🔬 Research Summary by Gregory Gondwe, an Assistant Professor of Journalism at California State University – San Bernardino and a Harvard faculty Associate with the Berkman Klein Centre.
[Original paper by Gregory Gondwe]
Overview: This study investigated ChatGPT usage among journalists in sub-Saharan Africa and its implications for misinformation, plagiarism, and stereotypes. The research highlighted the challenges posed by ChatGPT’s reliance on limited and non-representative databases for African contexts, particularly the limitations related to language and language-switch codes. Despite these challenges, ChatGPT offers opportunities for effective journalism practice in the region.
Introduction
The increasing reliance on generative AI tools and the aspiration for a connected world have sparked debates about the Global South’s ability to effectively engage with new media technologies. Some scholars highlight the Global South’s lack of resources and technological skills as a barrier to optimal AI utilization. In contrast, others argue that the “Global Village” concept suggests active participation in these debates, emphasizing interconnectedness beyond cultural and geographical boundaries.
Skeptics challenge this vision, viewing the networked world as a manifestation of capitalism that exploits the Global South’s data. Concerns about technology’s potential exclusion of marginalized communities in system design and data biases are raised. The study aims to investigate the integration of generative AI, specifically ChatGPT, in the practices of journalists in five sub-Saharan African countries. Through interviews, the research explores the challenges and potential benefits of ChatGPT use in contexts with underrepresented databases, examining the Global South’s involvement in generative AI, the representation of the Global South corpus within these tools, and potential concerns among journalists regarding their utilization of generative AI tools in their work.
Key Insights
Online Databases Perception
The study revealed contrasting views among journalists in sub-Saharan Africa regarding the nature of online databases. While some believed in their existence, citing increased social media presence and data availability for crowd coding and crowdsourcing, others perceived them to be almost non-existent, expressing concerns about the reliability and representativeness of online content. Journalists emphasized the need for clear and relevant information from reliable sources, often disregarding content in local languages or code-switching.
Challenges with Internet Connectivity
Journalists in sub-Saharan Africa faced significant challenges with unreliable internet connections and slow download speeds. This posed obstacles in their utilization of generative AI tools like ChatGPT. Despite the availability of alternative solutions, such as live coverage through Facebook or WhatsApp, the study highlighted the need for better internet infrastructure to support seamless journalistic practices in the region.
Awareness of ChatGPT Inaccuracies
Initially, journalists exhibited limited awareness of ChatGPT’s inaccuracies and were impressed by its ability to organize thoughts and provide basic information. However, as they gained more experience with the tool, they became cautious about its accuracy and reliability. Careless errors raised concerns about the trustworthiness of ChatGPT in providing accurate information, prompting journalists to question its reliability for professional use.
Perpetuation of Stereotypes
The study shed light on ChatGPT’s perpetuation of stereotypes about Africa, including poverty, corruption, and gender issues. Journalists expressed caution and concern about how the tool portrayed African countries and leaders. Additionally, ChatGPT was perceived to be biased in favor of Western narratives, raising questions about AI’s potential to reinforce existing stereotypes. This finding emphasized the importance of critically examining and decolonizing AI tools to ensure fair and unbiased representations in journalism.
Between the lines
Journalists in Sub-Saharan Africa had varying perceptions of online databases; some considered them valuable for crowd coding and crowdsourcing, while others viewed them as unreliable. Connectivity challenges were common, with some needing to be made aware of the role of initiatives like Facebook Basics and zero-rating services. Initially, journalists found ChatGPT impressive but grew cautious about its accuracy over time, especially regarding misinformation. ChatGPT was deemed unsuitable for writing complete stories, and Google remained the preferred source for information. However, ChatGPT was used to organize thoughts. The tool was criticized for perpetuating stereotypes about Africa, raising concerns about bias and its alignment with Western narratives.