The technological cooperative specialized in artificial intelligence El Garbell has presented the 'FemFrame' project, an initiative that uses the gender perspective as a practical tool to detect errors in technology. The aim is to avoid structural errors that end up costing entities, institutions and companies time, money and credibility. In this sense, the 'FemFrame' project team states that there are inevitable errors and avoidable errors, which can occur in the process of creating a product such as, for example, virtual reality glasses that must be redesigned because the interpupillary distance has been based on male measurements and many women do not see well in them, personnel selection algorithms that must be deactivated because they have been trained with biased data and automatically discriminate or health applications that require urgent updates that have been developed without taking into account hormonal patterns.
In order to address these problems and many other foreseeable situations, the 'FemFrame' project team studies how decisions are made in each phase of technological development to detect where biases that can generate errors are introduced, develops a tool based on generative artificial intelligence that identifies and diagnoses these biases and offers concrete improvements and specific recommendations and accompanies companies, institutions and entities so that they can integrate this methodology in a structural and scalable way through advice, support, training and guides. Using open source and with criteria of ethics, inclusion and sustainability, the initiative places a total of five key and strategic phases: training and initial research, experimentation in real cases, creation of practical materials, development of the tool based on artificial intelligence and activation of a collaborative ecosystem.
For more information, you can consult the official website of the 'FemFrame' project.