Editor’s note: Last week Google hosted the annual Google For Africa event as part of our commitment to make the internet more useful in Africa, and to support the communities and businesses that will power Africa’s economic growth. This commitment includes our investment in research. Since announcing the Google AI Research Center in Accra, Ghana in 2018, we have made great strides in our mission to use AI for societal impact. In May we made several exciting announcements aimed at expanding these commitments.
Yossi Matias, VP of Engineering and Research, who oversees research in Africa, spoke with Jeff Dean, SVP of Google Research, who championed the opening of the AI Research Center, about the potential of AI in Africa.
Jeff: It’s remarkable how far we’ve come since we opened the center in Accra. I was excited then about the talented pool of researchers in Africa. I believed that by bringing together leading researchers and engineers, and collaborating with universities and the wider research community, we could push the boundaries of AI to solve critical challenges on the continent. It’s great to see progress on many fronts, from healthcare and education to agriculture and the climate crisis.
As part of Google For Africa last week, I spoke with Googlers across the continent about recent research and met several who studied at African universities we partner with. Yossi, from your perspective, how does our Research Center in Accra support the wider research ecosystem and benefit from it?
Yossi: I believe that nurturing local talent and working together with the community are critical to our mission. We’ve signed research agreements with five universities in Africa to conduct joint research, and I was fortunate to participate in the inauguration of the African Master of Machine Intelligence (AMMI) program, of which Google is a founding partner. Many AMMI graduates have continued their studies or taken positions in industry, including at our Accra Research Center where we offer an AI residency program. We’ve had three cohorts of AI residents to date.
Our researchers in Africa, and the partners and organizations we collaborate with, understand the local challenges best and can build and implement solutions that are helpful for their communities.
Jeff: For me, the Open Buildings initiative to map Africa’s built environment is a great example of that kind of collaborative solution. Can you share more about this?
Yossi: Absolutely. The Accra team used satellite imagery and machine learning to detect more than half a billion distinct structures and made the dataset available for public use. UN organizations, governments, non-profits, and startups have used the data for various applications, such as understanding energy needs for urban planning and managing the humanitarian response after a crisis. I’m very proud that we are now scaling this technology to countries outside of Africa as well.
Jeff: That’s a great achievement. It’s important to remember that the solutions we build in Africa can be scalable and useful globally. Africa has the world’s youngest population, so it’s essential that we continue to nurture the next generation of tech talent.
We must also keep working to make information accessible for this growing, diverse population. I’m proud of our efforts to use machine translation breakthroughs to bring more African languages online. Several languages were added to Google translate this year, including Bambara, Luganda, Oromo and Sepedi, which are spoken by a combined 85 million people. My mom spoke fluent Lugbara from our time living in Uganda when I was five—Lugbara didn’t make the set of languages added in this round, but we’re working on it!
Yossi: That’s just the start. Conversational technologies also have exciting educational applications that could help students and businesses. We recently collaborated with job seekers to build the Interview Warmup Tool, featured at the Google For Africa event, which uses machine learning and large language models to help job seekers prepare for interviews.