News GENERATIVE AI: OPPORTUNITIES & RESPONSIBLE USE
RAIN-Africa hosted a highly successful edition of our esteemed workshop series. This installment revolved around the theme “Generative AI: Opportunities and Responsible Use.” The event provided an exceptional platform for a lineup of accomplished panelists who graciously shared their invaluable insights on the subject. More specifically, the workshop featured engaging and enlightening presentations delivered by three distinguished experts: Renée Cummings, Darlington Akogo, and Lavina Ramkissoon.
Notably, our co-founders played an integral role in the event. Jerry John Kponyo inaugurated the workshop, setting the stage for the insightful discussions that followed while Caitlin Corrigan took on the role of discussion moderator.
This insightful event took place on July 26th, 2023. For those who were unable to attend, our team has prepared a detailed recap. Additionally, recordings of the event are now available, providing an opportunity to catch up on the enriching discussions and content.
“Data is suffused with cultural meaning independent from its computational meaning. So, data is cultural. Data is political. And we have to pay attention.” Renée Cummings, Professor of Practice in Data Science at the University of Virginia, is a Criminologist, Criminal Psychologist, Artificial Intelligence Ethicist, Therapeutic Jurisprudence Specialist, and Urban Technologist.
Renée highlighted that Generative AI has recently undergone a significant surge. Her presentation revolved around a fundamental question: How can we effectively harness the potential of generative AI while maximizing its benefits and minimizing its risks? She delved into the mechanics of large language models, like generative AI models such as ChatGPT, which are constructed by assimilating extensive historical data. Renée explained that this data includes a range of historical data harms, such as ingrained biases and prejudices, which have gradually permeated these datasets over time. She further expressed her concern that the use of data in large language models could be dysfunctional and dangerous, particularly for people of color, leading to discrimination based on race, gender, and sexuality. Thus, she underlined the paramount importance of acknowledging and addressing biases in the data used for constructing these models. Renée also touched on other associated risks, including privacy, misinformation, and disinformation. She explained how these aspects intertwine with democracy and decision-making, particularly in the context of the impending 2024 United States election. She concluded her presentation by pointing out that we have the established risks and the anticipated risks evolving along with this emerging technology and stressing the importance of responsibly creating generative AI.
“The possibilities are endless with the use of AI technologies. For example, in healthcare agriculture education.” Darlington Akogo, Founder, CEO, minoHealth AI, karaAgro AI, AI4Radiology Chair, United Nations ITU & WHO Focus Group on AI For Health
Darlington emphasized that we are currently in the age of Artificial General Intelligence. Experts are now building large language models and witnessing their capabilities. He mentioned that it is necessary to acknowledge the profound nature of AI systems and to shed light on the possibilities and implications they bring. Opportunities in various fields, such as healthcare, now allow for the creation of AI-powered medical assistants that can support doctors. According to Darlington, this technology represents one of the most significant advancements we have achieved and has the potential to redefine our society. He highlighted that unlike previous tools, this is the first time we have created a tool capable of generating other tools. However, there are also challenges, such as its impact on labor and issues related to bias. Yet, as long as we are honest about the power of the technology we have developed, we can have an open conversation about its associated risks.
“There are many ways actually to look at anything. I will be looking into three major aspects: the perspective of tech, human and social.” Lavina Ramkissoon, AI Ethics and Technology Policy Expert at the Africa Union, is our RAIN-Africa representative in South Africa.
In her presentation, Lavina pointed out that the technological perspective leads to real-time automation and immediate decision-making. This, in turn, raises concerns about the entire data lifecycle, encompassing its collection, utilization, consumption, and decommissioning, all of which are crucial considerations. The human perspective is closely related to how individuals interact with these technologies, encompassing issues of trust and reliance on these systems. On the other hand, the social perspective incorporates broader societal morals and values.
In the discussion that followed, Renée highlighted that using terms like “Global South” and “Global North” underscores the prevailing division. Meanwhile, Lavina emphasized the pressing concern of task automation. She noted that due to the rapid pace of technological advancements, people often adopt AI tools immediately upon release without considering potential implications.
Darlington pointed out the significant absence of representation from the African continent in the data used to train these models. He emphasized the necessity for Africans to create their own large language models as an alternative. Furthermore, he highlighted the substantial costs associated with building, operating, and maintaining these AI models. Consequently, only a limited number of entities are able to participate in the development process.
In the course of the discussion, the invited speakers voiced their concerns about various issues. These included the persistent lack of connectivity, which continues to impede the utilization of AI tools in certain regions. Additionally, they explored the technology’s transformative potential in key sectors like education, agriculture labor.
Renée emphasized the significance of recognizing that these AI models function as language models rather than knowledge models. Such recognition becomes pivotal when engaging with and leveraging the technology while Lavina underscored the necessity for cultivating novel skills to meet the demands of the evolving workforce landscape. Darlington discussed the imperative of engaging in an honest discourse regarding the future of work and the profound influence of technology within this domain. There was a collective agreement on the importance of self-upskilling to comprehend how Generative AI is revolutionizing all areas, including domains like copyright and intellectual property rights.
We extend our heartfelt gratitude to our distinguished guest speakers and the participants from around the world for their insightful questions and contributions.
