techcrisp

Significant issues facing the Global South and how AI can help.
Poverty, Food and Hunger, Health and Education
Water, Energy, Climate Change, Bio-Diversity, Democracy, and Materials we cannot do without

*|MC:SUBJECT|*

Article flow is:

Developments in AI

Significant issues facing the Global South and how AI can help

  • Poverty, Food and Hunger, Health and Education

  • Water, Energy, Climate Change, Bio-Diversity, Democracy, and Materials we cannot do without

Conclusion


Developments in AI

Since OpenAI launched ChatGPT in late 2022, there has been a significant increase in the market value of some technology companies associated with AI. Seven of these companies, listed in the USA, have gained a combined market capitalization of over US$ 13 trillion. This amounts to a 25% share of the US stock market, with a total market capitalization of US$ 51 trillion. The global market capitalization of companies is around US$ 109 Trillion. The combined revenue of these seven companies is US$ 1.77 trillion compared to US GDP of US$ 28 Trillion and global GDP of US$ 104 trillion.


Currently, AI solutions available are buggy in a giant sandbox development phase. A handful of, no doubt, brilliant technologists and financial wizards in Silicon Valley and Wall Street have allocated trillions of dollars, anticipating huge and fast returns despite little convincing data in hand. These numbers need coherence; they do not simply add up!


OpenAI (OpenAI valued at US$ 150 Billion), the agent provocateur of the AI revolution, projects a revenue stream of US$ 2 Billion in 2024. Microsoft depends on OpenAI technology. Google, Meta, Amazon, Nvidia, Elon Musk’s xAI, and Apple are offering or launching AI solutions. There is an extensive list of start-ups that have entered the fray. As per estimates in the “ State of AI Report 2023,” over US $ 234 billion of funding has flown into AI globally from 2021 to 2023.


Present AI technology is based on LLMs (ChatGPT, Gemini, Inflection, Anthropic, Mistral), which are large neural networks that work on statistical modeling. However, they lack experiential learning and tacit knowledge possessed by domain experts. They cannot access millions of peer-reviewed research papers on Science.org, Nature, Lancet, and many specialist journals and books. Currently, the data used in AI solutions is not validated and lacks quality.


Recent news reports that Google has partnered with Reddit to use its data for $60 million. Substack is also exploring this model.


You must undoubtedly be using one or more LLMs or CoPilot from Microsoft. While LLM bots can be helpful in general-purpose research and comprehension, CoPilot can improve the quality and productivity of a Microsoft Office power user. Neither LLMs nor CoPilot can improve an automobile company’s design, manufacturing process, or revenues at this stage. Present gains shall be limited to increased productivity of white-collar office workers.


According to OpenAI and Microsoft, approximately 18,000 companies use ChatGPT on Azure. This is a positive sign. It is essential to have early adopters who can experiment with new technology, conduct proof of concepts, and then mainstream it within their organization. Meanwhile, other organizations wait for success stories before diving in. This process can take 1-3 years to mainstream for early adopters and 3-7 years for others.


Presently, LLMs are the predominant solution available in AI. Domain-specific tools are being created using various mathematical modeling tools and AI algorithms.


Artificial intelligence (AI) has become crucial in the ongoing technology race between the USA and China. In the past year, their governments have pledged between $40bn to $50bn each for AI investments. Other countries are investing in AI to avoid being left behind or relying on foreign-controlled technology. In 2023, six more countries – Britain, France, Germany, India, Saudi Arabia, and the United Arab Emirates (UAE) – have committed to collectively investing around $40bn towards AI. This investment will mainly focus on purchasing graphics-processing units (GPUs) to train AI models and build factories to produce such chips. New AI-industrial complexes are taking shape.


What are the significant issues facing the Global South? How will AI assist in solving them?

Poverty

The world population is 8 billion, projected to increase to 10 billion by 2050.

10% of the World lives in Extreme Poverty – daily spending available is US$ 2.15 per day, also known as the Extreme poverty line.

25% live in poverty per Lower Middle Income criteria- Poverty Line at US$ 3.65.

47%  live in poverty per Upper Middle Income criteria- Poverty Line at US$ 6.85.

85%  live on less than US$ 30 per day- the Poverty Line for Developed Countries.

According to the World Bank classification, economies are classified by Gross National Income (GNI) per capita between certain thresholds. As of 2021, the following classification has been made:

Low-income economies: GNI per capita of $1,045 or less.

Lower-middle-income economies: GNI per capita between $1,046 and $4,095.

Upper-middle-income economies: GNI per capita between $4,096 and $12,695.

High-income economies: GNI per capita of $12,696 or more.

