Lesson 11: Quest for Social Justice in AI

Lesson 11: Quest for Social Justice

• The lesson explores social justice concerns in the age of AI.
• It examines the interplay between substantive and procedural justice in the real world, both technically and socially.
• The lesson identifies mainstream ways social justice and AI are understood and what people do about them.
• It highlights areas where these understandings are lacking.

Social Justice Issues in the Age of AI

• Criminal justice, socio-economic justice, racial justice, labour justice, environmental justice, and historical justice are all relevant concerns.

Examples: Biased Decision-Making

• Public services: Economic mobility and access to key public services are impacted.
• Case example: Dutch SyRI case: The SyRI risk profiling system, used by the Dutch government, is believed to have unfairly targeted low-income neighbourhoods based on socio-economic background. The system was deemed unlawful for violating privacy and equality rights.

Examples

• Criminal justice system, predictive policing: systems like DELIA in Italy use ethnicity to predict future criminality, targeting areas with racialised populations. The Dutch “Top 600” list similarly attempts to predict future criminality, with high rates of harassment.

How is Inequity Engineered?

• Bias is inherent in AI systems.
• This can be introduced purposefully, inadvertently, or emerge during application.
• Sources of bias include: statistical and computational biases, systemic bias, and human bias.

Types of AI bias

• Bias can be introduced, intended or unintended.

• Statistical and computational bias is found in the applications.

• Systemic and human biases are a frequent problem.

Theorizing Social Justice in the Context of AI Systems

• Principles of justice include basic liberties, equality, and freedom of thought.
• Social and economic inequalities should satisfy fair equality of opportunity, difference principle and more.
• It's important to consider the means as well as the ends when designing AI systems.

Theorizing Social Justice...

• Distributive justice and how institutions allocate resources.
• Social and technical elements interact dynamically to shape stable institutional practices.

Examples (continued)

• Workplace monitoring through AI systems: EU context requires employers to inform employees about the use of HR policies.
• Environmental justice is also a concern.

How are these social justice issues being addressed?

• This section asks about the methods for tackling identified social issues arising from and impacted by AI.

Approaches to address social justice issues.

• Legislation and regulatory instruments, adjudication (lawsuits), technical fixes, and design justice.

Legislation & Regulatory Instruments

• This section describes Al Act, DSA, DMA, GDPR, PLD, AILD, UNGPs, and mHRDD. Using these regulations to mitigate harms and protect human rights.

Adjudication

• Lawsuits are based on laws such as GDPR, DSA, and possibly PLD or AILD

What's Missing or Left Unsaid?

• Systemic and structural injustices, which are produced by multiple actors over a period of time and built into systems.

• Intersectional inequalities, which are particularly problematic.

Brainstorming

• This section encourages participants to brainstorm ways to address social justice using “design justice”.

What other ways are being imagined?

• Counterdata/coding (rewriting and controlling narratives) is one method

Community-Based Movements

• Data-based technologies alter lives.
• Understanding how data systems affect communities is important.

Indigenous Data Sovereignty

• Indigenous Peoples, communities, and Nations should govern the collection, ownership, and application of data related to them.

Key Takeaways

• Social justice in the realm of AI is not static, but continually emerging.
• Sources of bias, both technical and human, are interconnected and perpetuate systemic inequalities.
• Issues raised by AI must be addressed via social, ethical and technological means.
• Imaginative interventions like counter-coding, community-based action, and reclaiming agency are crucial for addressing inequalities.

This lesson delves into social justice issues in the context of artificial intelligence. It examines how AI intersects with various forms of justice, including economic, racial, and environmental aspects, while also highlighting gaps in public understanding. Case studies, such as the Dutch SyRI case, illustrate the potential bias in AI decision-making processes.