What PAIR is reading — April 2021
“What PAIR is reading” is a reading list series compiled by PAIR team members on a rotating basis.
Meet the author
Ned Cooper | Research Associate
I worked with PAIR on a research project to understand how community-based participatory research (CBPR) and its philosophy might be incorporated into the machine learning development process. CBPR is an approach to research that involves community members, organizational representatives, and academic researchers in all aspects of the research process. Something I’m familiar with, as I’m also a researcher with the 3A Institute at the Australian National University, where I explore questions related to artificial intelligence and cyber-physical systems.
Outside of work, I’m often trying to learn a new language or going on a hike. I’d also like to say I surf, but that would be overstating the slipping and falling I do in the water.
Here are a few things I’m reading — from papers and books on how CBPR can affect social change to captivating pieces about Wikipedia and what I think is the most significant document recently published in Australia.
What I’m reading
Barbara Israel, Amy Shultz, Edith Parker and Adam Becker: Review of community-based research: assessing partnership approaches to improve public health
If you’re interested in how to conduct CBPR, this paper outlines key principles of the approach. The paper is focused on public health, however the principles also have the potential to guide engagement with community members during the ML development process.
Peter Hovmand: Community-Based System Dynamics
There’s a persistent challenge to translate general discussions about community problems into components that may be ingested into technical development processes. In this book, Hovmand outlines one approach to consider — building systems dynamics models with community members.
ORES: Lowering Barriers with Participatory Machine Learning in Wikipedia
This paper explains how Wikipedia engaged a set of volunteer developers to participate in developing its algorithmic scoring service ‘ORES’ for content moderation, including developing multiple independent classifiers trained on different datasets. While the community the Wikipedia Foundation engaged with is quite technically savvy, it’s an intriguing example of the participatory design and evaluation of an ML system.
WeBuildAI: Participatory Framework for Algorithmic Governance
I think this paper demonstrates the potential for designers and developers to construct interfaces to allow community members to make preferences/trade-offs clear at key stages of ML development, without requiring community members to be too deep in the ‘tools’.
You’ll learn about a framework called ‘WeBuildAI’ which enables stakeholders to construct a computational model that represents their views, and apply the model to determine trade-offs for algorithms that govern a product or service.
Uluru Statement from the Heart
This is perhaps the most significant document to be published in Australia in the last 25 years. The statement calls for a First Nations Voice in the Australian Constitution that would empower Aboriginal and Torres Strait Islander people. I find the demand for sovereignty and clear articulation of the process for change very inspiring.
