When artificial intelligence works as intended, Silicon Valley guys often say it's "like magic."
But it is not magic. It is Brenda, a 26-year-old single mother living in Kibera, Africa's largest slum and perhaps the hardest area in the world, where hundreds of thousands of people live in a space not much larger than London's Hyde Park.
Every day Brenda leaves home to take a bus to the east side of Nairobi, where, along with more than 1,000 colleagues in the same building, she works hard on a side of artificial intelligence that we hear little about – and even less see.
In her eight-hour shift, she creates training data. Information – mostly pictures – edited so that computers can understand them.
Brenda loads a picture and then tracks pretty much everything with the mouse. People, cars, street signs, road markings – even the sky indicating whether it is cloudy or bright. The inclusion of millions of these images in an artificial intelligence system means that a self-driving car, to give an example, can "recognize" these objects in the real world. The more data available, the smarter the machine.
She and her colleagues sit nearby – often too close – on their monitors, zooming in on the images to make sure no pixel is mislabeled. Her job is checked by a supervisor, who sends her back if she's out of order. For the fastest and most accurate coaches, the honor of having your name on one of the many television screens in the office. And the most popular bonus of all: shopping vouchers.
"You can do something unique," Brenda told me as I visited the tiny house she shares with her daughter, her brother, and her mother.
"With my work, I work for something that will help someone in the future."
Brenda does this work for Samasource, a San Francisco based company that counts Google, Microsoft, Salesforce and Yahoo among its customers. Most of these companies do not like to discuss the exact nature of their work with Samasource – as is often the case with future projects – but the information that is presented here forms a crucial part of some of the largest and most famous of Silicon Valley's efforts the AI.
It's a kind of technological advancement that is unlikely to be felt in a place like Kibera. As the largest slum in Africa, more pressing problems are to be solved, such as a lack of reliable clean water and a well-known sanitary crisis.
However, artificial intelligence can not have a positive effect here. We drove to one of Kibera's few permanent buildings near a railway line that had been completely decommissioned by mud on that rainy day, but seemed to have been in regular use since its colonial days.
Nearly a year ago, this building was the dividing line between rock-breaking insurgents and the military. Today it is a thriving center for activities: a media school and a studio, a kind of cafeteria and on the first floor a room with PCs. Here, Gideon Ngeno teaches about 25 students the basics of using a PC.
The interesting thing about this process is that digital competence is high in Kibera, where smartphones are common and every other store sells chargers and accessories that are purchased with the MPesa mobile money system.
But much of Africa has skipped the era of desktop PCs. The combination of keyboard and mouse is a strange, cumbersome experience. A Samasource team member told me that she often saw trainees looking away from their PCs and picking up their phone when asked for information on the Internet.
The course taught here is specifically designed for those who wish to continue working with Samasource or any other digital business company. It costs 500 Kenyan shillings – around $ 5. This is a not insignificant figure for people who often live below the poverty line. In the past, the company offered the course for free, but without financial commitments, I was told that participation (and concentration) was at best unstable.
Well, the biggest challenge, Ngeno said, was the noise – as we talked, a group of eager kids did exactly what they would expect them to do when they presented a selection of musical instruments. Outside, a market full of activity.
A campus for California
In contrast, the Samasource office is located in a part of Nairobi that assures you that this is an up-and-coming city. The company spans four floors of a business park building, and the training data is made up of huge computer computers.
If you do not look out the windows, you might think you are in a Silicon Valley technology company. Walls are lined with corrugated iron, which in California is considered trendy trendy. Here, however, they remind of the environment from which many workers come: around 75% come from the slum.
Most impressively, Samasource has overcome a problem that most Silicon Valley companies are familiar with. Just over half of their workforce is women. This is a remarkable achievement in a country where starting a family often excludes the mother's career. A lactation room, up to 90 days of maternity leave and flexibility in shift patterns make the company an outstanding example of inclusiveness, not just in Kenya, but worldwide.
"As many people say, if you have a man in the workplace, he will support his family," said Hellen Savala, who is Human Resources.
"[But] If you have a woman in the workplace, she will support her family and extended family. So you have much more effect. "
"It would never work"
This balance is not just among the newcomers. In San Francisco's Mission District, an office much more modest than what the company has in Kenya, Samasource's chief executive Leila Janah beamed when she talked about the majority of the company's management team being female.
"It's very unusual in Silicon Valley, but especially in the field of artificial intelligence.
"We just think it's normal, it's a competitive advantage."
Samasource was founded in 2008 and received a frosty reception in the early days. In America, which was hit by the recession, outsourcing a large number of jobs to developing countries was not considered a welcome idea. It's probably not yet.
Those who liked the concept worried that there were too few people with the digital skills to bring the work to a standard that the tech giants would accept.
"Very smart people in the tech world and in the world of great philanthropy said it was a wonderful idea, but it would never work," Janah recalls. Today, Samasource is the largest organization of its kind in East Africa and also has facilities in Asia and North America.
Janah announces the company's record for accuracy and security as key reasons why Google and his colleagues come to this work. Of course, there is an obvious motivation for these companies to employ workers in parts of the world where wages are low and the locals are desperately looking for solid work.
Samasource is aimed at those who currently earn about $ 2 a day or less in the so-called informal economy from odd – or dangerous – jobs. Instead, Samasource provides a living of around $ 9 a day. This is an improvement, but still a treat for Silicon Valley.
"Yes, that's inexpensive," Janah said. "One important thing in our work, however, is not to pay wages that would distort local labor markets – if we had to pay people much more, we would shake everything off, which could have a negative impact on the cost of housing, the cost for food in the communities where our workers thrive. "
Then, of course, the question arises what happens when the work is no longer needed. After all, Samasource's core business is providing data for automated systems. What if the process of creating this data is also automated?
"That's the billions-technology question that everyone is paranoid about," Janah said.
"I think there's a lot of hype around this. But if you're actually talking to data scientists, the minds behind these algorithms are thinking, you'll find that the machine is way farther back than most people realize.
"We will need training data for a long time."
"It changed everything."
Expecting a data training is boring, repetitive, never-ending work. And if not in front of our cameras, some employees talked about how they were under pressure to work fast to achieve business goals, resulting in fewer disruptions. Some Samasource employees are freelancers who can work anywhere, but with a webcam watching them at work.
None of the staff we saw in the office had decent ergonomic support, which often squatted for hours, squatting and straining the eyes and body. The company said it would work.
Complaints about the work – which are certainly not unique in this industry – are quickly tracked down with stories about changed lives.
Samasource believes that it has affected nearly 50,000 people in developing countries. those who either worked at Samasource or are supported by someone who did that. She interviewed former employees and found that around 84% continue their formal work or complete higher education.
One of those workers who deals with bigger things is 25-year-old Idris Abdi, who was able to move out of the slum.
"It has changed my … everything," he said.
"It changed my perspective, it exposed me to see that there is no hope for living here anymore."
Follow Dave Lee on Twitter @DaveLeeBBC
Do you have more information about this or any other technology history? You can reach Dave directly and securely through the encrypted messaging app. Signal on: +1 (628) 400-7370