An Economy of Empathy - 2026
(link)Summary
This talk provides a critical examination of the modern technology industry, tracing its ideological roots from eugenics and systemic bias to contemporary techno-optimism. The speaker analyzes how the demand for data labor—often involving workers in the Global South sifting through traumatic content—perpetuates exploitation. The discussion connects historical eugenic practices (e.g., IQ testing, standardized testing) to modern tech narratives, including the push for Artificial General Intelligence (AGI) and transhumanism. The core argument is that the tech oligarchy is built on biased, efficiency-driven systems that marginalize women and people of color. The talk concludes by advocating for a shift toward an 'economy of empathy' in software engineering, emphasizing conceptual clarity, interdisciplinary collaboration, and persistent resistance against systemic oppression.
Key Takeaways
- The current data labeling economy exploits workers in the Global South, forcing them to process traumatic content for LLMs at minimal wages.
- The ideological foundations of modern tech bias trace back to eugenic principles, evident in historical academic practices and current AI development models.
- Techno-optimism and the pursuit of AGI often mask underlying biases, reinforcing existing power structures and marginalizing vulnerable populations.
- Building technology with empathy requires rejecting dominant industry narratives, adopting interdisciplinary perspectives, and prioritizing the needs of affected communities.
Sections
The Exploitation of Data Labor
The speaker details the conditions faced by data labelers (e.g., at Sama) who are paid less than $2/hour to sanitize data for major tech companies. These workers are forced to process extreme and traumatic content, leading to severe mental health consequences. This labor market is linked to larger, often secretive, military and economic projects, highlighting systemic exploitation.
Historical Roots of Bias: Eugenics and Standardization
The talk traces the intellectual lineage of bias from early 20th-century eugenics movements (e.g., Stanford University's early research) to modern academic and tech practices. Concepts like standardized intelligence testing and the emphasis on 'efficiency' and 'production' are shown to have historically been shaped by racial and class biases. This history is shown to underpin modern statistical methods used in data science.
Techno-Optimism and the AGI Narrative
The speaker critiques the prevailing techno-optimist narrative—championed by figures like Marc Andreessen and proponents of long-termism—which posits that technological progress (especially AGI) is the ultimate solution to human problems. This narrative often dismisses social and ethical concerns, framing technological advancement as an inevitable, superior force, sometimes leading to the disregard of human rights and marginalized groups.
Building with Empathy: A Path Forward
To counteract systemic bias, the speaker proposes an 'economy of empathy' for software engineers. This requires three actions: 1) **Conceptual Clarity:** Rejecting the dominant, often ambiguous, frames set by the tech industry. 2) **Interdisciplinary Collaboration:** Integrating expertise from sociology, history, and ethics, rather than relying solely on tech elites. 3) **Persistent Resistance:** Actively challenging systems of oppression and advocating for open-source solutions that empower human agency and prioritize the needs of the affected population.
Keywords: data labeling, eugenics, ai bias, techno-optimism, systemic racism, software engineering ethics, ai governance, global south labor