During the development of AI, large amounts of data must first be organized, annotated and processed, a task mostly performed by human workers. This makes human labor a critical component of the AI supply chain. These workers are sourced through digital labor platforms that are often unregulated and offer low wages, raising concerns about labor standards in AI-development. During this PhD trajectory you will study human labor as part of the AI supply chain. You will do this research as part of the pioneering Feminist Generative AI Lab, which offers an interdisciplinary environment combining artificial intelligence, design, and social sciences.
Job description We invite applications for a fully funded PhD position in the social sciences, focusing on human labor in the generative AI supply chain. The PhD Candidate is invited to prepare and execute an independent PhD project that should include empirical research among data workers (who generate, annotate, and enrich data). In most cases, data work is performed by workers from low-income countries, who earn poverty wages. For generative AI to be ethical and fair, it should account for the fair treatment of data workers. This research leads to insights into how fair treatment might look like according to the workers themselves.
Research Context Rapid advancements in generative AI have created enormous demand for manual data labeling and curation, tasks that are carried out by workers who are mostly invisible to the end users of AI. Data work is essential to the production and maintenance of AI-powered services. According to Tubaro et al (2020), there are three contributions data workers make to AI systems: generating and annotating data (AI preparation), verifying model output (AI verification), and by directly mimicking model behavior to produce a service (AI impersonation).
Studies focusing on data work spatial distribution highlights how AI companies in the global North outsource data work to Global South workers due to lower labor costs and advantageous labor laws.
This project examines how data work is outsourced in a broad sense by examining every data work process in the AI value chain. To date, ethical frameworks for AI have focused on important societal impacts of AI deployment, including user privacy, discrimination, and bias in AI-driven predictions and decisions. However, most frameworks for ethical AI remain silent on the role of the human labor in the development and maintenance in AI systems. Explicitly acknowledging fair treatment of data workers as a key component of ethical AI will raise the visibility of human workers among all stakeholders in the generative AI supply chain.
This part of the Feminist Generative AI Lab thus focuses on extending knowledge about human labor necessary for AI-systems to function. Such insights can inform a feminist approach of the human AI supply chain, and form the groundwork for platform design and activism.
Methodologies and Activities The research can be conducted through case studies, a combination of qualitative and quantitative research, digital methods, and participative methods. Activities will include:
- Design an independent PhD project
- Conduct empirical research (probably also abroad)
- Translate results into theoretical and practical lessons
- Engage and develop activities with the staff and other PhDs in the Feminist Generative AI Lab
- Publish and present research results on workshops and conferences
- Communicate research findings to a broader audience
- Participate in the ESSB graduate school’s training program