In the 1800s, most people in the world lived in extreme poverty. Various social and technical revolutions have reduced that number to 10 %. The AI revolution builds on developments of the last 200 years.

Satya Nadella, Microsoft CEO, was in India recently and publicly announced that India’s US$3.5 Billion GDP will move to US$5 billion in 2025, with AI contributing US$500 Billion. Coming from a person of his stature, I will not discount his projections and, for the moment, share his audacious view on a 10 % growth in GDP attributed to the adoption of AI.


Based on that premise, the Global South per capita incomes should move closer to present Middle-Class Incomes in the next five years if AI is adopted in earnest. We should move into the US$ 7,000- US$ 10,000 per capita GDP bracket for at least 80 % of the Global South population.


Climbing out of poverty is akin to climbing a ladder with greased rungs. The four essentials are adequate nutrition, good health, relevant education, and availability of employment opportunities. Let us go slightly deeper.


Food and Hunger


According to the World Economic Forum, 10% of the world’s population is undernourished, while 2 billion people are deficient in micronutrients. Researchers claim that the average cost of a nutritious diet worldwide is $3.54 per day. Three billion people need proper nutrition. By 2050, humans will need 60% extra food. Business as usual will not work.


AI and gene editing technologies can create fast-growing, productive, nutritious, disease-resistant seeds that need less water and fossil fuel-based nitrogen fertilizers. AI can also discover microorganisms that can fix nitrogen for plants directly. AI can help create better-quality, productive, fast-growing livestock, poultry, and fish, including dairy and egg production.


A recent study in India observed that farmers receive a meager percentage of the end product sold. The supply chain has evolved so that intermediaries, processors, and retailers receive the majority of revenues.


Supply Chain optimization on a local, regional, and global scale is an issue in the Food sector. With a worldwide revenue of US$ 5 Trillion and half the world’s population relying on the farm sector, AI can highlight issues in the Supply Chain and evolve solutions where farmers get an adequate share of the pie.


Field trials, Regulatory approvals, and the local Government’s role in extension work—providing seeds and other inputs, including finance, to poor farmers—are time-consuming, bureaucratic, and corruption-ridden. Technology must accompany country-specific social, political, administrative, and economic reforms. The affordability of wonder seeds can be a significant issue.


This would also require specialized AI tools to analyze massive local data sets on soil, weather, irrigation, etc. A large ecosystem of start-ups in various countries would be needed to create solutions based on local requirements rather than relying on companies such as Monsanto or Cargill who only provide seeds.


Health


Life expectancy. Africa-62; Asia-70; USA / Developed economies- 80 years.

Child Mortality: Low Income-6%, Lower Middle-4%, Upper Middle- 1.5%, and High-Income economies- less than 1%.


DALY’s Nos. – The most common way to measure the burden of disease is to estimate the number of years of life “lost” due to poor health, which is the so-called loss in “Disability Adjusted Life Years” (DALYs).

DALY index is around 20,000 in developed countries, 37,800 in India, and 40,000-70,000 in Africa.


More than half the world’s population is deprived of basic health amenities. The Global South has a tremendous shortage of doctors, affiliated healthcare workers, and infrastructure.


AI can upgrade primary and tertiary health providers with timely and accurate diagnostic tools, suggest medical treatment, and monitor progress. Such tools are available now on Mobile phones. In the case of advanced diagnostics of X-rays, MRIs, and Blood tests, it has been proven that AI can do a better job. In the long run, humans can have a personal AI Doctor Chatbot for routine medical advice and monitoring.


Drug discovery is an expensive exercise. Billions of dollars and many years are needed for costly clinical trials, regulatory approvals, and education of health care providers. Big pharma has to focus on first-world problems for their financial health. Consequently, the cost of drugs for many diseases is just too high. Most patients in the Global South cannot afford life-saving medicines.


AI can reduce the cost and time of drug discovery and go to market. AI can help focused, specialized pharma companies that incur lesser costs and provide customized drugs. This would need more start-ups with adequate funding in different countries or regions.


Education and Employment


8%, 58 million of the world’s 787 million primary school-age children are not in school.


After primary school, a significant percentage of children in Low Income (90%), Lower Middle (55%), Global Average (48%), Upper Middle (29%), and High-Income economies (9%) struggle with reading with proper comprehension.


86% of the world’s population are literate with some education. More than literacy is needed to provide employment, especially in the Knowledge economy.


According to data from UNESCO, as of 2019, around 7% of the global population aged 25 years and older had completed tertiary education (including college and university education). This figure varies widely by country and region, with higher tertiary education attainment typically found in wealthier countries.


It is estimated that 235 million, or 6.5 %, of the global labor force is unemployed. The ILO claims that 3.3 billion people who are technically employed are working under substandard conditions that offer too little pay, poor economic security, and little to no opportunity for advancement.


