Emma Lopez

Emma Lopez is a technologist with a focus on people and products.
Emma Lopez holds a Masters Degree in Computer Science. Throughout her career, she has participated in different roles focused on the development of technological products. She is an enthusiast of the agile principles and tries to bring transparency and continuous improvement to all the processes she is involved with. She is also committed to the promotion of local tech communities.

As Product Director at Eticas, she is in charge of improving the algorithmic audit methodology, as well as manage and participates in the ongoing audits.

A bottom-up approach to algorithmic auditing

Algorithmic audits conducted by independent third parties are emerging as one of the practices that can bring transparency to AI models. However, it is an uncharted landscape where multiple audit frameworks are emerging to try improve accountability, transparency and trustworthiness in AI models.

What most existing frameworks share is a top-down approach. At Eticas we have taken a different route. For the last two years we have been auditing algorithms with a few trusted clients in order to learn what works and what doesn’t. Our approach combines social science as well as computer science as the team performing them have different abilities. We have recently published a guide with our methodology and we would like to share it with the community for peer review.

During the session we will present the practical experience gained while conducting external audits as independent third parties of different models in areas like health, wellness and government social services, with algorithm types ranging from recommender to ranking algorithms, and from risk classifiers to NLP-based matching systems.

Co-presented with Dr Gemma Galdon.