CritML Workshop at CHItaly 2021
Critical Tools for Machine Learning: Figuring, fabulating, situating, diffracting machine learning systems design
This workshop draws on feminist and critical methodologies to construct interdisciplinary interventions in the design of machine learning systems, towards more inclusive, reflective and contextualized systems design. Tapping into feminist technoscience research, the project takes up concepts of “figuration”, “situating/situated knowledge”, “critical fabulation/speculation” and “diffraction” as theoretical tools that can help re-imagine machine learning systems design as a more contextualized, transdisciplinary process.
The purpose of the workshop is to experiment with how machine learning systems can be imagined and designed in a more situated, inclusive, contextualized and accountable way in order to reduce the systemic socio-cultural biases and develop more socially responsible frameworks of design. The premise of the work is that while computer science has developed sophisticated technical tools to improve machine learning accuracy and expand application fields, it is facing issues in particular with regards to systemic socio-cultural bias. Critical theories, particularly feminist and postcolonial critical theories, have developed tools to address societal bias and its embeddedness in systems of thought and technology, and to trace how these embeddings give rise to new and reproduce existing hierarchies of power in society.
This workshop thus aims to translate insights developed by those critical study fields into approaches in machine learning systems design through an experimental workshop. Concretely, the objectives of this workshops are: 1) to introduce critical methodologies as potential design intervention tools; 2) to generate creative ways to translate such methodologies into approaches to and tools for machine learning systems design.
The workshop will be divided into four main parts:
- critical concepts from feminist technoscience as well as the standard machine learning systems design chain will be introduced and discussed.
- Participants in smaller groups will chose one methodological concept to work with and will discuss concrete ways how this concept can be applied in machine learning systems design, relying on intervening into the more conventional machine learning systems design process flowchart as a basis for intervention.
- Participants will apply their critical concept as a methodological tool in a speculative machine learning systems design scenario.
- The results will be discussed collectively.
Workshop process will be documented on our research blog 'Engines of Difference' for critical computing.
Please note your interest in participating by filling this form. The workshop is open to CHItaly participants. However, registrations for single workshops are possible too.
Please follow the conference website for more information about registration.
There will be no collective publication following the workshop, as it is intended more as a hands-on activity. However, the process will be documented and published on the blog for critical computing.
This workshop is meant for computer scientists and social scientists who are particularly interested in critical computing and critical design methodologies and human-algorithm interaction as well as participatory algorithmic systems design.