AI in the Global South
What impacts - positive and negative - are we seeing? And how can MPs harness the benefits, mitigate the risks, and contribute to global AI governance?
This post introduces issues around AI in the global south, its potential, specific risks and governance challenges. Like other transformative technologies, AI will proliferate across borders and reap global benefits. However, after over a decade working with parliaments in the Asia Pacific region, I’ve seen the negative impact of the emerging technologies, introduced rapidly and misunderstood by lawmakers.
The current debate on AI governance is slanted towards North America, Europe and East Asia where industry is concentrated. However, the effects of AI will play out very differently across the world and lower and middle income countries (summarised using an admittedly clumsy term as the ‘global south’) are likely to have specific needs and concerns. In this post, we’ll also look at the role that parliaments and international development support can play to address these.
Looking first at potential benefits of AI in the Global South, the conversation has focused on development gains. Where agriculture is the primary source of livelihoods, AI can improve decision-making and increase yields, contributing to food security and economic development. For example:
AI is helping identify high performing plants and animals, with potential to “feed 10 billion people by 2025” in the face of threats to food security from climate change and new pests and pathogens.
Farmer.chat, a generative AI application, is supporting better agricultural knowledge in local languages across India, Ethiopia and Kenya. In India, KissanGPT gives farmers guidance on irrigation techniques, pest control, and crop cultivation.
Also in Kenya, AI supports early diagnosis of disease in crops including cassava. In Trinidad and Tobago, Crop Mate provides farmers with real-time AI-powered information about soil conditions and nutrients.
Beyond these applications, in future drones and sensors on farms can collect data and support precision agriculture, such as analysing soil conditions and predicting optimal inputs. Satellite and drone data can also help anticipate shortages based on patterns of settlements and vegetation and help predict fires and floods.
In certain countries, education and health services suffer from a lack of infrastructure, qualified practitioners and poor access in remote areas. How can AI help? In education, AI gives potential for more personalised tuition, can expand education services to remote areas and upskill workers across sectors. An example is Daptio in South Africa which uses deep analytics to provide personalised learning to teachers and students.
In healthcare, AI systems can support improved triage and diagnostics where access to medical advice is difficult. For example, in Kenya and South Africa AI-powered remote ultrasound machines are helping identify high-risk pregnancies.
AI can also help improve governance outcomes in the global south. AI-powered applications are providing new means for the public to access services. In India, MyScheme is available on WhatsApp and Telegram to help citizens access complex government schemes, and generative AI is helping people access legal advice and engage with authorities. In Kenya, an AI Chatbot helps people interact with local authorities to prepare for and recover from natural disasters. At the policy level, AI is being trailed by the African Centre for Economic Transformation to help improve economic forecasting and decision-making, with potential to contribute to more inclusive and sustainable economic outcomes across the continent.
These examples of AI applications are exciting. They can pave the way for innovation, empowerment, and inclusive development and contribute to achieving the SDGs. So why am I not feeling more optimistic?
Firstly, for all the beneficial applications, there are also numerous examples of inappropriate or dangerous technologies developed in advanced economies being tested in countries with weaker regulatory guardrails and oversight.
There is the example of Lendtechs across countries in the global south, which take advantage of data protection gaps and raise issues of reliability. Investigations have shown that they can use micro-behavioural data points - selfies, videos, apps installed, typing speed – to determine the credit-worthiness of the users, even exploiting borrowers’ contacts to call family and friends about loan repayment. Reports suggest that loan apps have plunged many Kenyans into deep debt and pushed some into divorce or suicide.
In South America, the human rights project NotMy.ai, has mapped 20 AI-related schemes likely to stigmatise or criminalise the most vulnerable people, developed together with software companies in the global north. In one case, an AI-based system introduced in Argentina forecast the likelihood of teenage pregnancy based on data such as age, ethnicity, country of origin, disability, and presence of hot water in the house.
