Data is fundamental to automated decision-making, and sits at the heart of its ethical and practical challenges. These range from well-known problems with bias and privacy to complexities of data ownership and flows across devices and national borders.

Examples include:

  • An algorithm used to assign school grades during COVID-19 is accused of
    bias against students from poorer backgrounds.
  • Driverless vehicles are unable to ‘see’ and avoid kangaroos because they
    were designed in Europe and trained with data about animals like moose.
  • Photos people posted online for social purposes have been collected by a
    company that markets facial recognition technology to the police.
  • As more workers telecommute due to the Covid-19 pandemic, new
    applications are being developed for detailed monitoring of remote
    employees. The new forms of data that are collected can help remote work
    become more efficient, but they can also result in wage theft and diminished
    autonomy in the workplace.

Whilst there are new Centres and Institutes emerging around the world to advance ethical data-driven AI and data governance, the ADM+S Centre tackles these challenges at an unprecedented scale of cross-disciplinary and cross-institutional collaboration.

The Data research program brings together data science with social science to find responsible, ethical and inclusive ways of constructing, sharing and using data to solve problems and automate decision-making systems; and uses critical data studies approaches to critically examine the data logics, infrastructures and flows that sit behind ADM systems.

This group will deploy new digital tools and methods to improve awareness of data-related issues and involve the public in research, as well as working closely with our partners to collaboratively develop and embed responsible, inclusive and ethical data practices in industry, government and community settings, leading to more effective automated decision-making with fairer outcomes.

RESEARCH PROJECTS

Political Economy of Sex Tech

Examining the political economy of ‘sex tech’ in order to identify how sexual technologies are being governed at scale, how sexual data is being collected, stored, shared and monetised, and how the material benefits of sex tech may be more equitably distributed.

Automating safety: developing better data models to help foster prosocial platforms

Indentifying how misunderstandings of harm and safety flow into flawed data logics and ineffective automated digital platform responses.

Automated Content Regulation (disinformation and political bias)

Evaluating the moderation of social media content, which has become radically more reliant on machine learning classifiers during the Covid-19 pandemic.

Responsible health consumer data analysis and ADM

Creating a replicable framework for building capacity (expertise, literacy, data partnerships and data governance) to unlock the social value and impact of advanced data analytics, AI and ADM across the not-for-profit sector.

Mapping the Public Conversation on ADM

Maping the extent, qualities and diversity of conversations about automation and ADM in Australia and globally, using key social media data sources including Twitter and Facebook to collect posts and a wide range of media articles.

Hybrid digital methods for detecting and managing problematic automated agents in social media

Operationalising new data partnerships and implementing data analysis to improve non-profit and humanitarian sector work.

Quantifying and Measuring Bias and Engagement

Recent developments in machine learning and information access communities attempt to define fairness-aware metrics to incorporate into these frameworks. This project will address a number of research questions related to quantifying and measuring bias and engagement that remain unexplored.

The Coronavirus Impact

Considering ethical approaches in the area of automated decision-making (ADM) and civic life with a focus on civic commitments and concerns.

When All Data is Health Data

Considering the ethical issues raised by new streams of health data, including how best to regulate the use of the data, its storage, and the infrastructures that collect it.

Data mapping and ADM to advance humanitarian action and preparedness

Operationalising new data partnerships and implementing data analysis to improve non-profit and humanitarian sector work.

Everyday Data Cultures

Exploring the role of everyday data practices and literacies in automated decision-making.

Mapping ADM Across Sectors

Examining common themes with respect to the issues raised by the collection, storage, and use of data for ADM across object domains.

Data Ethics, Rights, and Markets

Considering ethical approaches in the area of automated decision-making (ADM) and civic life with a focus on civic commitments and concerns.

Automated Content Regulation (Sexuality Education and Health Information)

How might platform regulators and moderators better distinguish between education and information content, and other forms of sexual texts and imagery on social media platforms?

Data capacity and collaboration for ADM in the community sector

Creating a replicable framework for building capacity (expertise, literacy, data partnerships and data governance) to unlock the social value and impact of advanced data analytics, AI and ADM across the not-for-profit sector.

Dark Ads Transparency Project

Providing strategies to address the potential harms posed by ‘dark ads’ and provide accountability and transparency mechanisms for targeted advertising.

Civic Automated Decision-Making

Considering ethical approaches in the area of automated decision-making (ADM) and civic life with a focus on civic commitments and concerns.

Infrastructures, ADM and sovereign capability

This project examines the relationship between telecommunications infrastructure and automated decision-making (ADM) as new infrastructures, such as 5G and Wi-Fi 6, support the intensification of ADM.

Assessing the Personalisation of Search Results from Major Recommendation Engines

This research aims to assess the extent to which search results are personalised, by various leading search engines and their algorithms, based on the profiles established by those search engines for their different users.