Informal Embodied Auditing

Exploring Facial Emotion AI (FEAI) through Community Workshops

Activities: Critical AI Literacy Workshops, Qualitative Analysis

Year: 2025-26

As AI systems become more integrated into everyday life, they’re increasingly tasked with interpreting human behavior, including emotions. Emotion AI technologies analyze signals like facial expressions, vocal tones, and physiological data to predict emotional states, but their accuracy and impact are highly contested. Facial Recognition Emotion AI (FEAI) is a type of Emotion AI that infers emotions from facial expressions, and it is increasingly ethically controversial, motivating a need for greater public understanding and critique. In response, we developed Explore-FEAI, an FEAI model and accompanying interactive website that offers open-ended exploration with FEAI firsthand. We designed a workshop where participants learn about FEAI using Explore-FEAI and discuss societal implications, partnering with local organizations to host community workshops. Our findings analyze participants' growing critical AI literacy through exploring inputs and outputs, mechanistic reasoning, data critiques, sociocultural critiques, ethical concerns, and embodied and material exploration of FEAI. Our discussion offers informal embodied auditing as an approach for critical engagement with AI through embodied and material exploration, as well as reflections on informal auditing for supporting AI literacy, informal auditing for questioning EAI ethics, and expanding participation roles for more holistic EAI training. This project is a collaboration with Xingyu Li, Zhiming Dai, Noura Howell, and Decatur Makers. 

Publications

Critical AI Literacy Workshops

tangibly exploring Emotion AI

We ran a series of workshops with local community organizations where participants could learn how Facial Recognition Emotion AI (FEAI) works, where it's being used today, and the ethical and social challenges it raises. During workshops, participants could explore FEAI using our provided tangible materials (clay, paper, pencils, markets) and embodied practices. At the end of our workshops, participants were asked to imagine alternative futures: how should (or shouldn't) these technologies shape our world?

Pictured is a workshop we ran at the Public Art Futures Lab, where we explored Facial Recognition Emotion AI (FEAI) with participants by inviting them to use our web-based tool. Participants took or uploaded photos to play with the tool, developing critical literacy around FEAI. Additionally, we encouraged participants to use materials such as clay, pencils, and paper to tangibly explore how FEAI works. 

Informal Embodied Auditing

exploring tangible sense-making

In our workshops, participants used tangible materials, such as clay, paper masks, and hand-drawn emotional expressions to explore FEAI. To deepen AI literacy education and algorithm auditing, our workshops point to supporting participants in using their bodies and tangible materials to intuitively explore and evaluate FEAI. Doing so leverages participants' embodied ways of knowing, as well as sociocultural understandings, to support critical AI literacy and question AI ethics.

Here is an example of a participant using tangible materials to experiment with FEAI. This participant crafted facial features out of clay and tested them in various iterations. Here, they compared placing them on white and black backgrounds, with different results.

Participants also uploaded hand-drawn features and cartoons, such as the one pictured here, to test FEAI's mechanisms, inputs, and outputs.

Conclusion

informal embodied auditing for FEAI

Our workshop findings reflect on how informal embodied auditing can support AI literacy, encourage critical engagements with Emotion AI, and empower informed, embodied, judgements around controversial technologies.

We contribute the following to situate informal embodied auditing for FEAI:

1. We suggest supporting participants in using their bodies to intuitively test FEAI inputs, and leveraging their embodied ways of knowing to evaluate outputs.
2. We recommend providing tangible tools and materials that support open-ended
   exploration and creative inputs in simple, intuitive, flexible ways.
3. Informal embodied auditing ultimately invites leveraging more expansive and inclusive ways of knowing—embodied, tangible, and sociocultural—as sources of ‘ground truth’ for algorithm auditing.
4. We recommend supporting participants in collaboratively and tangibly comparing and contrasting various AI inputs and outputs.
5. Informal embodied auditing empowers participants as situated interpreters, not just data labelers, whose embodied experiments and critical reflections illuminate what FEAI is and what it ought (not) to become.
   

We share the remainder of our findings and recommendations in our publication.

Email   ariggs00@gmail.com

LinkedIn  atriggs

Google Scholar

ORCID