Reflections from the First International Conference on Governance, Ethics and Technology (GET1) jointly organised by the Institute of Humanities (IOH), Pan-Atlantic University, Lagos, Nigeria and Centre for Leadership, Ashesi University, Berekuso, Ghana.AI for intelligence, investigations and public protection
During the First International Conference on Governance, Ethics and Technology (GET1), Prof. Luis G. Franceschi, the Assistant Secretary-General of the Commonwealth of Nations, presented a demo of a project at “Commonwealth AI governance”, with plans to create avatar ministers of health, law and finance. Their thought is that AI generates the “right” questions and answers based on available data, giving leaders the “accurate, data-driven solutions” needed to tackle challenges. He said there are hopes that in the future, these AI-powered ministers might become the next generation of government advisors, noting that
“AI would not replace humans, but humans with AI would replace humans without AI.”
The idea of having avatar ministers in Africa comes from decades of corruption in government that has plagued the region. AI avatar ministers may reduce personal corruption and increase evidence-based policymaking, but it would not eliminate corruption. If these avatar ministers get into the hands of malicious AI developers who can manipulate their training dataset, this initiative could be replacing one form of corruption with a more advanced one. Moreover, governing people involves values, humanity, and cultural sensitivity which AI cannot fully replicate.
There was also a panel discussion on “Ethical Governance and Human Flourishing in Africa.” Our discussions were centered around AI adoption in Africa. The main concern was that the data set used in training the AI technologies used by Africans, in Africa is unable to truly and effectively consider African norms and experiences. For instance, some widely used facial recognition systems have been trained primarily on images of faces with lighter skin tones. Even the NIST-constructed IJB-A dataset that was primarily created for geographical diversity has 79.6% of faces with lighter skin tones. Now, imagine using such facial recognition technology in Africa for border control or law enforcement. This could lead to wrongful arrests and have ripple, harmful effects on people. This takes us back to considering the uncertain or probable nature of outputs from systems supported by AI.
One of the central insights was the importance of cultural context in developing effective governance and technology solutions. Reverend Appiah, the immediate past president of Institute of Directors, Ghana captures this metaphorically.
“Anything you borrow is like a decorated donkey, it would always be an ass.”
Introducing the acronym ATTIRE (Accountability, Transparency, Trustworthiness, Integrity, Responsibility, and Excellence) as a framework for actions in governance and development, she highlighted the need for “mental decolonisation” and an understanding of our own identities. The panel also highlighted the necessity of addressing existing social power imbalances to prevent technology from perpetuating inequalities, especially for marginalised groups.
There were then recommendations around giving underrepresented voices the opportunity to sit at tables where decisions on how AI tools are developed and deployed are made. Contributing to this discussion, participants noted that storytelling would be an instrumental means of understanding lived experiences especially in Africa where storytelling (orature) has been a core part of the culture.
We also considered a participatory approach that brings technical and non-technical groups together. But there were concerns around non-technical members being able to follow conversations about AI tools. To address this, we considered different training opportunities, giving non-technical members basic knowledge to follow through with conversations and contribute meaningfully. Another recommendation was finding out beforehand how members wish to communicate and incorporating this into how we engage with them. Contributors believed that this would inspire confidence in members and encourage meaningful participation.
As one of the keynote speakers puts it, “responsible innovation is not just about creating new technology or new ideas. It is about ensuring that those innovations serve humanity, are grounded in ethical reasoning, and are sustainable in the long term.”
Participatory oversight is also central to ensuring responsible innovation. But, for oversight bodies to be successful and operate effectively, they must be independent, have a clear function, be representative, transparent, and equipped with adequate resources and outreach capabilities. Oversight bodies embedding a participatory approach create space for researchers, members representing commercial voices, practitioners, and underrepresented groups to engage in meaningful dialogue. These groups can collaboratively work together to expand the exploration and understanding of social issues like the use of AI in law enforcement and criminal justice.