Breadcrumb

Citizen science and AI in the spotlight at NASA

AI in citizen science can support and improve volunteer efforts in image classification, data analysis, and problem-solving. NASA Cit Sci Leader Series is organizing a webinar on this topic.

Image
Woman with green jacket and arms crossed outdoors
Associate Professor Marisa Ponti
Photo: Agnes Ekstrand

On February 6, 2025, a webinar* about the use of artificial intelligence in citizen science will be arranged by NASA Cit Sci Leader Series. The webinar will be recorded and available to listen to afterwards. Associate Professor Marisa Ponti is one of the researchers invited to speak at the webinar.

How come you were you invited to speak at this event?

I was invited because I co-edited a special collection on the future of artificial intelligence and citizen science that was released at the end of 2024 in the journal Citizen Science Theory and Practice*, a journal that is an integral part of the Association for Advancing Participatory Sciences* in the USA. My fellow co-editors are: Lucy Fortson, an astrophysicist working at the University of Minnesota and founding member of the Zooniverse project*; Kevin Crowston, a Syracuse University scientist working on human-machine collaboration; Laure Kloetzer, a learning scientist at the University of Neuchâtel, and myself, who have been studying the use of technologies in citizen science for some time now (I began with the use of serious games and now AI technologies). There have been a couple of articles I published in the last few years about human-machine integration in citizen science and the distribution of tasks between volunteers, experts, and algorithms, and how their epistemic agency influenced the distribution of tasks. This is probably why I was invited to co-edit the special collection in the first place. 

Perhaps you  are wondering why NASA is interested in citizen science. Astrophysics and astronomy have been incubators for exploring how humans and machines can work together for some time now. Additionally, volunteer amateur astronomers have traditionally been involved in these fields for a long time.

What topics will be covered?

During the NASA Cit Sci webinar, we will talk about how new research techniques powered by artificial intelligence (AI) could complement or enhance human efforts in citizen science/crowdscience/participatory science projects. Among the topics covered in the special collection there are early achievements of artificial intelligence, remaining challenges to its use, ethical implications, and ways to collaborate between citizen science and artificial intelligence in the future.  

What types of projects and tasks within citizen science do you think AI is particularly good at helping with?

We can see from the papers included in our collection that AI technologies, mainly machine learning (ML), are being used for automating data analysis, where algorithms can analyze large datasets from citizen science projects, identifying patterns and anomalies much more quickly than humans. Machine learning is also used to improve data quality, because AI can help check the accuracy of data collected by citizen scientists and identify potential errors. Let me give you some examples of AI in citizen science projects from the studies included in the special collection:

  • Identifying animal species in camera trap images: ML has been used to automatically identify the species of animals captured in camera trap images.
  • Classifying galaxies: AI has been used to classify galaxies based on their shape, helping astronomers understand how galaxies form and evolve.  
  • Detecting classes of glitches: AI has been used to sift through large amounts of data from gravitational wave detectors, identifying potential signals that might indicate that a glitch belongs to a known class.  

Scientifically, the papers in our collection are concentrated in ecology and biodiversity, astronomy and astrophysics, and biomedicine (one paper).

What do you see as the greatest advantages of integrating AI into citizen science projects?

At present, I see the main advantage of augmenting volunteer efforts by using AI technologies to perform tasks such as image classification, data analysis, and problem-solving.

Is there anything one should be cautious about when it comes to integrating AI into citizen science?

As guest editors, we argue that CS projects must strive for ethical ("moral AI") use of AI through transparency, more democratic, and trustworthy processes. To achieve this goal, volunteer participants and project organizers must overcome several challenges. One of them is the fear that AI systems could reduce engagement due to task simplification, or replace humans in performing certain tasks. We listed several challenges in our editorial*.

The Global South also needs special attention because of its disadvantaged positions in AI development and the threat of "data colonialism".  It will take more empirical studies to understand how AI can be implemented effectively and equitably. 

Another challenge is that AI technology is developing extraordinarily quickly. Constraints such as limited resources, expertise, and infrastructure can make employing cutting-edge technology impractical. However, since CS facilitates human validation of ML, the most sophisticated models may be unnecessary, which can be particularly advantageous for groups lacking computational resources. Using smaller or older models may also be more environmentally friendly.

Do you think the balance between human and AI-based work in citizen science will shift in the future? How?

It's a good question. I always find it hard to respond, but let's see what happens with generative AI. The papers in this collection do not report the use of generative AI. In addition to designing tasks that are enjoyable for volunteers to complete, it is important to enhance the results of automated approaches. Finding roles that citizen scientists can play meaningfully alongside experts and AI technologies is essential. Concerns have been raised over the potential of AI to disengage citizen scientists. For instance, the use of AI can reduce the range of possible volunteer contributions or make their tasks either too simple or too complex. Delegating interactions to AI might be efficient for science teams but distancing for volunteers.

By the way, our Faculty of Science and Technology has a great citizen science project that integrates machine learning: Matthias Obst, researcher in the Department of Marine Sciences, has set up the Koster seafloor observatory* years ago to study the impact of climate change and human activity on Swedish marine ecosystems.

Links related to the article*