The Project | Team | The Study | Results | Partners | Contact Â
Participatory Identification of Biases in Artificial Intelligence (AI-Bias)
Nowadays, and often without realizing it, many of the decisions we make in our daily lives are assistedâand even improvedâby Artificial Intelligence (AI) algorithms. These tools, sometimes misunderstood by the public, make our lives easier in different areas, as they are present in internet search engines, social media, dating apps, GPS systems, self-driving cars, and even in medical advances.
However, we must not forget that Artificial Intelligence is designed by humans and, therefore, reflects their unconscious biasesâtendencies or inclinations that give rise to stereotypes and discrimination within society. For this reason, we must remain critical and attentive to such issues in order to detect them, since, if discrimination is assumed to be the general norm in our society, systems could instead be designed to promote notions of equality (Broussard, 2018).
CONTEXT
AI-Bias addresses the challenge of strengthening citizensâ knowledge and training capacities to better identify and report biases through education in Artificial Intelligence, engaging the public directly in scientific production.
Artificial Intelligence has enormous potential to drive social growth across different sectors of society, due to its increasing presence in everyday activities and toolsâfrom internet search engines and personal assistants on mobile phones to professional applications such as the design of public health campaigns or disease treatment, thanks to its predictive capabilities.
While AI is a promising tool with the ability to foster social progress, it also has certain shortcomings and challenges resulting from algorithmic bias. According to the Royal Spanish Academy, a bias refers to an inclination or deviation. Macchiavelli (2021) notes that the use of the word âbiasâ is symbolic, referring to a tendency or leaning toward something. It is worth noting that UNESCO (2020) assigns a negative connotation to algorithmic bias, stating that such biases have the potential to reinforce or spread stereotypesâfor example, gender-based ones.
Within this context, two main concerns are identified:
(i) The need to understand citizensâ experiences with AI-based tools and platforms, and their ability to identify possible biases when using these technologies.
(ii) The need to provide citizens with training that helps them address the implications, applications, and ethical challenges of AI in their daily lives.
WHAT IT CONSISTS OF
The main idea behind the AI-Bias methodology is to focus on two fundamental and complementary actions.
The first action seeks to understand the perceptions and knowledge of Spanish citizens regarding Artificial Intelligence biases, in order to implement the second action, which aims to equip the Spanish population with tools to detect biases present in AI systems through citizen science.
Methodology
The project will be carried out using mixed research methods, achieving a comprehensive and multidimensional approach to adequately examine citizensâ perceptions and understanding of Artificial Intelligence biases from multiple perspectives.
The ultimate goal is to conduct a final activity in which participants identify biasesâmainly related to gender, race, and religionâin some open-access Artificial Intelligence platforms, after having received training in an AI workshop.
Three main methodological approaches are used:
 |
 |
 |
| Surveys & Focus Groups | Citizen science | Open online workshops |
OBJECTIVES
The main objective of the IA-Bias project is to provide citizens with basic knowledge of Artificial Intelligence so that they can actively participate in science from different social spheres and help identify biases in AI systems.
Specific objectives include:
O1. To identify citizensâ experiences with AI tools and platforms, as well as their perception and level of knowledge regarding the identification of possible biases encountered by Spaniards when using AI tools.
O2. To provide citizens with training that helps them address the implications, applications, and ethical challenges of Artificial Intelligence in their daily lives, prioritizing the education of citizens in AI bias detection. This objective also seeks to foster citizen participation in the scientific process by encouraging them to contribute data through the detection of biases in an AI platform.
INNOVACIĂN Y RETOS
The main innovation of the project lies in its effort to support and align with several Sustainable Development Goals (SDGs) of the United Nations 2030 Agenda, and to bring these closer to citizens through its two main actions. The project primarily focuses on strengthening citizensâ knowledge and training capacities to improve the identification and reporting of biases through education.
This focus is directly linked to Strategic Axis 2 of Spainâs National Artificial Intelligence Strategy:
âPromote the development of digital skills, enhance national talent, and attract global talent. It is essential to raise the level of technical AI training among the active populationâboth general users and specialistsâin order to facilitate access to new quality jobs and face the challenges of the future labor market.â
One of the most significant innovations compared to previous studies is the active involvement of citizens in scientific production, promoting a participatory and inclusive approach to AI research.
The project is structured around two main actions, employing both qualitative methodology (focus groups) and quantitative methodology (survey) to generate robust, mixed-methods research.
Another key innovation of this project is the inclusion of materials adapted to diversity, especially for people with visual or intellectual disabilities, through audiovisual pieces with simultaneous sign language translation and easy-to-read educational materials that follow accessibility and inclusion principles.
With the collaboration of:
Project funded by FundaciĂłn Española para la Ciencia y la TecnologĂa (FECYT) in the frame of “Convocatoria de ayudas para el fomento de la cultura cientĂfica, tecnolĂłgica y de la innovaciĂłn 2023-2024” [FCT-23-19454]