WHITE SAVIOR COMPLEX & AI
lpfdev ftn campaign: wsc & ai report cover
A visual outline of the role of the emergence of Artificial I Intelligence (AI) in the development sector and it’s importance in the use of imagery and perpetuating ‘stereotypical’ narratives in the international development sector.
This report is the first part of a series of campaigns led by Level the Playing Field’s campaign to ‘Flip the Narrative in International Development’.
*First edition published 17 June 2024
**LPFDev interchangeably uses some terms (see glossary) across our work and in this report. We are aware that some of these terms are contested, and have many limitations. We use these terms for the purposes of aligning with currently used terminology across institutions.
***Acknowledgments
– Campaign Development and Research: Liam Chikwana, Maud Kadye
– Contributors : Helen Coskeran, Shalom Mulinge
Summary
‘International development’ organisations based or spearheaded by the Global North have been grappling with the part they have played and are playing in perpetuating structural racism and colonialism for a number of years now. The conversation particularly picked up in 2020, with many continued discussions and efforts across various international development organisations to date, examples of a few great initiatives are: #shifthepower, aid-reimagined, Development Hub, RINGO, uncharitable and no white saviors.
Although there have been many efforts to change the system, racism and colonialism still exist in parts of the international development sector for a variety of reasons, this includes but is not limited to: funding, power imbalances and the White Savior Complex (WSC).
The narrative within the international development sector has been to portray Africa using shock tactics that are demeaning and often without dignity. This is often used as a way to get more funding, but how valuable is this if it dehumanises black/African people in the process? How has this affected the algorithms that feed into Artificial Intelligence (AI)? What role does the whole international development sector have to play in this?
GLOSSARY
Decolonisation
The action or process of a state withdrawing from a former colony, leaving it independent. Processes of decolonisation understand that colonisation is more than just a physical project – it has cultural and psychological components which determine whose knowledge is privileged. Decolonisation involves seeking restorative justice through cultural, psychological and economic freedom. Decolonisation is the process of deconstructing colonial ideologies regarding the superiority and privilege of Western thought and approaches. – Racism, power and truth: Experiences of people of colour in development
Global North & Global South
“Generally speaking, the countries of the Global North have well-developed, mature economies and are both wealthy and politically stable. They also tend to be the most technologically advanced countries and their population growth is low. Most are located in North America, Europe, and Northern Asia, although a handful of southern countries, including Australia and New Zealand, are usually considered part of the Global North. The Global North has roughly 25% of the world’s population, but earns 80% of the wealth and tends to dominate the Global South both politically and economically. Countries in the Global South, by comparison, are those whose economies are still developing. They tend to be located in Africa, South America, and Southern Asia—though, perhaps ironically, more Global South countries are located in the Northern Hemisphere than in the Southern Hemisphere. Global South countries also display faster population growth than those of the Global North” – World Population Review
It is worth noting that the above is still contested as it is based on the Brandt Line. “Today the world is much more complex than the Brandt Line depicts as many poorer countries have experienced significant economic and social development. However, inequality within countries has also been growing and some commentators now talk of a ‘Global North’ and a ‘Global South’ referring respectively to richer or poorer communities which are found both within and between countries. For example, whilst India is still home to the largest concentration of poor people in a single nation it also has a very sizable middle class and a very rich elite.” – Royal Geographical Society, The Global North/South Divide
International development
For the purpose of this report, this term encompasses (For Definitions see Appendix 2):
Advocacy
Cultural Heritage
Education
Humanitarian Aid
Poverty
Volunteering ‘abroad’
Racism
“Racism includes racist ideologies, prejudiced attitudes, discriminatory behaviour, structural arrangements and institutionalized practices resulting in racial inequality as well as the fallacious notion that discriminatory relations between groups are morally and scientifically justifiable; it is reflected in discriminatory provisions in legislation or regulations and discriminatory practices as well as in anti-social beliefs and acts; it hinders the development of its victims, perverts those who practise it, divides nations internally, impedes international co-operation and gives rise to political tensions between peoples; it is contrary to the fundamental principles of international law and, consequently, seriously disturbs international peace and security.” – United Nations, Declaration on Race and Racial Prejudice
White Saviour Complex
“The White Savior Industrial Complex is an outcome of racial capitalism. Like many other industrial complexes, White Saviorism, when interwoven within aid / charity / development work, deviates from any efforts aimed at actualizing justice to creating racialized profit economies. With race and capitalism as a constant factor, White Saviorism is built on the assumption that it is mainly individuals racialized as White who ‘can make a difference’ within Black and Brown communities.” – White Saviorism in International Development: Theories, Practices and Lived Experiences
introduction
We initially planned to launch a campaign to flip the narratives of commonly-used images in the international development sector.