The Global South has a significant Primary and Tertiary education issue due to a need for more infrastructure, teachers, and tutorial fees. The quality of education and skilling has to be higher, and most qualified graduates are unemployable.


Education is the magic potion for succeeding in the Knowledge economy and remaining relevant. Technology moves at the speed of light. Automation has, in the past, put people out of jobs in agriculture and industry. Productivity gains introduced by Technology have gone to Capital holders rather than Workers and continue to create stark economic inequalities. The Global South shall have to educate and upskill its workforce to reap the benefits of technology.


AI can redefine pedagogy and tailor learning for an individual. Mobile phones and the internet can reduce reliance on expensive education infrastructure. The initial success of ChatGPT and other chatbots has been in education.

As in other domains, particular purpose AI tools must be developed for different domains, skills, and learning paths. AI Chatbot acting as a personal trainer can upgrade knowledge and skills. AI can redefine learning in a big way for the Global South.


Based on my limited research, Unemployment is also caused by a Demand-Supply information mismatch. For example, I tried to find a Video Editor who could edit videos for YouTube. This job is part-time; one does not need a high school or college education. It takes about a month to learn open-source software and another month of practice. The equipment required is a PC. And you can be sitting anywhere. I used various sites to find an individual who could provide the service. Surprisingly, I could only connect to a few people across India, while I am sure tens of thousands of people have these skills.


AI can provide tools to match individual skills with potential employers. It can also continuously search for opportunities that potential contractors or employees need to take advantage of.  


AI can help humans lift themselves out of poverty and transition to a middle-income scenario through radical changes in agriculture, life sciences, health, and education. This would take 3-10 years to show results, provided significant social, political, administrative, and economic reforms accompany the initiative.


What are the other significant issues facing the Global South? How will AI assist in solving them?


Water


2.5% of the world’s water is fresh, but only 30% is usable. 70% of the freshwater is consumed in agriculture, 22% in industry, and 8% for domestic purposes. The daily water consumption per person varies depending on the country and economic strata. In developed countries, it can be as high as 100 liters per day, while in underdeveloped and arid countries, it can be as low as 3 liters per day.


By 2030, the worldwide water demand is estimated to be 6900 billion cubic meters, while the supply is expected to be only 4200 billion cubic meters, resulting in a 40% deficit. India, which has 18% of the world’s population, has access to only 3% of the world’s fresh water and is already in the midst of a water crisis. The Global South has a projected deficit of 50% or more by 2030.


AI can help reduce water wastage in various sectors, especially agriculture. Modeling of water basins and optimizing distribution are essential areas. Predicting rainfall and identifying retention areas to prevent run-off can conserve water. New technologies for water treatment and recycling are another area.


Energy


The majority of Primary energy comes from Fossil Fuels, Which also generate the majority of electricity. Energy is the driver of every economy, so energy use is increasing in the Global South. For example, India has a 7 % annual growth rate in power consumption.


There is a global initiative to move to Renewables at a fast pace. The transition from Coal, Oil, and Gas will not meet the expectations set up by IEA and others. The sunk capital cost of fossil fuel extraction, transportation, and use has been written off. New investment in Renewables will have to consider that. Most countries in the Global South plan to use Coal and Oil & Gas for at least 25-30 years. Essential materials like cement, steel, plastics, and ammonia have processes that need fossil fuels only.


The fossil fuel industry generates trillions of dollars in revenue. It sustains nations and determines geopolitics. Many countries in the Global South, including Africa and South America, have substantial Oil and gas reserves.


Renewables have a high implementation cost; they are intermittent and need cost-effective storage to run the power grid. Hydrogen as a clean fuel shows promise but is still nascent, with storage, transportation, and price issues.


How can AI help reduce Fossil fuel usage and transition to Renewables economically?

AI can work on making transportation, manufacturing, and services more energy efficient.

AI can discover new materials, reducing dependence on Plastics, Petrochemicals, and Nitrogen fertilizers.

Material discovery also involves superconductors, new battery chemistries, efficient processes for producing Hydrogen, Biofuels, and technologies for Carbon Capture and Sequestration.


Climate Change and Bio-Diversity


The Paris Accord set a target of restricting temperature rise to a maximum of 2 degrees Celsius and zero carbon emissions by 2050. The investment needed is in trillions of dollars per year and requires a significant technology shift to renewables, battery storage, hydrogen, and  CO2 capture/sequestration.


The effects of climate change on the Global South is huge. Weather patterns affect agriculture; droughts, floods, and increasing sea levels displace people. The Global South does not have the money to invest in mitigation solutions.


