Nearshore Americas

Breakdown: Something Inside the “Ethical Sourcing” Model is Just Rotten

Meta’s legal troubles in Kenya and a series of media firestorms have put the claims of “ethical sourcing” trumpeted by several BPOs and data labeling firms under scrutiny. 

Questions keep being raised about the fairness, sustainability and actual impact of these companies’ operations on communities from the Global South, which has become one of the go-to sources for the labor that sustains increasingly lucrative AI technologies.

In short: A lawsuit by Kenyan content moderators against Meta –Facebook’s parent company– and outsourcing firm Sama put the spotlight on the outsourcer’s labor practices.

  • Sama provided data annotation and validation services to Facebook. Such tasks are necessary for the training of machine learning algorithms, which have grown crucial to keep social media feeds clean of noxious content.
  • Sama uses a crowdsourcing model to execute these tasks. It’s been reported that the company paid less than US$2 an hour to Kenyan workers for the job.

Catching up: Sama informed in January that it had dropped its contract with Meta, closing its content moderation business in Kenya altogether and laying off 200 employees.

Other latitudes: Similar legal and media firestorms have hit other major tech companies and their third party providers in developing nations. 

  • Scale AI and its subsidiary Remotasks –which combines crowdsourcing with a freelance/gig model– were accused of running “digital sweatshops” in the Philippines, where they allegedly offered remuneration which amounted to less than minimum wage in the country.
  • We’ve spoken with freelancers from Venezuela and Mexico who use Remotasks’ platform for work. Some claim to have earned US$32 a week for 10+ hours of work each day. 
  • TikTok and Teleperformance got in trouble with Colombian authorities following an exposé which alleged very low pay, union busting and a traumatic working environment for content moderators.

Feet to the fire: The crowdsourcing model has come under scrutiny as of late, particularly when applied to the lucrative business of data labeling services, which are an important link in the (also very lucrative) supply chain for AI. 

  • Crowdsourcing provides an efficient and affordable alternative for the execution of crucial tasks for the development of AI models. 
  • Nevertheless, the nature of the crowdsourcing model –breaking down bigger tasks into much smaller ones– lends itself easily to the justification of low wages, prompting a race to the bottom.  

Firms such as Sama built a reputation in the business world as providers of opportunities for the Global South, positioning themselves as proponents and practitioners of “impact sourcing”.

  • While ESG compliance has yet to become a must for third party providers, the bigger players in business are increasingly interested in partnering with vendors aligned with their social and environmental agendas. Thus, responsible sourcing becomes part of the sales pitch.

Blast from the past: Sama (formerly known as Samasource) has been using impact sourcing as part of its branding for over a decade.

  • NSAM has spoken with several of Sama’s executives over the years, including late founder Leila Janah. In every instance, the concept of ethical sourcing was brought up as a major part of the organization’s philosophy.

Growing distant: Some BPO and data labeling firms have leveraged the advantages of remote work, a global labor market and gig economies to crowdsource freelancers for data labeling and similar jobs without little to none of the responsibility that comes with being an employer.

  • Jack Madrid, President of the BPO industry group from the Philippines (IBPAP), said that Scale AI/Remotasks was “not directly connected to our industry”. That is, the companies operated remotely and had no contact with either IBPAP or Filipino authorities.
  • “I would say it [freelancing] is something that we need to pay attention to. But it is a force that is hard to stop,” Madrid said in another interview. “I think as an industry, we need to accept it as a challenge because that was catalyzed by the pandemic. It was catalyzed by the fact that we can now work from home.”

What’s fair?: There has yet to be a serious discussion on what constitutes fair pay for data labeling and other crowdsourced jobs in the AI supply chain.

  • The Allen Institute for AI set its own guidelines for ethical pricing: US$8.50 an hour for workers in the US and US$3-US$4 an hour for workers outside the US (in developing nations). That was back in 2019, though.

NSAM’s take: The topic of labor exploitation in the AI supply chain keeps gaining traction in media and academic circles. Expect to see even more horror stories published by major outlets as usage of AI grows among businesses, pushing the demand for data labeling and similar services higher than it already is.

If enough of these exposés pile up, we might see governments and regulators jump into action. The White House recently put out an executive order on AI, which makes several references to ethical development and application of the technology. The cases of Meta/Sama in Kenya and TikTok/Teleperformance in Colombia reached international audiences, prompting local authorities into action.

At least two major companies already got in trouble due to alleged malpractice by their third party providers, which points to apparent carelessness in their procurement processes, but also to a troublesome understanding of what “positive impact” entails in developing countries and vulnerable communities. 

Sign up for our Nearshore Americas newsletter:


 

The outsourcing industry already has a dark reputation when it comes to its workings in the Global South. It’s hard for the general public –which includes potential workers and investors– to buy into claims of positive impact and responsible sourcing when labor arbitrage keeps driving businesses in search of cheaper, increasingly efficient and more lucrative work arrangements. 

Business leadership will have to take an honest look at the actual impact of their companies’ on the communities where they operate, as well as doing proper due dilligence regarding third party partnerships. If they don’t, their claims of ethical sourcing will be taken for what they are: vacuous statements. That could eventually lead to more PR firestorms, increased government scrutiny and, most importantly, exploitation of already vulnerable communities.

Cesar Cantu

Cesar is the Managing Editor of Nearshore Americas. He's a journalist based in Mexico City, with experience covering foreign trade policy, agribusiness and the food industry in Mexico and Latin America.

Add comment