IBM Research Enables Prosperity From the Tip of Africa
Downtown Johannesburg is baking in the summer heat. The city is often described as a melting pot, because, as the economic powerhouse of the continent, it attracts people, skills and investment from all over the world. On the day that we caught up with Dr. Solomon Assefa, director of IBM Research Africa, the description was literal. Tar was sticky in the streets.
It couldn’t be more different to the clear, cold mountain tops of Switzerland, where Assefa headed shortly after our interview to participate in the annual World Economic Forum. At the conference, he talked about many projects the lab has undertaken. One favorite is an application that can analyze mobile phone usage of unbanked citizens across Africa, in order to provide accurate credit scoring.
Working with local banks, it’s a system that could enable investment into the small, entrepreneurial businesses that are critical to providing jobs. Up to 60 percent of youth in South Africa, for example, are unemployed, which is viewed by many as a root cause of inequality and social unrest.
“The world has changed a lot since last year’s World Economic Forum event,” Assefa says. “Last year, the theme was about technology and governance and responsibility, but the world has become much more fractured and there’s more risks of conflict coming up. I hope the discussions will be about technology and how we can directly impact the lives of the bottom 2 or 3 billion people. How can we be seeds for change?”
Assefa’s certainly in the right place to provide answers to that question. A prolific research scientist in the field of phototonics with more than 50 patents and 150 co-authored articles to his name, he works with government, academia and enterprise to understand the potential impact for cloud computing, artificial intelligence (AI) and big data to impact and improve the lives of the world’s poorest.
IBM’s Bold Investment
Three years ago, IBM announced that it would be investing in a second African Research Lab to join its existing facility in Nairobi, Kenya. The new lab is based in Johannesburg’s Tshimologong Precinct.
The precinct had been recently purchased by the nearby University of the Witwatersrand (Wits). Professor Barry Dwolatzky of the School of Electrical and Information Engineering had somehow managed—over the course of several years—to convince his employer that it should engage in an ambitious plan to turn a shuttered nightclub and disused offices into a thriving innovation hub, around which it was hoped a high-tech business cluster would emerge. Its name, Tshimologong, is a Sesotho word that means “new beginnings.”
Remarkably, Dwolatzky’s vision of a space where academia, multinational technology firms and innovators from the city meet, mingle and share ideas is coming to fruition. Today, the precinct is home to several dedicated startup accelerators run by Wits, in diverse partnerships that include the Dutch Embassy, Ryerson University, the city of Johannesburg, South Africa’s state-owned telecommunications provider, Journalists for Human Rights and, of course, some of IBM’s best researchers.
Standing on the corner of the precinct, the lab provided the catalyst for further investments in the area. The IBM branding across the facade emboldens IBM’s “THINK” motto, which is appropriate following IBM’s 10-year commitment to the precinct that was announced in August 2016 with great fanfare. It signalled the start of partnerships with not just Wits, but with the Departments of Science and Technology and Trade and Industry. The lab was to be a cornerstone not just of the new Tshimologong Precinct, but big government-led programs in healthcare, urban transport and smart cities, and skills development.
Assefa says that the original vision has been more or less unchanged. The emphasis on research is important, Assefa says, as it means the lab can focus on understanding problems rather than fall into the common trap of rushing through “scattergun solutions” that may or may not be appropriate. The Kenyan lab, which opened in 2012, is only now commercializing some of its work, but patience is paying off as solutions are better suited to local needs.
“In our work on cancer research, we are leading the discussions with healthcare providers around the continent, but it’s been important to bring all stakeholders together right from the start so that they’re involved in the definition of the solution itself.”
—Dr. Solomon Assefa, director of IBM Research Africa
“We get the space to develop, there’s no pressure to hit headlines,” Assefa says. “But it’s a unique environment. It’s not researchers working by themselves; we’re very much in the thick of it. For example, in our work on cancer research, we are leading the discussions with healthcare providers around the continent, but it’s been important to bring all stakeholders together right from the start so that they’re involved in the definition of the solution itself.”
That’s not to say that there hasn’t been a lot of science—the lab’s output has been prolific. In healthcare, it’s produced research that maps the spread of communicable diseases in public hospitals using Arduino-based tracking devices to monitor patient movements. And in doing so, they’ve also produced valuable data from which they can analyze and improve efficiencies within the way hospitals work and also begin to map the relationships between communicable and non-communicable diseases.
“It’s well-known that a third of all cancers in Africa are tied to infectious diseases,” says Dr. David Moinina Sengeh, head of healthcare research at the lab in South Africa. “At the core, we’re building and exploring network graphs for various conditions. If you have deep insights in cancer and tuberculosis, for example, you are able to abstract relationships that we don’t usually talk about.”
Within the lab, an Advanced and Applied AI team is working on multiple issues. One team is currently building a network of air quality sensors around Johannesburg and helping the city to predict “bad air days” and their impact on public health, while another has developed a free, web-based mobile app that can help Africans understand the needs and costs of installing solar panels based on their GPS location and specific energy needs.
One project has applied machine learning to the issue of wildfires in South Africa. Currently, emergency services and provincial governments use a crude system developed in the 1970s to assess the risk of dry season wildfires and allocate resources. Models based on historical data have already proved more accurate.
Dealing With Data
Possibly the highest profile project that the lab is working on is around preparation for the nation’s largest science project, the Square Kilometre Array (SKA) radio telescope. Under construction in South Africa’s Karoo desert (and sites in Australia and New Zealand) the SKA is the world’s largest big data project. Its receivers are expected to generate upwards of 5 exobytes (i.e., 1 million TBs) of data every single day. By comparison, the Large Hadron Collider (LHC) at CERN generates an average of 20 PB per day (although when the LHC is running, it generates a petabyte per second of raw data).
The Johannesburg lab is developing ways to deal with that volume of data. It’s already produced techniques for filtering out local radio noise. “It’s the ultimate big data project,” says Assefa. “But with the right partnerships we could position South Africa as a leader in the AI techniques used to filter it. And if we can develop skills in that area we can apply them to other big data projects for banks, retailers and the environment.”
And that’s key to understanding the future vision that Assefa has for the facility. As much as the research is valuable, it’s positioning the lab physically and societally as a catalyst for change. With the facility now well embedded into the local ecosystem for research, the next goal is to increase its involvement with the startup community and improve the number and quality of data scientists in the region.
“We want to grow,” Assefa says. “We have exceptional talent that we’ve hired locally from South African universities. We have lots of interns, lots of young faces at masters and Ph.D. levels, who come in on six-month or year-long contracts to specialize in machine learning or AI. Our hope is that many will remain.”