ACHIEVING SOCIAL JUSTICE FOR DATA WORKERS: IS THERE A ROLE FOR HARMONISED STANDARDS?
summary
Panel documents that training data and annotation rely on global manual labor — estimated 165 million data workers — often paid one to two cents per task or $1.34–$2/hour, and doing jobs like simulating smart cameras or labeling images. Large models cite Common Crawl and GPT‑3’s 176 billion (GPT‑4 ~1 trillion) parameters as built on that labor. EU harmonized standards can become de facto rules, but committees run via national delegations, industry-heavy membership, limited civil society voting, and only an opinion channel for outsiders.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 5:13Hidden labor behind AI
- 26:31Standards and worker gaps
- 34:47Power in standards
- 46:07Human rights in AI standards
- 51:48Standards vs regulation
- 1:00:41Reforms in standardization
- 1:05:53Audience questions and closing remarks
related talks
36mGold standard data: lessons from the trenches
01h 14mA Call for Data Dignity Is there a Need for a New Right to Be Seen in the Age of...
01h 16mTHE AI ACT AND THE CONTEXT OF EMPLOYMENT
11mData-centric AI: A new paradigm
01h 12mAI in the Urban Landscape Navigating Data Governance with Multiple Stakeholders
01h 21mBalancing Data Protection and Workplace Justice GDPR’s Role in Employee Rights ..
56mGlobal Bias in AI: When Western Data Shapes the World
31mHelp! There Are Humans in My Data!