IMPACT COMMUNICATION
AND INNOVATION
AND INNOVATION
We are becoming increasingly aware of the issue of technological bias, whether it is when technology misinterprets gender issues or when it reproduces structural racism as an integral part of its algorithms. More recently, we learned not only how the app FaceApp was a threat to its users' privacy who jumped on the aging filter trend, but also that this same program offers a beautifying filter called “hot” which, however, ends up lightening the skin of Black people.
On the other hand, what we see with the recent publication from the website Real Life is an extensive and in-depth article authored by Abeba Birhane, in which she describes how technologies like artificial intelligence are reaching the African continent with individualistic and capitalist biases that do not align with the sociocultural context of these countries.
As the author describes, the most recent edition of CyFyAfrica 2019, the Conference on Technology, Innovation and Society, provided a dynamic, vibrant, and diverse gathering that included lawmakers, UN delegates, ministers, government officials, diplomats, media, tech company representatives, and academics from over 65 countries, the majority of which were African, followed by Asian. Its main objective was to advocate for a global discourse starting from the African continent, in order to highlight the importance of amplifying African voices so that their social, educational, health, economic, and financial issues gain more prominence in the industry. Therefore, the conference served as a platform for connection between the continent and the rest of the world.
Just as in Brazil or in Latin America as a whole, there are also different hubs in Africa equivalent to the American Silicon Valley. Birhane cites as examples Sheba Valley in Addis Ababa, Yabacon Valley in Lagos, and Silicon Savannah in Nairobi: all of these are African centers that are developing cutting-edge innovation in sectors such as banking, finance, health, and education.
“The continent, in fact, benefits from these various technological and artificial intelligence developments: Ethiopian farmers, for example, can use crowdsourced data to predict and harvest better crops. And this data can improve services in the education and health sectors, such as this World Health Organization report detailsData can help resolve major inequalities that affect all social, political, and economic spheres. By revealing pervasive gender disparities in key positions, for example, data can also help bring inequalities to light and, in doing so, support social and structural reforms.”
Despite this, this is not Birhane’s point in her piece for Real Life. After all, as the author points out, there are already many tech enthusiasts who believe the continent would have all its problems solved if they simply adopted any solution involving the tags “AI,” “smart,” or “data-driven.” That is, whenever we enter the topic of technological innovation, we are inevitably confronted with the argument that the digitalization of everything is the best solution for everything. This is a very common reality in the rest of the world, and one that also appeared in the presentations made during CyFyAfrica 2019, which means the entire tech industry is doomed to blindly accept anything premised on the buzzwords “innovation” and “disruption” without any skepticism or criticism. “You don’t have to look much further than the presentations at the largest AI or machine learning conferences happening across the continent: ethical considerations, data privacy, or AI without unintended consequences are rarely a topic at these events. The next conference Deep Learning Indaba is a typical example,” Birhane criticizes.
Thus, there is already room for a critical view of these discourses, and therefore, to spare Africa from having to experience the dark sides of technology that are already seen in applications outside the continent. Although data and artificial intelligence can function as a possible and quick solution to complex social problems, these same technologies are also being used around the world as part of decision-making processes in areas such as insurance, mobile banking, healthcare, and education. This trend is also reflected on African soil with startups like Printivo in Nigeria and Mydawa in Kenya. Likewise, these companies also aim for the largest amount of data possible to analyze, infer, and deduce various behaviors and habits of the users of their applications, systems, and tools.
However, Birhane points out, in this race to develop the most innovative tool, these startups end up losing sight of the people behind the data being collected.
“‘Data’ is treated as something that must be collected, something that, without a shadow of a doubt, belongs to tech companies and governments, thus erasing individuals completely. This makes it easy to ‘nudge behavior’ or ‘harass’ people, often with the goal of generating profit for companies rather than for the people. As ‘nudge’ mechanisms become common to ‘correct’ people’s behavior, whether for their eating or exercise habits, private sector engineers end up developing automated systems aimed at deciding what is ‘correct.’ In this process, individuals who do not fit into stereotypical images of what is, for example, a ‘fit body’ or ‘healthy eating habits’ end up being punished or forced further into marginalization. The rights of the individual, the long-term social impacts of this system, and its consequences, whether foreseen or not, especially affect the most marginalized—if they are even part of this discussion at all.”
