3.03 AI-powered citizen science

The rise of digital citizen science

Citizen science, a collaborative research approach involving public volunteers, has been revolutionised by digital technology. The advent of smartphones and online platforms has democratised scientific data collection, enabling anyone with an internet connection to contribute to research projects. Now, a new evolution is underway: the integration of Artificial Intelligence, giving rise to “AI-Powered Citizen Science”. This synergy combines the vast scale of public participation with the formidable analytical power of AI, accelerating our ability to understand and address complex environmental issues.

AI enhances citizen science in several critical ways. First, it dramatically improves the efficiency of data analysis. For instance, AI-powered image classification algorithms can automatically identify species from photographs submitted by volunteers, a task that would otherwise require countless hours from human experts. Second, AI can improve the quality and reliability of the data collected. Algorithms can be designed to validate submissions, flag potential errors or anomalies, and ensure data consistency across large projects. Finally, AI can be used to create more engaging and effective experiences for the volunteers themselves. AI can use a deep learning algorithm to assess each volunteer’s skill level and create a personalised training pathway, which has been shown to significantly improve both classification accuracy and volunteer retention.

Platform showcase: Tools for people-powered research

Several platforms exemplify the power of AI-enhanced citizen science. Zooniverse stands as the world’s largest hub for people-powered research, hosting projects that span disciplines from astronomy to biodiversity monitoring and the transcription of historical documents. The platform’s philosophy is to use AI not to replace human intelligence but to augment it, often using volunteer classifications to train machine learning models that can then handle the more routine tasks, freeing up human volunteers for more complex or ambiguous cases.

iNaturalist operates as a global social network for biodiversity observers. Its core feature is a computer vision AI that suggests species identifications from photos uploaded by users. This AI model is trained on the vast, community-verified dataset of observations on the platform itself. When a user uploads a photo, the AI provides a list of likely species, weighted by visual similarity and the user’s geolocation. The user can accept a suggestion or make their own identification, which is then subject to verification by the wider community of experts and enthusiasts. This creates a virtuous cycle: community engagement generates the data needed to train a more accurate AI, and the AI, in turn, makes it easier and more rewarding for new users to engage, thereby generating more data.

This dynamic illustrates a powerful socio-technical feedback loop. The collective intelligence (CI) of the community builds the dataset that trains the artificial intelligence (AI). The AI then provides a technological scaffold that supports and expands the community’s collective intelligence. This self-improving system offers a scalable and effective model for community-driven environmental monitoring. For a community-based project, this model is directly applicable. A custom mobile application could be developed where community members along a river photograph and log observations of key indicator species – plants or animals whose presence or absence signals changes in the ecosystem’s health. An AI, trained on images verified by local biologists and knowledge holders, could provide initial identification suggestions, empowering local residents to become active stewards and data collectors for their own ecosystems.

The following table offers a comparative overview of key platforms, helping learners to identify potential tools for their own community-based environmental projects.

Table 1. AI-Powered Citizen Science Tools.