The Living Planet Index (LPI) measures the average change in the number of individuals across the world’s animal populations. The Living Planet Index indicates an average relative decline of 69% since 1970. I am sure the situation in the plant world is even worse. Humans are intimately linked to every plant and animal species for survival. Biodiversity has to be protected, especially when it comes to agriculture. Bees and insects pollinate plants. The felling of one tree can affect more than 100 species in the ecosystem.


AI can help create better climate models and identify areas where optimizing materials, processes, and supply chains can reduce fossil fuel usage. AI can model and predict areas in the Global South where Climate Change can cause maximum damage. At least those countries will be able to triage and save vulnerable regions.

Where elimination of plant and animal species is concerned, since AI can handle large volumes of multi-modal data, models based on satellite remote sensing, GIS, real-time sensors, and on-ground data on deforestation, urban sprawl, changes in hydrology, etc., can provide visualization to various stakeholders’ about localized threats, raise alarms, and provide suggestions on mitigation.


Human Rights and Democracy


The V-Dem Index ( v-dem.net ) captures the extent to which people are free from government torture, access to justice, political killings, forced labor, property rights, and freedom of movement, religion, expression, and association. More than 50 % of the global population has an Index of 0.5 or less. Developed democracies like the USA and Western Europe enjoy an Index of 1.


Freedom of thought and expression, justice, equality, and secularism are essential for human well-being. The image below (courtesy of Economist magazine) starkly reminds us of today’s situation.


AI can be a positive tool to ensure that focused and correct information is disseminated to individuals. In the long run, the masses create social and political change no matter how disempowered an individual may be. Democracy has to survive and flourish in the Global South. Otherwise, fundamental problems will remain.


Materials We Cannot Do Without


In 2019, the world consumed about 4.5 billion tons of cement, 1.8 billion tons of steel, 370 million tons of plastics, and 150 million tons of ammonia. These depend on fossil fuels, and their demand shall grow.


Global production of these four materials claims about 17 percent of the world’s primary energy supply and 25 percent of all CO2 emissions originating in the combustion of fossil fuels.


In the move to Renewables, Copper, Cobalt, Nickel, and Lithium have become essential inputs.


They are not readily replaceable by other materials—certainly not shortly or globally. No commercially available and readily deployable mass-scale alternatives exist to displace these established materials or extraction and manufacturing processes.


As the Global South grows, demand will increase. It is impossible to meet the increasing demand sustainably over the next 25 years.


AI can help optimize extraction and manufacturing processes and supply chains. Most importantly, discovering new sustainable materials is an area suitable for AI.


Conclusion


  1. At first glance, AI implementation seems like a simple, no-brainer solution. However, AI data centers use expensive, high-end technology. AI needs a huge amount of validated data, which is difficult to collect and validate. Validation is time-consuming and mind-boggling. Ultimately, training the model is complicated, specialized, and costly.


  2. As mentioned above, Global South problems are not just IT-related; they are complex and interlinked in a complicated mesh of historically inherited problems, regional conflicts, and unstable social, political, and administrative networks.


  3. AI will need to be accompanied by social, political, and administrative reforms to positively impact the resolution of issues faced by the Global South. Depending on the problem being confronted, this may take months to years.


  4. While IT solutions such as setting up an ERP system may be implemented within a company’s premises by trained professionals, AI solutions are not limited to the Data Centre alone. They must be implemented in the real world, and their impact must be measured through transformation. This can be a challenging and strenuous process as AI solutions touch the real world.


  5. Silicon Valley-based companies spearheading the AI revolution have to give returns to their shareholders. Their solutions may become too expensive for mass penetration in the Global South. For example, CoPilot’s annual subscription sells for US$290 in India, which is too expensive for the average user.


  6. Setting up AI data centers of scale and training models requires billions of dollars. This infrastructure must be local for data security. Financial returns will take time. Investment must come from either the Government, which has its own problems, or the private sector. The Global South will struggle to obtain finances.


  7. AI implementation skills must be indigenous to create solutions with local context and on-ground implementation. Unfortunately, capability creation has yet to start. Skilled resources prefer to migrate to the Western world.


  8. The creation and absorption of local domain-specific solutions will take many years. VC funding will be needed. Economic benefits in most areas will take 3-10 years to fructify.


  9. The Global South should adopt AI and go through the learning curve. Low-hanging fruits such as education and health should be chosen first, where implementation is more manageable and provable.


  10. Will AI be a Chimera? Will Satya Nadella’s optimism about rocketing GDP ring true? Is all this going to be a pie in the sky?


  11. The Global South should dive in, create a large pool of capability, and start implementation with limited resources supported by long-term financial and development institutions for the first few years.


Comments

Leave a Reply

Your email address will not be published. Required fields are marked *