Living Wage Commitment

In the AI Story

The commitment distinguished Samasource operationally from the broader outsourcing and data-labor economy. Industry-standard practice paid data annotators fractional wages that pushed organizations toward maximum extraction per worker-hour. Samasource paid wages calibrated to local cost of living with explicit attention to dignity — wages that were below Silicon Valley rates but genuinely sustaining in the communities where workers lived.

Janah framed the commitment as a business argument precisely because she understood that moral arguments alone would not scale. 'If your self-driving car or defect-detection algorithm is fed the wrong training data,' she wrote in 2018, 'there will be disastrous consequences for your business.' The argument located the case for dignity in the self-interest of the companies consuming the labor — a framing designed to persuade procurement departments that moral arguments would not reach.

The post-2020 trajectory of Samasource tested the commitment to destruction. Muldoon and colleagues documented in 2023 that Sama workers labeling data for OpenAI and Meta were paid approximately $2/hour while the outsourcing firm billed up to $12/hour to the technology clients. The gap between Janah's founding commitment and what the organization became demonstrated that the commitment required continuous leadership maintenance to survive the market pressures that erode it by default.

The broader lesson for the AI transition concerns what kinds of labor conditions are structurally compatible with what kinds of AI output. Models trained on data produced under exploitative conditions inherit the quality consequences of those conditions. The human labor supply chain is not separable from the machine output it produces, and the industry's tendency to treat the labor as invisible infrastructure is a quality risk as much as it is an ethical one.

Origin

The commitment was formalized at Samasource's founding and maintained through Janah's lifetime as a non-negotiable operational principle.

The post-2020 erosion of the commitment at Sama — documented in the Muldoon study and in worker lawsuits — became the empirical case for how institutional commitments require continuous stewardship to survive.

Key Ideas

Dignity as quality strategy. Living wages produce better output than exploitation, converting moral commitment into operational necessity that procurement decisions can evaluate.

Foundation consequences. The data the workers produce is the foundation AI models are built on; exploitation in the supply chain has quality consequences downstream that the industry is slow to reckon with.

Requires maintenance. The commitment does not sustain itself; it requires leadership willing to hold the line against market pressures that erode it by default — as the post-2020 Sama trajectory demonstrates.

Regulatory implication. If markets alone cannot sustain the commitment, then regulatory frameworks and worker organizations become necessary rather than optional — the countervailing institutional architecture the Muldoon study explicitly calls for.

Debates & Critiques

The living-wage commitment raises questions about whether the business case Janah made is sufficient to sustain dignity at scale. Her own organization's post-2020 trajectory suggests the business case alone is insufficient without countervailing institutional pressure. Contemporary policy debates about AI training data increasingly treat labor standards as a regulatory rather than voluntary question, reflecting the empirical lesson of Sama that voluntary commitments erode when leadership changes.