The case of Worldcoin showed how companies in the global north are seeking to obtain training data for AI in countries with fewer legal protections. It promised to enable a cryptocurrency-based universal basic income by collecting the biometric details of users, however it was criticised for using deceptive marketing practices, collecting more personal data than it acknowledged, and failing to obtain meaningful informed consent. It violated privacy and data collection laws and ended up being banned in Kenya.
These examples demonstrate issues of privacy, agency and consent in the global south. We should also raise questions around ‘technology solutions for the developing world’ as companies that develop AI products also often lack significant incentives to produce high-quality datasets that truly reflect local needs. AI technologies created for but not by lower and middle income nations, with data stored predominantly in the global north, risk exacerbating power imbalances and bias by promoting products that are divorced from the realities on the ground.
“All this data that is being generated through AI is data that is freely being collected from our countries and monetized …. there are also issues of manipulation, especially for us as politicians. We’re talking about politics, in terms of the impact on democracy. There are also issues on enhancing the gender inequality because unfortunately, the current data that has been fed is already gender-biased.” Tanzanian MP Neema Lugangira
Secondly, in its current state, the global AI industry risks negatively impacting the labour market in the global south. Low-wage and low-skill workers in countries without strong labour protections are already seeing the effects. Generative AI is already impacting workers including in India, Pakistan, and Bangladesh as some of the largest hubs for freelancers on online gig work.
We are also seeing labour exploitation in the AI supply chain. While AI producing countries are benefitting by developing and deploying AI systems that enable economic growth, there is a rise in the global south of industries that engage cheap, low-skilled workers to perform data labelling and correction. This industry is expected to reach $17.1 billion by 2030 and is part of AI’s ‘vast underbelly’. In the Philippines, more than two million people are working to annotate datasets and in Venezuela, they are paid an average of 90 cents an hour. In Kenya, workers paid as little as $1.50 an hour are having to flag inappropriate content including videos of murder and rape and report deep psychological impacts.
“The tech companies will behave extremely well within the EU, within the U.K., within the U.S. … But the same companies, when they come to Africa, their behavior takes a 360-degree turn and they behave accordingly, only because we don’t have the same laws as the other Global North countries.” Tanzanian MP Neema Lugangira
For the foreseeable future, countries which are constrained by limited funds and technological expertise will need to rely on a select group of affluent corporations and nations for access to cutting edge AI. This threatens to exacerbate inequality and the digital divide. While there are forecasts of AI adding $15.7 trillion to the global economy by 2030, countries in the global south will experience more moderate increases and the IMF suggests a severe widening of the gap between developing and advanced economies as a result.
Finally, AI technology also has potential to undermine democratic development. AI-powered surveillance technologies and predictive analytics can help increase security, safety or traffic control, but the same capabilities be used to control citizens through blanket surveillance, to stifle free speech, crack down on protest and spy on political opponents. These capabilities may be increasingly tempting to governments in fragile democracies seeking to consolidate control.
Generative AI’s potential to supercharge disinformation may also be felt even more in the global south where lower levels of digital literacy and a less robust press may struggle to push back. I have witnessed the effect that disinformation spread on social media can have on interethnic conflict and violence while living in Myanmar and in Ethiopia, and on democratic debate in Kenya.
When practiced cautiously, certain AI applications can strongly benefit the global south. However, the cases above point to a trajectory that could threaten inequality, social unrest and democratic regression. So what might be required for the global south to reap the benefits and mitigate the risks? And how can democratic institutions and elected representatives in the global south take action?
Starting at home, by articulating a vision for a future with AI through national strategies. While many countries currently lack the resources for a robust local AI industry, these strategies can shape the global AI discourse and attract international aid to nurture homegrown AI solutions tailored to local contexts. They can be based around a local understanding of ethics and responsible use, with UNESCO’s recommendations providing a basis, integrated into national development plans and aligned with SDG priorities. Non AI-producing countries can also craft regulatory responses focusing on consumer protection, data governance, and effects on the labour market, social equality and job security, as seen in examples from South America. Engaging the public and civil society in forming national responses, and in ongoing oversight of AI deployment will be critical.
