Participatory AI: For whom? For what?
On June 20th 2024, the IEAI hosted a virtual Speaker Series on Participatory AI. We were honored to have Daniel Gatica-Perez, Head of Social Computing Group at the Idiap Research Institute and Professor at EPFL School of Engineering and College of Humanities as our speaker to give us insights into his work on participatory AI.
Prof. Gatica-Perez highlighted that AI has been perceived by the general public quite differently compared to those working in the field. Moreover, he stated the US and the UK have dominated data-generation for AI systems, leading to the relative dominance of these two nations compared to other parts of the world. This highlights inequities in power relations in terms of geography, gender, race, etc. that the technology industry might exacerbate. Still, human experiences are what matter along the journey.
In his description of AI, Prof. Gatica-Perez reiterated the need for AI to embody key attributes such as trustworthy, participatory, equitable, and explainable. He acknowledged the current reality that we are still striving to achieve these ideals in AI and believes more serious efforts should be invested into researching them.
Within the domain of participatory AI, he referred to the undeniable social and human aspects of the AI pipeline. This has given rise to three questions: 1) What human forces shall we amplify?; 2) How can we accompany people in the loop?; and 3) Who decides and benefits from answering these questions and vice versa? When speaking of human factors in AI, Prof. Gatica-Perez summarised with three important facets to be distinguished: AI literacy, AI governance and participatory AI.
Next, Prof. Gatica-Perez presented three case studies on the deployment of AI systems using participatory research methods allowing many people to participate.
The first project was SenseCityVity, which was carried out in a city in Mexico with a large youth population (34% of the residents were under 18). The idea of SenseCityVity was to create a survey asking the youth about their perspectives towards the urban environment in their city and construct a computer vision based on their description. Ultimately, his team sought to analyze how local youth could be supported in engaging with documenting and reflecting upon their urban environment. Additionally, the data collected helped the team to understand what issues are the most sensitive to young people.
A second project, IcARUS, conducted across six institutions and universities in Europe looked into co-designing approaches to urban security with the participation of both authorities and citizens. In so doing, four areas of concentration were addressed: juvenile delinquency, radicalization, safe public spaces and organized crime. Prof. Gatica-Perez explained the details of the framework, underlying how people could be included by using design thinking methodology. He then presented an Evidence-Informed Participatory Governance (EIPG) model and Sbocciamo Torino tool – both of which encourage users to have more human interactions. Interestingly, the outcome in this case was a universal choice not to employ AI-enabled tools to achieve the solutions they were seeking in urban security, and thus illustrating how a participatory approach may lead to unexpected outcomes.
In the end, it is important to think that AI is a thing, an object. It’s nothing else than that.
The third case study focused on Generative AI Literacy. The project emerged after observations of the MyCity Chatbot in New York City found the chatbot to provide users with answers containing inaccurate content and encouragement to break the law. In response to this accusation, the city of New York decided to add a notice stating: “May occasionally produce incorrect, harmful or biased content.”. This example points to a larger problem of AI literacy. Here, Prof. Gatica-Perez argued a tailored approach has become more necessary due to legal and ethical reasons. He along with colleagues hence developed a GenAI Competencies Framework for GenAI literacy.
We thank Prof. Gatica-Perez for his insightful presentation on participatory AI. The recording of the event can be found here.
