In recent years, the few papers or surveys that have to some degree either directly or indirectly addressed public sentiment towards AI in Africa include Neudert, Knuutile and Howard's (2020) large-scale survey of global perceptions of this technology’s risks, Gwagwa et al.’s (2021) summary of the state of AI in Sub-Saharan Africa, and a 2020 poll of South Africans’ perceptions of AI conducted by Viacom Global Insights South Africa. In their working paper, Neudert et al. (2020) stress that across the globe, soci-eties are fairly ambivalent about AI’s development, generally perceiving it in terms of risks rather than benefits, although they also note some key dif-ferences between regions. In contrast to other countries, people inhabiting North America and Western Europe, for example, are of the view that AI and robotics will bring about more harm than good. In terms of perceptions of threats associated with automated decision-making, people who live in Africa and East Asia, for instance, are less concerned about its pitfalls than those who occupy Europe or Latin America. Although they do not explicitly address public opinion, Gwagwa et al. (2021) note that according to a 2021 UNESCO survey of AI capacity building needs authored by Sibal, Neupane and Orlic, a mere 21 African countries have prioritised AI in their plans at national level (p. 6). Such a lukewarm response does not bode well for mar-ginalised populations, who rely on the government to exploit AI’s benefits while also mitigating its risks. Sibal er al. (2021, p.66) cite fears around AI that include issues related to privacy, lack of human agency, bias, and gender discrimination. A 2020 poll conducted by Viacom Global Insights South Africa reports that although individuals aged 18–49 are equivocal about AI’s capabilities, uncertain as to how the technology will directly affect them as well as concerned about possible job insecurity and forfeiture of human agency, most expressed the opinion that AI will ultimately be more benefi-cial than detrimental. In this regard, they cite some of AI’s main advantages as enhancement of productivity on the part of humans and the eradication of faults as some of AI’s main advantages (Dias, 2020). While useful, this poll targeted the online community, thus excluding at least 36% of South Africans who do not have access to the Internet (Kemp, 2020). Other Africa-centred industry reports briefly refer to people’s fears around AI as they pertain to income inequalities, uneven development, and the entrenchment of injustice (Butcher, Wilson-Strydom and Baijnath, 2021; Pillay, 2020; Schoeman et al., 2021).
An informative study that explicitly addresses ordinary citizens’ percep-tions of AI in a framing theory approach is one by Donald Malanga (2019), who examined how Malawian women from academia, government, civil society, and the private sector conceptualise AI and human rights. Overall, perceptions of AI were mixed. Respondents were of the view that AI could improve their social and economic circumstances on condition that it was implemented according to ethical guidelines that stressed the protection of human rights. However, given that Malawi’s constitution pays lip-service to gender equality in light of the perpetuation of patriarchal views (Malanga, 2019, p. 172), it is unsurprising that the respondents also identified specific fears around AI’s deployment. These included concerns about facial or speech recognition software being exploited to curtail freedom of expression and encroach upon their privacy and about AI being used to discriminate against women based on race, gender, and ethnicity, effectively impeding them from gaining access to jobs or places at institutions of higher learning. Of signifi-cance is that Malanga (2019) not only reports that the respondents exhibited a basic knowledge of what AI entails but also notes low levels of awareness among them of what kinds of AI projects are currently being deployed in the country. What is more, none of the women reported participating in any of Malawi’s AI initiatives, which revolve around machine learning, biometrics, and drones, to name a few (Malanga, 2019, pp. 174–175).
The sprinkling of academic studies or media reports that call attention to African perceptions of AI are those in the realm of the professions. This is particularly evident in the field of radiography, although it must be acknowl-edged that deployment of AI systems in healthcare on the continent at this stage pivot around pilot projects that are limited in nature (Owoyemi et al., 2020, p.1). In studying African radiographers’ perceptions of the use of AI in medical imaging practice, Antwi and his team found that while radiogra-phers were generally positive about AI’s application, noting improvements in diagnosis, patient safety, and research, among other things, they were fairly concerned about a number of issues. These included lack of proper training and data protection, fears around job loss, and concerns about radiographers’ core skills and roles vanishing (Antwi, Akudjedu and Botwe, 2021, p. 1). In the field of academic librarianship, Lund et al., (2020) mea-sured academic library employees’ perceptions of the use of AI across six continents including Africa. They concluded that academic librarians – particularly those described as early adopters who tend to enthusiastically espouse innovations after careful deliberation – were overwhelmingly posi-tive about AI’s integration in the day-to-day operations of libraries. This finding is in marked contrast to one by scholars in Nigeria who found that while librarians acknowledged that AI could enhance their productivity and increase satisfaction among library users, they were concerned about being replaced by machines (Abayomi et al., 2021). With regard to the use of AI in the legal profession, Adeyoju (2018) notes that while references to AI may engender fear among lawyers, it is nevertheless being embraced – albeit on a small scale – by law firms in Kenya, Nigeria, South Africa, Tanzania, and Uganda to carry out mundane tasks related to legal research, processing of historical data, analyses of statistical information, and the like. In the field of journalism, perceptions of AI’s role in journalistic practices are somewhat mixed. Allen Munoriyarwa and his colleagues point to "a deep-seated skep-ticism with AI in South African newsrooms" (Munoriyarwa, Chiumbu and Motsaathebe, 2021, pp. 15–16), which is partially the result of fears relating to ethical issues and job insecurity, while the media themselves refer to loss of decision-making on the part of editors and journalists as a significant concern (Daily Maverick, 28 September 2021). Some media outlets display positive sentiment, citing AI as a useful gatekeeper for moderating online news consumers’ (vitriolic) comments (News24, 15 July 2020). Others offer a tongue-in-cheek view of AI, with Jay Caboz from Business Insider SA allaying journalists’ fears of job insecurity by describing the paper’s use of an AI writing assistant as a failed experiment, given that what it generated was "vague and flowery" as well as factually inaccurate (Business Insider SA, 17 October 2021). In the accounting profession, Kamau and Ilamoya (2021, p.6) aver that while AI is a disruptive technology in Kenya, it should not pose a threat to basic tenets inherent in accounting, which is a view shared by the South African Institute of Professional Accountants (SAIPA) (Moyo, 2019), although Professor Rashied Small, executive for education and training at SAIPA, warns that accountants will have to complete technology-orientated training to keep pace with the profession’s rapid digi-tisation (Small, 2019). Adopting a gender lens to interrogate the impact of AI systems on financial technology ecosystems (FTs) in Ghana, Kenya, Nigeria, and South Africa, Ahmed (2021) notes that "gender workforce dis-parity […] contributes to gender barriers in the FTs that impact women’s access to many financial services and resources" (p. 15). Ahmed (2021, p. 15) further predicts that AI’s applications in such a setting will not remove biases related to gender and race.