Georges Vivien Houngbonon invites to qualify the recommendations of the AI Forum

0

In this interview with fo-Koffi Djamessi of scidev.net, the researcher in digital economics, Georges Vivien Houngbonon at the Paris School of Economics, explains that the recommendations issued to the African States by the participants in the first Forum on artificial intelligence (AI), organized by UNESCO on 12 and 13 December 2018, should be qualified for the countries of sub-Saharan Africa. One of the recommendations of this forum is to rely on artificial intelligence (AI) to address current development challenges in Africa. 

Researchers, economists and diplomats had insisted on this occasion on the urgency for African States to promote the integration of AI into African economic systems.

What are the major obstacles to the development of AI in African countries?

The first of these concerns the local expertise deficit. Recent initiatives such as the Google AI Center in Ghana, the master in machine intelligence in Rwanda or the master in data science from the National Polytechnic Institute in Côte d'Ivoire remain very modest in the face of the magnitude of the challenge. Secondly, the lack of national and regional data-development strategies accentuates the cost of learning AI programmes in the African context. The success of the AI also depends on advances in connectivity. However, on this point, African States are not progressing fast enough, with the penetration rate of the Internet being just above 20%.

Second, the lack of national and regional data-development strategies accentuates the cost of learning AI programs in the African context. »

Georges Vivien Houngbonon, researcher in digital economics at the Paris School of Economics

Finally, access to electricity is a necessary condition, but it still remains out of reach of 60% of Africans, with an average progression of just 1 point per year. In addition to these surmountable obstacles, there is a specificity of Africa that diminishes, at least in the short term, the importance of AI for its development. This is the preponderance of the informal sector, characterised by urban trade and crafts and family-type agriculture in rural areas. This situation limits the penetration of AI into the economic sphere.

Should it be inferred that African States should curb their enthusiasm towards the AI?

Not exactly. Under current conditions, the AI will have more impact in modernising public administration, through better prevention of offences, a reduction in tax evasion, a customization of education and the relationship with public service users. In the private sector, for example, the introduction of AI into the Internet of things (IoT) will optimize the costs of logistics, one of the factors of competitiveness in agro-business. It should be said that the impact of AI in Africa will be manifested in the medium or long term in Africa. However, the medium term here refers to a horizon of less than 10 years; Therefore, it is now necessary to initiate the appropriation and domestication of the AI in Africa.

In the current context, what is the issue of artificial intelligence for the development of African countries?

Today, the challenge for Africa is to take ownership of this technology, to domesticate it and to make it a powerful lever for development. On the economic level, the AI, substituting for repetitive tasks, allows to do more with less. For States that need more resources to finance social priorities, it is an opportunity to save money on routine tasks.

"The challenge for Africa is to take ownership of this technology, domesticate it and make it a powerful lever for development. »

Georges Vivien Houngbonon, researcher in digital economics at the Paris School of Economics

For the private sector, this is mainly an opportunity to reinvent new economic models to serve customers who have hitherto been excluded from formal transactions because they are too costly to reach. This is the case for credit services for which the AI will increase the scoring capacity of the clients. Access to credit will be crucial for the deployment of services that remain out of the reach of the poorest, including e-commerce, energy, health and education.

In a study published in April 2018, researchers at the International Development Research Centre warn that the employment situation could deteriorate by 2030 in the countries of the South, as a result of the use of AI. What are the provisions that African States should take to deal with this situation?

The share of destructive jobs is certainly very high, but it affects very few people in Africa because of the preponderance of informal employment, which still occupies more than 80% of the workforce. However, this does not allow African States not to prepare their citizens in the era of artificial intelligence. Concretely, there is an urgent need to reinvent education, focusing on interpersonal relationships, critical thinking, intuition, entrepreneurship and leadership. Like the learning of foreign languages, the introduction to digital literature must become a transversal competence acquired from the earliest teaching cycles. In addition, more effort should be made on continuing education for all.

At the African Forum on AI organized by UNESCO on 12 and 13 December 2018 in Morocco, some specialists drew attention to the risk of "cybercolonization" of Africa by the digital giants because the continent does not yet have the control of its data. Do you share their fears?

Africa, like Europe, runs the risk of a loss of digital sovereignty. However, this risk does not stem from the lack of control of its data, but rather from a lack of local expertise capable of appropriating the data and making it a tool for development. The data remains widely available to States, and what has changed is that it is increasingly being produced by private entities, very often outside the States. While more data centres are installed on the continent and regulatory texts strictly govern access and use of data, States have not yet developed the expertise to control data exploitation. Meanwhile, digital giants are expanding their grip on data that is now an integral part of the business model. The traces left by subscribers, including location data, are particularly useful for the construction of socio-economic indicators or the development of artificial intelligence applications. Banning their uses would be tantamount to limiting innovation to the detriment of African economies. Therefore, in order to counter a loss of digital sovereignty, Africa should promote open data and massively train local experts in their treatment.

One of the major recommendations issued to the African Union at the end of the Forum is the need to organize on a continental scale for the development of the AI. What do you think the shape of this Organization should be?

The AU's role in promoting AI should be the allocation of financial resources so as to encourage local research and development teams to produce AI models capable of meeting the specific needs of their countries and to influence the definition of international standards and standards. R&D in IA should therefore be conducted nationally or at most regionally, as needs are specific. Such decentralization also has the advantage of creating an emulation between regions, a source of excellence and innovation.

What about the role of African researchers?

African researchers and engineers need to be more oriented towards the mastery of the AI. At present they are in significant numbers in the countries of North Africa and South Africa, but very few are active in sub-Saharan Africa. The establishment of a research network on a continental scale could be a way to initiate a dynamic in the field of AI. In the medium term, it should be integrated into international networks and work closely with the private sector and the State.

What should be their priority research areas, for example?

Data science should be the priority research area. But for it to be relevant, it must be carried out in partnership with the producers of massive data. Public promotion of open data is essential to this end. Secondly, social science research, particularly in economics and sociology, applied to digital should be more encouraged. Finally, research in design, communication and digital media also offers promising prospects for African economies.

LAISSER UN COMMENTAIRE

Please enter your comment!
Please enter your name here