The last couple of months have been very exciting in artificial intelligence (AI), especially as we are getting more of a glimpse of what AI can do. The tangible potential of AI for the field of applied social and behavioural science became real to me once I started building custom Generative Pre-trained Transformers (GPTs). As Motalebi and Verity (2023) put it, “generative AI tools provide a multitude of opportunities and use cases for large scale projects as well as generic organizational workflows.”
A big topic has been around what all of this progress means for the majority of the world that is non-Western and what it could potentially mean for the 30% of the world’s population that does not have access to the internet. Based on several papers that I have been reading on AI in the humanitarian sector and beyond, I summarised the systemic and behavioural drivers that we can start to take into consideration when we are thinking about:
I used the socio-ecological model as an entry point. The analysis is primarily based on Ade-Ibijola and Okonkwo (2023) book chapter on the emerging challenges of AI in Africa. I used the paper to identify both the potential barriers and facilitators.I
The key barriers at the individual level include skills acquisition, uncertainty and ethical concerns. The need for theoretical and practical skills for AI development and use is crucial, yet there’s a skills gap, especially in programming and AI-related competencies. Business leaders and stakeholders are uncertain about AI’s benefits, leading to hesitancy in adopting AI technologies. Ethical challenges regarding privacy, data usage, and cultural alignment raise concerns among potential AI users.
Facilitators at the individual level include interest in AI skills development and a positive perception of AI innovations. There is a growing interest among business leaders and stakeholders in acquiring AI-related skills and knowledge, which impacts how young people look at the role of AI in their future. This in turn leads to increasing awareness and positive attitudes toward AI technologies among individuals can facilitate adoption.
At the structural and organisations levels, key barriers include lack of structured data ecosystems, insufficient infrastructure and network connectivity and ethics in technology development. AI systems need quality and diverse data. The lack of comprehensive data repositories reflects limitations in AI system development. There is emerging evidence of how current large language models are simply not representative of the Global Majority (Atari et al., 2016), which puts into question the extent to which they are currently of use. Furthermore, inadequate technological infrastructure and limited internet connectivity hinder the effective deployment and use of AI technologies, and ethical principles guiding AI development are not fully integrated, affecting trust and acceptance of AI technologies. This includes how data is currently being mined from the Global Majority with no checks and balances.
On the other hand, some of the facilitators are: AI-powered solutions being trialled in various sectors, the emergence of technology hubs and research groups and the incorporation of AI in education and training. Ade-Ibijola and Okonkwo (2023) offer examples of successful deployment of AI in sectors like finance, agriculture, and healthcare, especially in Kenya, Nigeria, and South Africa. They also mention that the establishment of technology hubs and research groups in South Africa and other African nations encourages AI development and use. There are also initiatives to integrate AI skills and knowledge into educational curricula from secondary levels and in professional training.
Lack of relevant government policies and challenges in policy and governance are the two main policy level barriers. The absence of structured AI implementation strategies and policies in most African countries hinders the development and adoption of AI. There is a need for robust governance structures to support AI innovation and its ethical, legal, and societal implications remain unmet.
To address some of these barriers, there is progress being made on the continent including the development of national AI strategies as well as ramping up data protection laws. Countries like South Africa, Nigeria, and Kenya implementing data protection laws, contributing to a more structured approach to AI deployment.
I am an AI enthusiast and optimist, and I really believe that AI can boost our work in all sectors, and help many African countries leap forward. However, I think that attention needs to be paid to the systemic and behavioural drivers in order for us to be successful in leaving no one behind. For this reason, social and behavioural scientists need to have a prominent seat at the table to contribute analysis and strategies for how to get the most out of AI for the Global Majority.
What other factors do you think need to be taken into consideration to make AI work for everyone? Share your thoughts here!
Ade-Ibijola, A., Okonkwo, C. (2023). Artificial Intelligence in Africa: Emerging Challenges. In: Eke, D.O., Wakunuma, K., Akintoye, S. (eds) Responsible AI in Africa. Social and Cultural Studies of Robots and AI. Palgrave Macmillan, Cham. https://doi.org/10.1007/978-3-031-08215-3_5
Motalebi, N. & Verity, A. (2023). Generative AI for Humanitarians. Digital Humanitarian Network.