101 total views, 1 views today
Last month, Google opened its first Artificial Intelligence centre in Africa, situated in Ghana. Moustapha Cisse, head of Google’s AI Research Lab in Accra in an interactive session with journalists including BusinessDay’s Caleb Ojewale, gave insights into how it will impact Africa. Excerpts:
What is the objective for setting up this centre in Africa?
Our primary goal is to advance the science of artificial intelligence in Africa. The results of our research will hopefully be used in applied problems to enable and accelerate finding new solutions to them. Most of what we do in our research centre in Google, not just in Accra, we publish it, and is available to the larger public.
We also open source code so that everybody can just go out there, take the code we publish and use it to build all sort of things. In fact, we have seen many times, surprising ways of using code that we open source. For instance, a team of researchers from the University of Pennsylvania and IITA-Tanzania, used TensorFlow, a tool we have open sourced, to build new artificial intelligence models that are deployed on phone to monitor crop disease.
We did not do this, but people who just used the tools we built. When we do science, the results of our research, usually and hopefully because it is of good quality, goes way further than what we expect. We are hoping to see the same things happen in Accra and across Africa in general.
One of the biggest challenges foreseen is building trust in this technology because data is involved, how will this be addressed?
One of the areas of research I am personally interested in is what I call axiological machine intelligence. I believe that AI and any other technology that we build should be aligned with the best value of the society to which we identify ourselves. Only this way will we be able to build trust between the people, us who are building the technology, and the technology itself that we are deploying.
This means that from the beginning, when in the ideation phase, we design the technology while keeping in mind important and critical issues such as algorithm fairness. How do we design and build technologies in ways they do not discriminate against certain categories of people. Or how do we design new algorithms that from the beginning incorporate notions of privacy that do not breach the privacy of people. Or how do we design the technology in a way that the predictions and decisions that are taken by the algorithms that learn from the data are transparent and how to make these algorithms reliable. All these are important questions being investigated here by our researchers.
I am from Senegal, and I can tell you we did not pick this place because Ghana has better Jollof, because the best Jollof is in Senegal (laughs). That being said, I think there are many countries in Africa that are good candidates for a research lab like this one. I think Ghana is just one good candidate among all these candidates. Lagos could have been a very good candidate and in fact, Google’s launch pad accelerator is in Lagos. So, there are different programmes in different countries and Ghana is just one good candidate among others.
Many multinationals have been sort of accused of bringing western solutions to Africa and they do not work. What are you doing differently?
First of all, I am not a westerner (laughs). So, I cannot answer from a westerner perspective. From an African perspective, I think it is important that we design solutions that fit our problems, that we give ourselves the tools and capability to build our own solutions. That is exactly what we plan to do here because there are many Africans in this team as well.
Do you accept proposals from external bodies, for instance if the police comes to ask for assistance to use AI in deterring crimes?
We have our AI principles at Google and we only work on projects that comply with our AI principles, so we are extremely careful about any project that may be conflicting with our AI principle. That being said, for the greater good, if we can help NGOs or governments improve certain aspects of people’s lives, we are very happy to contribute our expertise in order to advance to those areas. Like in any application that may put at risk the privacy or lives of inidivduals, we are very careful and chiefly want to be in compliance with our AI principles.
Apart from health, in what other sectors are you working on solutions?
We primarily try to advance the foundational aspects of artificial intelligence, and use those advances in different fields. Apart from health, there are people here interested in using AI to improve various aspects in agriculture. There are people interested in using AI to analyse satellite imagery to support census and inform the decisions that will be made by policy makers. There are many different things that we can do with AI today and we are looking at all the exciting and challenging problems that we can put our hands on.