479
AI cannot benefit the world if technology leaders ignore 85% of humanity in the Global South, writes Wharton’s Cornelia Walther.
The following article was written by Dr. Cornelia C. Walther, Visiting Scholar at Wharton and Director of the Global Alliance POZE. Walther, a humanitarian practitioner who spent more than 20 years at the United Nations, is currently working with the United Nations in Morocco and the Sunway University Center for Planetary Health in Malaysia to develop a national blueprint for prosocial AI to be designed, delivered, and deployed to benefit people and the planet. Her new book, Artificial Intelligence for Inspired Action (AI4IA), will be published in February 2026.
We are the last analog generation, the last population to remember what life was like before algorithms began mediating our relationships and choices. The first iPhone was released exactly 18 years ago. Today, artificial intelligence is changing the way we think, make decisions, and experience reality, from dating to credit decisions to medical diagnoses. This unique position gives us an important perspective and allows us to compare the world before and after the explosion of generative AI. This gives us a compelling opportunity to forever shape our hybrid future and leave behind an algorithmic architecture for future generations that helps people and planet thrive. However, current trends do not match that vision.
What’s the problem?
Three major paradigms of AI are emerging. They are the US market-driven model, China’s state-coordinated approach, and the European Union’s regulatory framework. But for 85% of humanity in the Global South, none of these models adequately addresses the interrelated challenges of development, dignity and sustainable growth.
They are asked to choose a side, but none of the options are really helpful.
Why is this happening?
This gap exists because current AI developments, largely driven by the Global North, overlook important dynamics. The ABCD of this underappreciated AI problem affects everyone, but the Global South has the most to lose and perhaps the most agility to find solutions. The Global North largely set the conditions for the first three industrial revolutions, but it is often locked into a model that has left our global society in a precarious place. This Fourth Industrial Revolution cannot be left to the private sector alone. It must be co-created by all parts of society: public, private, academic and private, across all continents.
Agency decline:
As we delegate more of our cognitive effort to AI, our ability to make independent decisions weakens. Personalization systems limit our exposure to diverse perspectives and artificially reduce the range of choices we have for everything from news to job postings.
Bond erosion:
AI-mediated interactions and digital companionship can reduce empathy and increase polarization. In regions where strong community ties form the primary social safety net, this algorithmic isolation has existential consequences.
Climate challenges:
The environmental footprint of AI systems is growing. From the emissions generated by training large language models to the use of land for data centers and the water needed to cool them, we are witnessing a dynamic in which the needs of AI systems are prioritized over human needs. This uncertain and growing burden affects the very countries least responsible for their historic emissions. Moreover, they face pressure to implement technologies that not only exacerbate the serious challenges they face, but also often do not meet their needs.
Divided society:
When 98% of AI research comes from wealthy institutions, the resulting systems are embedded with assumptions that are irrelevant or harmful elsewhere. This is digital colonialism by code. With billions of people in the Global South still struggling to obtain basic necessities and huge amounts of money being poured into generative AI, we must ensure that the resulting advances benefit those most in need.
How can we build a better alternative?
This is not a new problem. In 1955, 29 newly independent countries convened in Bandung, Indonesia, for a purpose that skeptics dismissed as impossible: to create an alternative to Cold War-era polarization. Rather than choosing sides, leaders like Sukarno, Nehru, and Nasser formed the Non-Aligned Movement (NAM).
NAM demonstrated that 120 countries can unite around common principles of sovereignty and mutual benefit and influence international affairs through solidarity.
Today, we need a “Fourth Way” of AI inspired by the same spirit. This path is prosocial AI. That is, a system that is intentionally tailored to local conditions, trained on representative data, tested for impact on equity, and aimed at collective flourishing. This approach offers a different compass: serving people, serving the planet, and contributing to possibilities.
This “4T framework” is already in practice.
Customization: AI systems should reflect local language, context, and values. Kenya’s M-Pesa revolutionized financial inclusion, not by copying Western banking apps, but by understanding how money moves through social networks in a cash-based economy.
Trained: If a facial recognition system is trained on primarily light-skinned faces, it will fail significantly on darker-skinned people. Prosocial AI requires representative data, fair compensation for the communities that provide the data, and inclusive development teams.
Tested: Evaluations should not only assess technical performance, but also social impact. Will this financing algorithm perpetuate discrimination? Will this transportation optimization sacrifice environmental justice for efficiency? Testing must ask who it works for and who it might harm.
Targeted: The system must aim for collective prosperity. AI managing a nation’s power grid can aim to optimize purely for cost, or simultaneously balance equitable access, renewable energy integration, and climate resilience.
NAM’s greatest strength was South-South cooperation. This principle is the catalyst for prosocial AI. This leadership is already taking shape at the local level. The African Union (AU) and the Association of Southeast Asian Nations (ASEAN) are both developing continental AI strategies. Unlike dominant paradigms, these frameworks are built from the ground up to prioritize inclusive growth, data sovereignty, and solutions to common challenges.
Within these regions, countries like Malaysia and Morocco have emerged as potential catalysts. Countries with middle-income economies, young and digitally native populations, rich linguistic diversity, and long cultural heritages are uniquely positioned to champion the Fourth Way. By fostering a vibrant, values-based AI ecosystem, we can demonstrate why and how the successor to the Sustainable Development Goals (SDGs) should be the Hybrid Development Goals (HDGs). Hybrid Development Goals (HDGs) are goals that aim to harmonize with technological progress and human prosperity while preserving the dignity of the planet.
These efforts show that unity turns weakness into strength. If Nigeria develops an AI system to diagnose malaria and shares its methodology with Indonesia, it could benefit from collaborative learning in a way that is not possible in a competitive commercial framework. While individual developing countries cannot dictate their terms with the tech giants, 120 countries representing 85% of humanity can claim multiple seats at the table on data sovereignty, algorithmic transparency, and governance.
The moment to choose hybrid humanity
The algorithmic architectures that are now taking shape will shape humanity for decades in tangible and immaterial ways. It not only affects how we think, feel, choose and interact, but also what decisions are made by whom, for whom, and about what. In that context, the question is not so much which country will gain AI advantage first, but how we can structure hybrid communities that offer every human being, every country, every continent the opportunity to fulfill their unique potential, without endangering the ecosystems on which future generations will depend. Can we (and are we willing) to shape a system that best respects humanity, or can we embrace a system that is optimized for user engagement and profitability?
Seventy years ago, Bandung’s leaders declared the impossible possible. Today we can declare that a fourth way is possible. It’s an approach to AI development that advances people, the planet, and potential.
