During the last day of the ICEGOV 2024 conference in Pretoria, South Africa, yesterday, October 4, 2024, a panel of ICT experts discussed the potential benefits of the deployment of Artificial Intelligence (AI) in public services, but also cautioned against blind adoption of this transformative technology. They highlighted the need for careful consideration of AI’s potential impact on citizens and managing expectations against the limitations presented by context and the environment.
In her keynote lecture, Research Director in the Center for Technology in Government at the Rockefeller College of Public Affairs and Policy in the United States, Ms. Mila Gasco-Hermandez, said the deployment of AI for public value is not a stand-alone tool but has systemic dependencies and success factors.
“AI encompasses various aspects, but I want to emphasize that it involves the integration of machines, algorithms, and data. It tries to imitate human intelligence and has the capability to learn and adapt,” Ms. Gasco-Hermandez explained in her keynote lecture.
Ms. Gasco-Hermandez emphasized the importance of making a distinction between “adopting” and “using” AI, saying that her research, which was the subject of the lecture, focuses on the decision-making process that precedes the practical deployment of artificial intelligence, as opposed to merely taking a policy position on AI.
Wary of the tendency to use off-the-shelf ICT solutions, the Center for Technology in Government’s Research Director also stressed the importance of the environmental context in which AI is being deployed.
“My research is primarily based on information from Europe, particularly in the UK, indicating that context influences the adoption and implementation of AI technologies,” she stated.
Ms. Gasco-Hermandez spoke of the “drivers of adoption” and gave advice on what countries should consider when adopting AI in public service organizations. Several contextual factors, such as the sales pitches of IT industry vendors, could inform decision-making, potentially leading to the adoption of inappropriate solutions if not carefully considered, she explained.
According to Ms. Gasco-Hermandez the “determinants of success” in the deployment of AI in government institutions cover data and information-related factors, from lack of understanding of the data requirements to issues of data quality and embedded bias. The technological success factors include lack of skills and computing incapacity, which might result in frequent system crashes.
“It is important to have the technical knowledge to be able to build an AI system or to manage it even if it is outsourced,” Ms. Gasco-Hermandez.
She added that government institutions that are contemplating the use of AI for the public good must consider their country’s regulatory posture to data related issues. For example, a regulatory approach on whether data is a human right, or a commercial asset is key to the deployment of AI in government, she said.
Ms. Gasco-Hermandez’s lecture also touched on the importance of creating citizen’s awareness of the capabilities and limitations of AI to better manage public expectations.
“The expectations, preferences and needs of citizens need to be taken into account in the case of AI not only in terms of the service offering but also to co-create the solutions,” said Ms. Gasco-Hermandez.