How Ethical Is Your AI?

Conscious consumers demand fair-trade when it comes to products like coffee, and when it’s quality coffee, they are even willing to pay more for it. When it comes to our technology products though, many consumers don’t even know that “fair-trade” is possible. Behind many acts of AI “magic”, there is a human in the loop. They are often under-paid and over-tasked. In May, Facebook was forced to pay a $52M settlement to content moderators who, in exchange for wages as low as $28,000 per year, were constantly exposed to violent images which caused PTSD. There has to be a better way. Samasource is setting out to prove that ethical data-labeling is not only good for people, but is also good for business.

Samasource provides “high-quality training data and validation for the world’s leading AI technologies” and, according to their website, serves 25% of the companies on the Fortune 50 list. Founded in 2008 by Leila Janah on the principle that while talent is evenly distributed, opportunity is not, the humans in the loop are young Kenyans and Ugandans who get access to training, meaningful work, and a career path within the company. Janah passed away earlier this year at the age of 37 after a battle with epithelioid sarcoma. The company is now run by Wendy Gonzalez, formerly VP and COO of Samasource.

Gonzalez joined the company because she had a “deep desire to marry her passions of social impact, science, and technology.” This comes through strongly in the balanced approach that she brings to her work. She focuses on how the innovation that Samasource offers their clients is strengthened by the way it positively impacts their workforce. “The real power behind our solution is the combination of our software platform and our humans,” she said. To support their impact mission, Samasource brings their workers into facilities where they have secure access to computers and quality internet. Caroline Ayieko, an agent at Samasource said, “What has surprised me is unlike other companies, Sama makes the office space more safe to feel like home.” In Nairobi, they work with youth and women from the Kibera informal settlement where only 20% of dwellings have access to electricity and unemployment rates are as high as 50%.

This type of data work can be seen as a commodity. Over 80% of time on an AI project can be spent preparing and labeling the data set for use. This had led many startups to turn to a distributed network of gig workers who annotate data for cents on the dollar. But bringing their employees into a “highly secured facility with biometric access”, Samasource also allows for stronger data privacy standards and increased trust with their clients. The fact that they formally employ their workers (and gives all workers the same access to benefits regardless of if they are in Kampala or San Francisco) also reduces a revolving door which allows their labelers to build skills and codify those learnings in their technology platform. “They are not doing the same work they were doing years ago. They’re doing increasingly highly complex and interesting work. And we continue to move them up the value chain,” said Gonzalez. This has allowed Samasource to compete on quality, instead of price, and fulfill their social mission in concert with financial goals.

Samasource also doesn’t just offer their employees jobs; they offer them careers. Grace Irungu, a Samasource Agent, said, “Joining Sama back in 2007 was a turning point in my life…I grew skills-wise. Samasource exposed me to technology which is ever-changing. I admire the leadership qualities that are practiced in Sama as well as the equal opportunities that are offered up the corporate ladder. I have been exposed to various environments, creative minds, new skills, and different trainings.” Gonzalez shared that 99% of people who enter pre-employment training complete it because, “people know that if they’re going to the training, there’s a good opportunity, they could be considered for a position.” About 80% of trainees go on to be employed by the company. Samasource promotes internally, but even for people who leave, career prospects are strong. “Roughly 82% go into higher paying jobs or university education and of those that go to jobs, well over 50% stay in the ICT space,” said Gonzalez.

Their care for their people is apparent in how Samasource dealt with the outbreak of COVID-19 in Kenya. One can imagine that their secure centers, once a source of competitive advantage, could easily become sources of an outbreak. “First, we created SamaHome to responsibly respond to the COVID-19 crisis by providing our employees with a safe, hygienic place to live and work during the pandemic,” said Gonzalez. The company has also been actively piloting different work-from-home solutions which include providing employees with the equipment they need to do their jobs alongside with care packages of sanitization products and guides on how to stay healthy.

The employees have noticed this care. “With the recent pandemic that has caused many employees to lose their jobs, surprisingly all Sama staff are still working. Considering the way the economy has drastically gone down, Sama still found a way to give to its employees who are not working, something at the end of every month, including the relief bag of goodies that they received when the office was closed,” said Ayieko.

Samasource has been successful not in spite of their investments in impact, but because of them. And so, they take impact very seriously. Each quarter they do a “learnings call” alongside their earnings calls to report on their impact. And about 5 years ago they started the process for a Randomized Control Trial, the gold standard in evaluating social impact, in partnership with two professors at MIT. The results should be out later this year. Beyond all of the good jobs that they are creating, this could be the real legacy that Janah left when she founded Samasource. Samasource is on track to prove empirically that creating good, high-impact jobs and pushing for an ethical AI-supply chain is not just feasible but a good business strategy. Gonzalez hopes that this will show that “ technology is not only disruptive in a good way for business but can be disruptive in a good way for people.”

Follow me on Twitter or LinkedIn

 

Original post: https://www.forbes.com/sites/meghanmccormick/2020/06/21/how-ethical-is-your-ai/

Leave a Reply

Your email address will not be published. Required fields are marked *