Will artificial intelligence widen inequality?

Artificial intelligence (AI) holds a lot of promise but also carries enormous risks. Aptly titled “The Great Divergence,” the UNDP Regional Human Development Report (RHDR) warns that, without decisive action, many developing countries in the region, owing to their weak capabilities, risk being left behind in the AI race. They will be unable to harness the upside of AI while mitigating the disruptions that have often accompanied frontier technologies. The risks are high as AI emerges amidst growing socioeconomic disparities among and within economies.

Like every major technological revolution—from steam power to electricity—should we expect a repeat of past trends from the emergence of AI, where inequality initially rises before benefits begin to diffuse? But we are entering the age of AI when inequality is already rising, exacerbated by the weakening relationship between GDP growth and job creation. For example, between 1995 and 2020, the top one percent captured 38 percent of global wealth, while the bottom 50 percent accounted for just two percent.

RHDR calls for embedding equity into policymaking. Equity cannot simply mean waiting for AI’s accrued benefits and redistributing them. It begins with AI that serves people and enhances human capabilities—a message that lies at the heart of human development. However, it cannot create equal opportunities for all if existing deficits in human, institutional, and financial capital are not recognised. Disparities in human development could be amplified by AI, a technology potentially as transformative as the steam engine and electricity. It is becoming a critical infrastructure that will determine the nature and pace of future development.

As the UNDP report shows, the Asia-Pacific will be a testing ground to see if AI-led development converges development outcomes or drives countries further apart. Countries in this region lie across a broad development spectrum: very large economies, small island nations, landlocked countries, and high-income economies (greater than $90,000 per capita GDP), alongside least developed countries (less than $500 per capita GDP).

From labour-intensive manufacturing, Bangladesh aspires to move into high-value production of goods and services—towards a more diversified, globally competitive economy, a country with a social safety net that supports not only those temporarily affected by the vagaries of the market but also those who are vulnerable and excluded. Moreover, a country with a public administration that is accountable and capable of delivering good quality services at scale. In all of this, AI could potentially play a critical role.

A forward-looking AI agenda for Bangladesh should focus on aligning it with the needs of people. Without a purpose-driven focus, there is a danger that AI won’t deliver its promise to improve human welfare. A good starting point for public policy is narrowing the digital gap. While connectivity has expanded in Bangladesh over the years, gaps still persist across rural areas, income groups, and gender. Women in Bangladesh are less likely to own a smartphone. Only 38 percent of women use mobile internet compared to 52 percent of men, severely limiting their ability to benefit from AI.

There may not be consensus on whether the net benefits of AI will outweigh its disruptions in the labour market, but many agree that countries must rethink education and training systems. Previous waves—such as industrial automation in the 1970s, ICT diffusion in the 1990s, or the introduction of robotics in manufacturing—initially caused job losses but eventually contributed to productivity booms. We cannot necessarily take past evidence and extrapolate it to the future. However, a key lesson is that education and training systems should be adaptive to rapid technological change, embracing lifelong learning. It serves as an insurance policy against joblessness as the labour market adjusts to technological change.

Furthermore, AI in public services should be underpinned by transparency. AI-assisted services should be clearly explained, open to scrutiny, and corrected when errors or biases occur. For example, Canada provides a set of guidelines for the use of AI and automated decision-making.  Public trust must not be undermined. Without safeguards, biased algorithms can erode trust and cause lasting harm.

Bangladesh is in a region that is likely to drive future global growth. According to recent estimates, the Asia-Pacific region will contribute 60 percent of the global GDP growth. Asia is also emerging as an important hub of AI investment and AI patents. A large country with a young population, Bangladesh has a stake in championing responsible AI for prosperity, while strongly advocating for minimum standards and collective safeguards essential for human-centric AI development. Bangladesh has already conducted an AI Readiness Assessment that provides a solid foundation for developing an AI roadmap for the future.

Owais Parray is country economic adviser at United Nations Development Programme (UNDP), Bangladesh.

Views expressed in this article are the author's own. 

Follow The Daily Star Opinion on Facebook for the latest opinions, commentaries, and analyses by experts and professionals. To contribute your article or letter to The Daily Star Opinion, see our guidelines for submission.