In this way, Birhane provides a more didactic perspective on buzzwords like big data or artificial intelligence in order to understand how they permeate different disciplines and how they can appear in different guises in our daily lives. That is, these technologies can refer to everything from the robots that use Facebook algorithms to dominate our timelines to our “smart” refrigerators. Both researchers in the field and tech reporters are responsible for the hype created around these issues, Birhane points out. This is how we generally came to believe that artificial intelligence could have a kind of “superpower” that puts it on a divine level. However, researchers like Melanie Mitchell have said before that we are very far from this “superintelligence.” Similarly, Birhane reminds us, Jeff Bigham also commented on how these supposed autonomous systems, whether they are robots or voice recognition algorithms, are often the product of human labor that is, in fact, very poorly paid.
“The exaggeration of the capabilities of artificial intelligence systems dissolves attention from the real threats they pose, which are much more invisible, nuanced, and graded than anything like ‘killer robots.’ AI tools are often presented as objective and value-free. In fact, some automated systems used for hiring or policing have already been scrapped due to the explicit accusation that these tools eliminate human bias. Automated systems, after all, apply the same rules to everyone. But this is one of the most damaging and erroneous misconceptions about automated systems. As Harvard mathematician Cathy O’Neil explains in Weapons of Math Destruction, ‘algorithms are opinions embedded in code.’ Under the guise of ‘AI’ or ‘data-driven,’ systems are presented as politically neutral, but because of the inherently political nature with which data is interpreted, collected, and used to produce certain results (which align with these analyzed and controlling data), these systems alter the social fabric, reinforce stereotypes, and cause even more disadvantage to those who are already at the lowest positions of the social hierarchy.”
In other words, it is necessary to reclaim the notion that these systems, however automated they may become, are systems, in the first instance, created by humans and, therefore, we are responsible for them just as we are for any other tool. These systems, therefore, present and reflect our inconsistencies, limitations, biases, and political and emotional desires. “How we see the world and how we choose to represent it is reflected in these algorithmic models of the world that we build. Thus, who builds these systems and selects the types and sources of the data used will profoundly affect the influence these systems have. The results that AI-based systems produce reflect social and cultural stereotypes, not objective truths,” argues Birhane.
Consequently, Birhane observes that the use of technology in the social sphere often ends up, intentionally or accidentally, focusing on punitive practices, whether by predicting who will commit the next crime or who will not be able to pay the next installment of their home loan. For this reason, the author comments that constructive and rehabilitative questions about why people commit crimes or what can be done to rehabilitate and support those who have just left prison are questions that are almost never asked. “Technological developments created and applied with the goal of bringing ‘security’ and ‘order’ often aim for punishment rather than rehabilitation. Furthermore, these technologies necessarily include cruel, discriminatory, and inhumane practices for some. The cruel treatment of the Uighurs in China and the unfair disadvantage of the poor are examples in this regard. Similarly, as cities like JohannesburgandKampala introduce the use of facial recognition technology, unfair and persecutory discrimination against minority groups becomes inevitable,” the researcher comments. Anticipating some of these problems, cities like San Francisco, for example, have already banned the use of facial recognition in their security cameras.
Therefore, it is not commercially advantageous for artificial intelligence companies to acknowledge the biases in their products. When this happens, the blame usually falls on specific “users,” “as if people only existed as objects to be manipulated,” reflects Birhane. In this way, these same companies take for granted that this data will automatically be their property, which demonstrates how the discourse and practice of data mining or data-driven content is inconsequential from this point of view. For the author, the very term “mining” is a reminder of the colonizing attitude that declares humans to be raw material — their data is, therefore, just another extension of the possibilities for exploiting the individual. “As we leave decisions about social problems to automated systems developed by profit-driven corporations, we not only allow our social concerns to be dictated by corporate incentives (profit), but we also allow complex moral issues to be handled by the corporate world.”
