The farmer used a TensorFlow-based application on his mobile phone. TensorFlow is Googles outsourced artificial intelligence (AI) machine that helps developers create solutions to real world problems.
John Quinn, an AI researcher, said that when it comes to AI, people are most likely to think of scenes in science fiction movies. But in fact, AI has been used in everyday life, from virtual assistants to language translation in Google search.
Google now wants to position itself as AI Priority and set up research centers around the world, such as Tokyo, Zurich, New York and Paris. Last week, the technology company opened its first AI research center in Africa in Accra, Ghana.
The benefits of AI
Moustapha Cisse, a scientist in charge of Googles AI research in Africa, said AI could be used in agriculture, health and education. The goal of the Cisse team is to provide developers with the necessary research to develop products that can solve the current problems facing Africa.
Figure 2: Googles Moustapha Cisse in charge of AI projects in Africa
Most of the work we do at Google Center is not limited to AI research, but we also publish these things and open source code so that everyone can use them to build a variety of things, Cisse said.
Cisse mentioned the application used by the Tanzanian farmer to diagnose cassava disease, an example of his teams plans to launch products in collaboration with relevant departments and agencies.
A joint team of the University of Pennsylvania and the International Institute of Tropical Agriculture is using TensorFlow to build new AI models that are integrated into mobile phones to diagnose crop diseases, Cissel said. This is not our credit, but the research and development results of people who use our research and development tools.
Cisse continued: When we do scientific research, our research usually has a good prospect, because it performs very well, even better than we expected. We hope to see the same thing happen in Accra and throughout Africa.
Cisse points out that the Accra AI Center works directly with researchers at African universities by providing grants and doctoral scholarships to those interested in various fields of AI. Google also supports the robotic intelligence graduate program of the African Institute of Mathematical Sciences in Rwanda.
Cisse leads a team of nine scientists and software engineers from nine different countries, such as Lesotho, Uganda and Ireland. Its a diverse team, Cisse said, and the most important thing is that Africans are at the forefront of solving the problems facing the continent.
The Accra AI Center will also work to improve Googles translation capabilities so that it can translate African languages more accurately. Cisse said that Africa, with more than 2,000 dialects, should be better served.
Google has joined Facebook and other technology companies in launching projects in Africa to keep a close eye on the continents growing youth population.
The Prejudice of Algorithms
Even so, Silicon Valley is still trying to diversify. According to a recent CNN Business report, many employees hope that after years of oral discussion on this issue, their company will make a real difference. Nyalleng Moorosi, a software engineer at Accra Center, said that because AI works essentially on input data, it sometimes shows a real-world bias that does not include all end users and does not represent all end users.
Shortly after its launch in 2015, Google Photos had labeled black images as gorillas and sparked criticism, which kept the company busy addressing the problem. Nyalleng Moorosi said she believed that this was a problem facing the entire human race and needed to be solved through the diversification of data collection samples.
Nyalleng Moorosi said more Africans would be included in data collection to provide accurate user representation. When you build something, you think it will only serve the world you know and your neighbors, she said. But you forgot that it could radiate and eventually be deployed to foreign communities. The best way is to have a diverse team study these algorithms, and then we can make more progress.
AI works because a lot of personal data is entered into it, and the privacy issue is what happens when people hand it over to a big company like Google. Cisse said that Googles AI principle is to build trust between people who create technology and end users, and technology should be consistent with social values. We must establish fair, respectful and transparent algorithms, Cisse said.
As the Accra AI Center came into operation several months later than originally planned, Googles bet that AI will have a transformational effect in Africa will be put to the test. The whole African continent is waiting to see Googles AI efforts.