Much of the current debate focuses on generative AI: tools that produce text, images, audio or video. These systems are linked to concerns about large data centres, energy demand and environmental claims.
This article focuses on a different use of AI: machine learning for finding, classifying and understanding plastic pollution. These task-specific systems are trained to recognise patterns in photographs, sensor readings or laboratory data. They use substantially less energy and resources than generative AI; they are designed to perform a defined task rather than generate new text, images or video.
Machine learning can help identify litter in images, distinguish types of plastic during sorting, examine underwater debris and speed up microplastic analysis. For example, “BePLi Dataset v1: ” describes a dataset of more than 3,700 real-world coastal photographs with manually created annotations of visible plastic litter. A later update expands the types of litter that can be recognised, including bottles, bags, rope, fishing nets, buoys, foam and fragments.
Subscribe to our fortnightly newsletter
Get our top stories straight to your inbox every two weeks and stay informed about plastic pollution!
These tools do not replace action to reduce plastic production, improve waste services or hold polluters accountable. But they can help communities, researchers and public bodies collect and organise stronger evidence about where plastic pollution is happening.
From observation to action
AI-supported monitoring can turn community and research observations—such as litter photographs, underwater debris or collected samples—into useful evidence when records include location, time, source, setting and the type or amount of plastic. Machine learning can speed up image review, litter classification and pattern detection, but people must verify significant or uncertain findings. Traceable, confidence-rated evidence can then guide clean-ups, pollution investigations, better waste services and follow-up monitoring.
Empowering citizen scientists: A web app for plastic litter classification describes an AI-enabled tool designed to help citizen scientists classify litter observations.
AI is only one part
Machine learning can speed up plastic-pollution monitoring and analysis, (see references below) from citizen reports and beach surveys to recycling and microplastics testing, but it should support rather than replace human judgement. Trusted results require traceability to the original observation, human review of important or uncertain findings, and translation into practical action.
The real question
Seek reliable AI evidence that supports action against plastic pollution. When you see an AI-supported pollution claim, map or report, ask:
Where did the information come from?
Has it been checked by people?
Can the evidence be traced to the original observation?
What action will follow?
+++++++++++++++++
For more information contact: info@scarabtrust.org.uk
Images:
plastic pollution Accra beach - Image by Muntaka Chasant (ST ref: 1404)
Graphics - BFFP
ID: 240305-A
Comments