Refugees in Kenya's Kakuma camp, who have been a source of labor for global tech companies, are now facing reduced pay and precarious employment. Entry-level digital jobs, once a lifeline, have diminished by approximately 50% since 2022, according to estimates from the International Trade Centre. This decline is attributed, in part, to the increasing capabilities of artificial intelligence, which is automating tasks previously handled by human workers.
For refugees like Grace, who requested anonymity, the search for work involves tasks such as using ChatGPT to research hymn translations in unfamiliar languages. She submits this information to platforms like AOP Connect, a research platform owned by RWS. While her work could potentially earn her a "discretionary reward" of up to $500, there is no guarantee of payment, and many earn significantly less without explanation. This model of compensation, where wages are uncertain, contrasts with earlier opportunities where workers could earn between 500 to 1,000 Kenyan shillings ($3 to $6) per day.
Organizations that train refugees for tech outsourcing, such as the Solidarity Initiative for Refugees (SIR), have seen a shift. Previously, finding work in data annotation, transcription, and translation was common. Now, SIR staff note a scarcity of such roles, with platforms like Remotasks, a subsidiary of Scale AI, having ceased operations in Kenya in 2024. This has led training groups to focus on more specialized skills, though access to these higher-skilled jobs remains challenging for many.
The situation highlights a broader trend of exploitation within the tech outsourcing industry. Reports indicate that some companies have paid significantly less to workers in countries like Kenya than they received from their clients. For instance, OpenAI reportedly agreed to pay an outsourcing firm $12.50 per hour per Kenyan worker, while the workers themselves received approximately $2 per hour. Activists have criticized such practices, noting that companies may not engage in similar wage disparities in their home countries.
The nature of the work itself can also be demanding and sometimes disturbing. Some data annotators have reported being tasked with labeling images and videos depicting violence, graphic injuries, and pornography. The pressure to work faster for less pay has led some individuals to work long hours, sacrificing sleep to meet the demands of online platforms.
While some highly skilled refugees have found success in roles like web development and software engineering, the majority remain confined to low-paid microwork. The diminishing opportunities and uncertain pay in the tech sector underscore the vulnerability of gig workers, particularly those in refugee camps with limited legal avenues for employment. Organizations like Action for Refugee Life (AReL) continue to offer training in areas like data analytics to equip refugees with market-relevant skills, aiming to improve their employability in the tech industry.