Thus, when we have a personalized service based on the study of user behavior, what we have is, in other words, the “extraction, simplification, and instrumentalization of human experience for capitalist ends,” something Shoshana Zuboff details in her book The Age of Surveillance Capitalism. Curiously, this type of thing seems banal compared to the dystopian speculations of science fiction, which imagines that humanity could be eradicated by killer machines from the future. However, Birhane says that what is most frightening is how these current systems of ours, despite being so unintelligent and automated, are already capable of removing people's autonomy by treating them as mere data to be processed. “The inferences that algorithmic behavior models make do not reflect a neutral state of the world or offer any deep causal explanation; instead, they reinforce social and historical injustices that are already firmly entrenched.”
“Technology is never neutral or objective; it is a mirror that reflects social biases and injustice. For example, during the conference, a UN delegate referenced work being developed to combat online terrorism, but unfortunately, it focused explicitly on Islamic groups, thus presenting an unrealistic and dangerous image of online terrorism: in the last nine years, there have been 350 terrorist attacks by white extremists in Europe, Australia, and North America; one-third of all such attacks in the United States during that same period were caused by white extremists. This illustrates the worrying point about stereotypical views that perceive a problem and what types of technology we develop in response. We then realize that, through the mirror of technology, there is evidence of our biased intuitions, which further reinforce stereotypes.”
Birhane mentions some key companies in the tech industry, such as Microsoft and Google’s DeepMind , as well as academic hubs like MIT as initiatives that have begun to create ethics centers so they can deal with the potential catastrophic consequences their products may generate in society. Different perspectives are directed toward this, but this multiplicity is actually a strength, not a flaw: it is truly necessary, as the author points out, for different “remedies” to be used to address the ethical, social, and economic problems that artificial intelligence is capable of generating in different contexts and cultures. However, these ethical frameworks or even the main AI ethics guides often end up focusing on the status quo as a benchmark, and thus, solutions end up concentrating on just one approach. “Insisting on a one-size-fits-all framework for AI, as one of the academics participating in the conference advised, is not only unattainable but could also worsen the problems it is supposed to address in contexts that could not have been anticipated.”
Thus, in the African context, the continent is not only trying to keep up with global technological innovations, but there is also an effort to protect its most vulnerable citizens from the consequences these new technologies may cause. By reinforcing the mission to protect and respect the rights, freedoms, and privacy of Africans, CyFyAfrica brought to the table what needs to be prioritized on the continent. And this does not just mean safeguarding individual rights and freedom, but also including critical voices that can be part of the technological narrative. According to Birhane, the event did indeed give space to critics on its panels, but those few voices were drowned out by the overwhelming number of tech enthusiasts.
Even so, after the event, the speakers still had the opportunity to share more of their thoughts in a semi-academic publication sponsored by the Observer Research Foundation, in which Birhane also participated as an author. Curiously, the researcher comments that in the first version submitted, the reviewers considered her tone too critical and, therefore, it did not fit the publication's editorial line. “Ironically, the platform that claims to give a voice to young Africans only does so when that voice aligns with the narratives and motives of powerful tech companies, lawmakers, and governments,” the researcher criticizes.
“The issue of the ‘technologization’ and digitalization of the continent is a question of what kind of society we want to live in. African youth are solving their own problems by deciding that we want to amplify ourselves and show ourselves to the rest of the world. This also means not importing the latest machine learning systems or any other AI tool without questioning what the inherent purpose is, who is benefiting, and who might be disadvantaged by the implementation of these tools.
Furthermore, as African youth lead the AI sector, this means creating more programs and databases that serve various local communities without blindly importing Western AI systems founded on individualistic and capitalist goals. This also means that we will investigate the systems we develop ourselves to set ethical standards that serve specific purposes rather than accepting Western perspectives as the standard. In a continent where many narratives are nullified by negative images such as migration, drought, and poverty, using AI to solve our problems means using AI the way we want, to understand who we are and how we want to be understood and perceived: a continent where community values triumph and no one is left behind.”
