Samasource
Samasource (later renamed Sama) was the operational vehicle through which Janah tested and proved her hypothesis that talent is equally distributed but opportunity is not. Founded in 2008 with a $50,000 grant, it grew into East Africa's largest training-data company, serving twenty-five percent of the Fortune 50 and providing the human annotation work that taught the machine-learning models at Google, Microsoft, Walmart, and eventually OpenAI and Meta what they knew. The organization employed more than 2,500 workers at its peak and claimed to have lifted over 50,000 people out of poverty. It also became, after Janah's death in January 2020, the subject of academic research and worker lawsuits documenting conditions — low wages, insecure contracts, content-moderation trauma — that contradicted nearly every principle Janah had articulated during her lifetime.
In the AI Story
Samasource's founding hypothesis was that workers recruited from Nairobi's Mathare Valley, Kibera, and Dandora — communities systematically written off by the global economy — could perform sophisticated digital work at quality levels competitive with established outsourcing markets. The hypothesis was tested across thousands of workers and hundreds of client relationships over more than a decade. It held. The data annotations produced in Nairobi met and in several measurable dimensions exceeded the benchmarks set in San Francisco.
The organization's institutional architecture was its most consequential innovation and the element most resistant to replication. Samasource built extensive training systems, quality frameworks, cultural bridges, and market-access mechanisms. For every dollar invested in the technology platform, the organization spent three to five dollars on this institutional apparatus. The ratio was not a measure of inefficiency; it was a measure of the true cost of converting formal access into effective access.
After Janah's death from epitheloid sarcoma in January 2020, the organization continued operating but underwent a gradual transformation. By 2023, the Muldoon study in AI & Society documented conditions at Sama's East African delivery centers that included low wages relative to client billing rates (approximately $2/hour to workers while clients paid up to $12/hour), insecure work, tight labor discipline, and gender-based harassment. A group of 184 content moderators filed suit alleging unfair termination and inadequate support for psychological trauma arising from exposure to violent content they had been hired to moderate for Meta.
The trajectory illustrates what Janah herself had warned against: the tendency of market pressures, absent countervailing institutional constraints, to erode the dignity that impact sourcing had been designed to provide. The organization did not fail technologically or commercially. It succeeded commercially to the point where its commercial success became inconsistent with its founding values, and the values gave way. This is the lesson the Orange Pill reader is asked to absorb: dams require maintenance, and maintenance requires maintainers.
Origin
Janah founded Samasource in 2008 after consulting work exposed her to the concentrated wealth dynamics of the traditional outsourcing industry. The name combined the Sanskrit word 'sama' — meaning 'equal' — with 'source,' signaling the organization's commitment to redistributing the sourcing of digital work toward communities the existing industry had excluded.
Early operations focused on data entry and digitization projects. The organization pivoted decisively into machine-learning training data around 2013–2015 as the AI industry's demand for human-annotated datasets exploded, positioning Samasource as a critical supplier to the same technology companies whose models would eventually reshape the global economy.
Key Ideas
Proof of universal talent. Samasource demonstrated empirically, across thousands of workers and a decade of operations, that cognitive capability for sophisticated digital work exists in communities the global economy had written off.
Three-to-one institutional ratio. The organization spent three to five times more on the institutional apparatus around the technology than on the technology itself — a cost structure that reveals the true price of converting access into outcomes.
Commercial success of dignified labor. The company reached twenty-five percent of the Fortune 50 and generated substantial revenue while paying living wages, refuting the premise that exploitation was economically necessary.
Post-2020 erosion. The documented deterioration of conditions after Janah's death provides the sharpest available test case for the thesis that institutional infrastructure requires continuous human stewardship to survive market pressure.
Debates & Critiques
Samasource's trajectory has become a contested reference in debates about ethical AI and the data-labor supply chain. Defenders argue that even the post-2020 conditions, while imperfect, represented substantial improvement over alternatives available to workers in the same communities. Critics, including the Muldoon researchers, argue that the erosion of Janah's founding principles demonstrates the structural impossibility of sustaining dignity within market-driven outsourcing without external regulatory and collective-bargaining constraints. Both readings are compatible with the empirical record; the debate is about which countervailing institutions are required to prevent the trajectory from repeating at larger scale.
Appears in the Orange Pill Cycle
Further reading
- Billy Perrigo, "OpenAI Used Kenyan Workers on Less Than $2 Per Hour to Make ChatGPT Less Toxic," Time, January 18, 2023.
- James Muldoon et al., "The poverty of ethical AI," AI & Society, 2023.
- Niamh Rowe, "Meta faces Kenya lawsuit from moderators who say they faced severe trauma," The Guardian, 2023.
- Leila Janah, "Building Jobs with Dignity," Stanford Social Innovation Review, 2018.