In launching our campaign, our primary objective was clear: to use an immersive and impactful approach in order to foster understanding and provoke thoughtful reflection. The aim of the campaign was to use imagery to demonstrate how damaging and powerful imagery within the international development sector is in shifting narratives about Africa and how that perpetuates the White Savior Complex (WSC). So we decided to centre the campaign on ‘reverse white saviorism’ – to provide a perspective in a different lens for the often overlooked, yet demeaning, undertones in images associated with traditional white saviorism (e.g. the typical pictures of children starving in Africa or the depiction of dirty environments in Africa). Our intention was not merely to spotlight disparities but to instigate active engagement and encourage a collective shift in perception, paving the way for a more inclusive and empathetic discourse on racial dynamics.
The core idea was to get people thinking about how they would feel if the shoe was on the other foot i.e.
How would people in the Global North feel if they were portrayed by people in the Global South in the same manner?
What impact does it have for people in the global south to see that they are portrayed in a limited lens and used as a tool in international development?
Due to ethical reasons, we decided that AI would be the best option to demonstrate ‘reverse white saviorism’ for our campaign. The idea was that we would harness the power of AI-generated images to challenge prevailing narratives and prompt a fresh perspective across diverse communities, however, we discovered that was not possible due to the discriminatory make-up of AI.
methodology
We initially started by looking at the whole international development sector and identified the most common areas of thematic focus (For Definitions see Appendix 2):
Advocacy, Cultural Heritage, Education, Humanitarian Assistance, Poverty, Volunteering ‘Abroad’
A sample of organisations whose work falls within the specific categories were selected for each theme. This comprised a mix of international organisations, charities and social enterprises.
A random selection of images were collected from the organisations’ publicly available material: websites, social media, ads, Google search.
Common patterns were identified and these were grouped accordingly into:
Children
Medical Aid
Education
Infrastructure
Environment
Celebrities
Volunteering ‘abroad’
Based on the above patterns of commonly used images, we developed a set of questions and statements to be inputted on an AI platform, with the intention of generating images for our campaign.
*A sample of publicly available images collated as part of the research:
results
Category: Children
Category: Medical Aid
Category: Education
Category: Infrastructure
Category: Environment
Category: Celebrities
Category: Volunteering Abroad
Findings & Conclusions
AI and Bias in Image Generation
In exploring the realm of artificial intelligence, we encountered significant limitations rooted in pre-existing biases.
This stark representation not only mirrored societal and historical stereotypes but also illustrated the inherent biases embedded within the algorithms and AI.
Despite our explicit requests, the AI consistently struggled to recognise the narrative we aimed to convey. Instead, it persistently depicted black communities in a perpetually oppressed light, reinforcing the very stereotypes we aimed to challenge. This encounter shed light on the challenges inherent in altering ingrained biases within AI systems, raising critical questions about their ability to transcend and reshape prevailing narratives.
The Challenge of Flipping Narratives
Navigating the intricacies of AI, we found a profound challenge in its capacity to comprehend and reverse entrenched stereotypes, especially within the realm of white saviorism. The AI’s struggle to break free from existing narratives became evident as it grappled with the task of altering deeply ingrained societal perceptions.
The complexities involved in this process are multi-faceted, touching upon the nuances of historical biases and cultural representations. AI, being a product of the data it learns from, inherently reflects and perpetuates the stereotypes present in the training data. As a result, attempting to redefine narratives through AI-generated content requires a delicate balance between recognising these biases and actively working to reshape them.
The difficulty lies not only in overcoming the limitations of algorithms but also in addressing the broader societal factors that contribute to the perpetuation of biased narratives. This underscores the need for a nuanced and comprehensive approach to leverage AI as a tool for positive change in challenging established stereotypes.
Acknowledging and Addressing Bias in AI
As we grappled with AI-generated content, it became clear that acknowledging biases in AI algorithms is crucial. The consequences of perpetuating stereotypes were evident as the AI depicted ‘African’ people and ‘White’ people through a one-sided vision lens, thereby reinforcing societal prejudices. This underscored the urgent need to confront biases within AI systems, which drove us to write this report.
