A Response from the Global South on the article “Getting Grip on Data and AI”

This article is a response from a developing country researcher. It supports the current advancement and opportunities of Artificial Intelligence (AI) but seeks cautionary measures before it is too late. 

The fourth industrial revolution that includes data and AI offers great possibilities, as the author, Jean-Claude Burgelman, outlines. The optimism that comes from these possibilities should be guided by the reality that most in the global south have little voice or representation in global policy decision making. I ask whether we should be cautious about the lessons of past industrializations. For example, while there were visible, positive economic impacts from them in the global north, many in the global south are facing some of its worst impacts, including climate change (1,2,3). I argue that we should think ahead before tackling a future of large-scale uncertainty. 

Wealthy nations have already prepared and put in place policies to protect data and support ethical decision making. A large amount of data is needed for AI-based decision making, or to underpin generative AI based models like ChatGPT. However, nations from the global south that have these AI based tools have inadequate data protection laws. They also face a wider problem, namely lack of awareness about the possible negative implications of loose data protection.  

This should be of great concern given the impact we have already seen of the misuse of information spreading faster than it can be validated. Minority communities have been abused with technology tools that have manipulated their emotions without regulatory oversight or accountability (e.g., (4, 5, 6)). Any AI tools must thus be regulated in a way that protects the weakest individual with the faintest voice. I strongly agree with the author about the requirement of a global regulation agency. 

Nonetheless, I also strongly believe that any regulatory approach should add few more details to what has already been presented (7), such as: 

1.      That data be FAIR (Findable, Accessible, Interoperable, Reusable), as argued by Burgelman, while at the same time user consent be guaranteed. In the global south, users often do not have adequate knowledge about consent and its implications (8, 9, 10). There must be a process in place to ensure the consensual granting of access to one’s data, given data is the fuel of the current revolution.  

2.      That AI systems make themselves known, while at the same time algorithms be geared towards explaining how and when AI employs data for training (to ensure minimal bias and abuse), along with information on the algorithms used in that training. 

3.      That there must be a generic, ethical global framework that works closely with local communities and civil societies and protects free speech and an author’s opinion which may face abuse in countries with authoritarian regimes.  

4.      And that the requirement for an implementation agency encompasses extensive engagement on ways to include voices of a diverse set of people in a bottom up approach rather than a top down, institutional fashion exemplified by the UN. 

As a researcher from Bangladesh, I may not have a strong voice or have enough evidence in hand to predict the challenges we face. But I strongly believe we must take a collective approach in which the poorest regions of the world can benefit from open data and AI backed by global solidarity. 

Let’s call a truce on AI development to guarantee we can take bold, inclusive steps together.

References

 1. Martinez, L. H. (2005). Post industrial revolution human activity and climate change: Why the United States must implement mandatory limits on industrial greenhouse gas emissions. Journal of Land Use & Environmental Law, 403-421.

 2. Zhang, D. D., Lee, H. F., Wang, C., Li, B., Zhang, J., Pei, Q., & Chen, J. (2011). Climate change and large‐scale human population collapses in the pre‐industrial era. Global Ecology and Biogeography, 20(4), 520-531.

 3. Yoro, K. O., & Daramola, M. O. (2020). CO2 emission sources, greenhouse gases, and the global warming effect. In Advances in carbon capture (pp. 3-28). Woodhead Publishing.

4. Kyaw, N. N. (2019). Facebooking in Myanmar: From hate speech to fake news to partisan political communication.

5 . Wasserman, H., Madrid-Morales, D., Mare, A., Ndlovu, K., Tully, M., Emejei, E., & Uzuegbunam, C. E. (2019). Audience motivations for sharing dis-and misinformation: a comparative study in five sub-Saharan African countries. In Comparative Disinformation Workshop, Harvard University. Available at: https://cyber.harvard.edu/sites/default/files/2019-12/Audience Motivations for Sharing Dis-and Misinformation. pdf.

6. Khan, S., Mishra, J., Ahmed, N., Onyige, C. D., Lin, K. E., Siew, R., & Lim, B. H. (2022). Risk communication and community engagement during COVID-19. International Journal of Disaster Risk Reduction, 74, 102903.

7. Burgelman, J. C., & Wang, K. (2023). Data science and artificial intelligence for (better) science. Frontiers in Research Metrics and Analytics, 8, 1177903.

8. Sambasivan, N., Batool, A., Ahmed, N., Matthews, T., Thomas, K., Gaytán-Lugo, L. S., ... & Consolvo, S. (2019, May). " They Don't Leave Us Alone Anywhere We Go" Gender and Digital Abuse in South Asia. In proceedings of the 2019 CHI Conference on Human Factors in Computing Systems (pp. 1-14).

9. Sambasivan, N., Checkley, G., Batool, A., Ahmed, N., Nemer, D., Gaytán-Lugo, L. S., ... & Churchill, E. F. (2018, August). " Privacy is not for me, it is for those rich women": Performative Privacy Practices on Mobile Phones by Women in South Asia. In SOUPS@ USENIX Security Symposium (pp. 127-142).

10. Saha, M., Varghese, D., Bartindale, T., Thilsted, S. H., Ahmed, S. I., & Olivier, P. (2022). Towards Sustainable ICTD in Bangladesh: Understanding the Program and Policy Landscape and Its Implications for CSCW and HCI. Proceedings of the ACM on Human-Computer Interaction, 6(CSCW1), 1-31.