Why AI might widen Asia's inequalities

What is Artificial intelligence?

Artificial intelligence, or AI, is a beguiling term, which means different things to different people. It is a catch-all phrase that groups together a variety of unrelated technologies. In other words, AI is not one thing!

In a recent report on AI in the Asia-Pacific, the United Nations Development Programme (UNDP) focuses on “generative AI” (GenAI). This refers to AI systems that can create new content (text, images, music, code, and more) that is often indistinguishable from human-generated content. They note that GenAI is a “general-purpose technology”, and is likened to general purpose technologies like the internet or electricity.

How can we say that a machine was intelligent? In 1950 Alan Turing proposed a test (the Turing test) for assessing if a machine was intelligent. A machine passes the test if a human cannot consistently tell the difference between its responses and a human’s. ChatGPT, released in November 2022, passed the Turing test and was the fastest adopted technology in history – one million users in five days (GPT stands for generative pre-trained transformer).

Compared to previous machine learning neural network models, transformers can much more efficiently “learn” by training on vast datasets in parallel, more effectively identifying complex patterns and underlying structures in human language, visuals, or other inputs.

How do AI models operate?

In essence, a GenAI model builds an internal model of the world (or whichever domain it is trained on) to generate novel outputs following the patterns of that world. A GenAI model can produce a plausible essay on sustainable farming, develop a new recipe, write and debug code, or hold a conversation in the style of a particular person.

What makes GenAI (and AI broadly) different from other previous general-purpose technologies is how it augments or automates cognitive work and creativity, with some referring to it as functioning as a new method of invention.

Tasks that previously required human intelligence – from composing an email, to translating a paragraph, or identifying patterns in a medical image, or predicting market trends, can now be done in part or whole by AI systems. This doesn’t mean humans are obsolete, but it means our productivity in these tasks can skyrocket.

How could the Asia-Pacific benefit from AI?

The Asia-Pacific could potentially reap great benefits from AI. Developing countries could use AI to help transform education, healthcare, and agricultural management.

AI can widen people’s freedoms and capabilities. In classrooms it can tutor in local languages and lighten teachers’ workloads. In clinics it can read images, triage symptoms, and support overstretched staff. In social protection it can spot needs faster and efficiently. In climate and biodiversity, it can turn satellite, acoustic, and sensor data into early warnings and smarter stewardship.

But the Asia-Pacific region is hugely diverse, both between countries and within them. And a few advanced economies are positioned to capture most of the gains, while others still struggle with power, connectivity, and skills.

Risk of the Next Great Divergence

Asia and the Pacific faces a clear choice: build an inclusive AI era or slide into a “Next Great Divergence” where overall progress happens but capabilities concentrate among those already ahead, both across and within countries. The idea of the Next Great Divergence harkens back to the first Great Divergence as Western Europe and North America pulled ahead of most other countries from the eighteenth century thanks to the Industrial Revolution and its technological advances. This was followed by a Great Convergence during Asia’s economic renaissance in the second half of the 20th century.

Countries like Singapore, Japan, Korea and China stand at the global frontier of AI research and adoption and are in a better position to capture the benefits of AI, based on the IMF’s AI preparedness index. They benefit from robust digital infrastructure, advanced STEM education systems, and dynamic technology ecosystems that attract investment and talent. In sharp contrast, Afghanistan, Papua New Guinea, Maldives and Myanmar, risk missing early AI dividends as they still struggle with skills, supply of investable resources, reliable power and connectivity.

Within countries, gains are landing first in urban hubs, large firms, and among highly skilled workers. Rural communities, small businesses, and people in routine roles face more disruption. Consider Malaysia: Kuala Lumpur’s income levels support rapid adoption, while those of the rural state of Kelantan are closer to the Philippines, illustrating sharp internal gaps.

Jobs will both be created and reshaped – new roles in data work, model safety, and data-center maintenance even as some clerical and content tasks shrink. The gig economy shows both sides: a freelancer in Jakarta can reach global clients, yet AI tools that let one worker do the job of many can compress rates and raise insecurity. Weak rules, scarce technical skills, and uneven capacity make it hard to manage algorithmic bias, cyber threats, and over-reliance on automated scores.

Minimising the risk of the Next Great Divergence

AI is becoming the region’s next essential infrastructure, like power, roads, and schools, with faster upsides and sharper risks. Uneven starting points mean cities with fiber, compute, and skills can surge ahead while remote districts, priced out of connectivity and missing from datasets, risk being invisible. If adoption follows these fault lines, the “Next Great Divergence” becomes a reality.

Leaders need to move quickly on two tracks: hard infrastructure (reliable connectivity, affordable devices, secure data and compute) and soft capacity (skills, capable institutions, clear rules, open competition, and genuine participation). That means investing in digital access, updating education and training for future jobs, extending social protection through the transition, and deepening regional and international cooperation to share practice, data, and technology while aligning on ethical standards emphasizing trust, interoperability, and closing regulatory fragmentation.