Organisations and researchers are refining training datasets and algorithms to ensure AI reflects real-world diversity without perpetuating harmful stereotypes. These initiatives align with our campaign’s mission of challenging and reshaping narratives by promoting more ethical and inclusive AI. But, this begs the question: how can there be true diversity and consideration for this, when the people inputting the data hold their own biases?
Empowering Communities through AI
In the pursuit of empowering minorities through AI, several strategies can enhance the depiction of scenarios where black communities are sources of help and leadership. One approach is to diversify the training datasets used to teach AI algorithms, ensuring they include a wide range of representations of black individuals in positions of authority, expertise, and support. Additionally, implementing bias detection and mitigation techniques directly into AI systems can help identify and correct any prejudiced or inaccurate portrayals.
By employing these solutions, AI has the potential to positively impact societal perceptions. By accurately depicting black communities as sources of help and leadership, AI can challenge existing stereotypes and promote a more inclusive representation of diversity. This not only fosters a more accurate reflection of reality but also encourages empathy, understanding, and collaboration across different communities. Ultimately, AI’s ability to present diverse narratives can contribute significantly to fostering a more inclusive society, where all individuals are recognised for their contributions and capabilities, regardless of race or background.
As the world is increasing the use of AI, the international development sector also has a responsibility in its use of stereotypical and non-representative imagery which feeds into AI models. We also need to start having more conversations on how representation matters in the developments of tech as the world moves that way.
Appendices
Appendix 1 – Additional reading recommendations
Becoming locally led as an anti-racist practice a guide to support iNGOs
Racism, power and truth: Experiences of people of colour in development
Race and representation in NGO storytelling: responses to 5 frequently asked questions
The story of a charity that did well when it was supposed to be doing good
White Saviorism in International Development: Theories, Practices and Lived Experiences
Appendix 2 – Definitions
Advocacy – “Advocacy is the process of using information strategically to change policies that affect the lives of disadvantaged people. It often involves lobbying northern development and political institutions. Increasingly southern NGOs are developing advocacy skills to challenge local, national and international policies. This involves strengthening the structures through which poor people can participate in the formulation of the policies that control their lives, for example developing strong local networks and representation on local and national civic institutions. Projects which involve the people affected by policy change in developing, implementing and monitoring advocacy work are more likely to achieve concrete change on the ground” – The How and Why of Advocacy, Guidance Notes – Megan Lloyd Laney, Jane Scobie and Alastair Fraser
Cultural Heritage – “Cultural heritage includes artefacts, monuments, a group of buildings and sites, museums that have a diversity of values including symbolic, historic, artistic, aesthetic, ethnological or anthropological, scientific and social significance. It includes tangible heritage (movable, immobile and underwater), intangible cultural heritage (ICH) embedded into cultural, and natural heritage artefacts, sites or monuments. The definition excludes ICH related to other cultural domains such as festivals, celebration etc. It covers industrial heritage and cave paintings.” – UNESCO Institute for Statistics, 2009 UNESCO Framework for Cultural Statistics
Humanitarian Assistance – “Defining humanitarian assistance Humanitarian assistance is intended to save lives, alleviate suffering and maintain human dignity during and after man-made crises and disasters caused by natural hazards, as well as to prevent and strengthen preparedness for when such situations occur. Humanitarian assistance should be governed by the key humanitarian principles of: humanity, impartiality, neutrality and independence. These are the fundamental principles of the International Red Cross and Red Crescent Movement (RCRC), which are reaffirmed in UN General Assembly resolutions and enshrined in numerous humanitarian standards and guidelines. In this report, when used in the context of financing data, humanitarian assistance refers to the financial resources for humanitarian action.” – Defining Humanitarian Assistance
Poverty – “Poverty entails more than the lack of income and productive resources to ensure sustainable livelihoods. Its manifestations include hunger and malnutrition, limited access to education and other basic services, social discrimination and exclusion, as well as the lack of participation in decision-making” – United Nations, Ending Poverty
Volunteering Abroad – “An international humanitarian trip where you travel, generally to developing countries, around the world to volunteer your time, skills or labor.” – IFRE Volunteers, The Ultimate Guide To Volunteer Abroad