As an MP, what can I do?
MPs in the global south can take the lead in pioneering national AI governance models tailored to their realities. They can:
Craft, debate and pass legislation to implement national AI strategies, including on ethics and data protection.
Stay updated on the rapid changes in AI development and deployment, including building connections with sources of expertise such as tech hubs, universities and entrepreneurial groups.
Incorporate AI oversight and policy into the work of parliamentary committees, which can investigate and publicise issues around AI deployment and engage diverse stakeholders.
Educate and informing the public and constituents on AI awareness, identifying gaps in understanding in areas such as data privacy and algorithmic bias.
Advocate for increased budgetary allocations to AI development, monitoring and audit, and for public AI literacy efforts.
Prioritise identifying and consulting vulnerable groups of society who might be disproportionally impacted by AI.
Working internationally, countries in the global south can contribute to a vision and universal principles for AI governance. At present, debate on AI governance is mostly centred in the global north, and countries in the global south should not just be onlookers. Without active participation in global governance discussions, inequities are likely to persist in how AI is built and deployed globally. A first step is building partnerships - sharing best practices, collaborating on regulations, and addressing common concerns across jurisdictions. Various initiatives are underway, with some links below. Regional bodies such as the African Union and Association of Southeast Asian Nations also have an important role to then play in pooling resources and sharing expertise and providing opportunities for a collective voice in a global AI governance discussions.
As an MP, what can I do?
Inter-parliamentary exchanges are going to be an important element in crafting inclusive and responsive global AI governance. MPs can:
Familiarise themselves with international AI governance debates and emerging frameworks, pushing for representation through national and regional channels.
Harness parliamentary diplomacy and membership in international bodies such as the Inter-Parliamentary Union to raise issues around AI development and deployment and advocate for AI governance that is responsive to local realities and needs.
Catalogue and share lessons learned on the impact of AI in society, across sectors and on the labour market to help inform global AI governance.
Urge regional bodies to assemble expertise to provide policy and regulatory direction to guide member countries.
Call to action for international development
International development support has a critical role to play empower global south parliaments to harness the potential of AI and mitigate risks. It can provide:
Capacity development for MPs and parliamentary staff on technical aspects of AI and its implications. This can better inform national policy and legislative responses and enable stronger participation in global governance discussions.
Platforms to convene elected representatives for South-South exchange and to tap into networks of expertise including technology companies, civil society and grassroots organisations. This can include networking across parliamentary committees tasked with overseeing the effects of AI deployment.
Public education and AI literacy including in local languages and for diverse and marginalised groups.
MPs can also seek international support for international investment in local AI infrastructure, helping with technical expertise and data infrastructure to develop appropriate, home-grown AI. In this way, countries in the global south can be assisted to reap the benefits from the AI revolution.
Examples of international and regional initiatives
The Igarapé Institute and New America Global Task Force on Predictive Analytics for Security and Development convenes digital-rights advocates, public-sector partners, tech entrepreneurs, and social scientists from the Americas, Africa, Asia, and Europe to define principles for the use of predictive technologies in public safety and sustainable development in the Global South.
SMART Africa Alliance is a flagship project of South Africa in collaboration with German Development Cooperation, helping develop policy frameworks ready for AI.
The World Economic Forum’s National AI Strategy Peer Network provides a platform for policymakers and practitioners to share lessons, challenges and best practices on AI.
The BRICS group of developing countries has created a committee to study the implications of generative AI and “track and evaluate the development and evolution of AI technologies”.
The Virtual Parliamentary Exchange on Artificial Intelligence and Digital Rights by ParlAmericas Open Parliament Network has discussed the effects on social well-being of AI. A digital caucus is developing legislative proposals on AI.
The African Parliamentary Network on Internet Governance comprises 35 lawmakers from 30 countries who aim to incorporate African voices in global discussions on digital policy, including on reining in artificial intelligence.