The Intel

AI doesn’t just enhance human productivity—it replaces certain types of labour, and it’s starting at the bottom. In the U.S., entry-level hiring in tech has fallen by 50 percent, as employers quietly hand foundational work—operations, communications, and analysis—to machines. By absorbing what used to be “learning-curve” work, AI is silently raising the bar for where careers begin. And in many of the places we work, those on-ramps were already fragile. If we’re not careful, we’ll wind up encouraging young people to upskill for roles that no longer exist or are harder to attain.

In fragile contexts, the risks aren’t theoretical and go well beyond the labour market. In Papua New Guinea, an AI-generated deepfake showing the Bougainville President punching the PNG Prime Minister went viral—garnering nearly 400,000 views—before fact‑checkers confirmed the video was fabricated.  

But it gives as much as it takes. In Nigeria, just six weeks of ChatGPT‑supported tutoring delivered staggering gains in English skills—equivalent to as much as two years of conventional schooling. Elsewhere, chatbots have been used to cut conspiracy-believers’ convictions by about a fifth. AI is also showing incredible diagnostic power. Research indicates that certain tools are outperforming doctors when run solo, but offering little change when paired with them. This suggests the problem isn’t AI’s intelligence, it’s how we partner with it.

With Big Aid receding and AI-driven disruption on the advance, both the ceiling of what’s possible and the floor of what’s secure are eroding. If we don’t find a way to adapt, we may discover that the middle space in which we operate could be threatened. This is the critical space for localisation, political savviness, mentoring, and adaptive delivery.  

The question is whether we will shape that change or be shaped by it. Can we direct it toward inclusion, equity, and smarter delivery? Can we soften the blow for those it leaves behind? These aren’t side conversations—they’ll define our work for decades to come.

Artificial Intelligence has stimulated very different strategic and policy responses around the world. Some governments, usually with more established economies, are taking a narrow productivity and competitiveness approach. From this perspective, AI is a way to accelerate existing economies and industries, but will fundamentally maintain status quo, scarcity-based economic paradigms.

Many other countries, especially throughout the Asia-Pacific, are choosing a different path: to embrace the disruption of AI as a launchpad to explore and implement an "AI economy". This very different viewpoint is driven by a surplus mindset, and the search for positive opportunities to reshape society as a whole.  

It is inspiring to see Malaysia, Singapore and many countries of the Global South leading the way with AI. Many of these countries are seeking to actively transform entire sectors, including the public sector, and are looking to improve social protection and equitable economic outcomes for society as a whole with AI.

Big questions loom. How meaningful will traditional industrial economic measures be in a world where value generation is fast and widely accessible, where hours are not tethered to effort, where individuals can orchestrate an entire workforce of AI agents to design, deliver and optimise an entire industry?

AI could also be an opportunity to reshape societal norms to address social vulnerabilities and fragmentation. Should we be using this moment to reshape modern economies to work less hours, to allow people to invest time into their families and communities, and to bolster civic participation? Should we properly incorporate quality of life measures into how we measure a healthy economy?  

Emergent AI systems might well force us to confront these questions.

The use of AI is accelerating, but not for everyone. I'm especially interested in two groups where uptake is slow. First: the international development sector, including people and organisations with ample resources to engage. And second: those without the same resources and access, who stand to be most severely affected.

Let’s start with the first group. I think it’s fair to describe the disposition towards AI of many in the development ecosystem as classically avoidant behaviour. This is understandable. AI feels deeply antithetical to development values: humanity, participation, equity. Add to that minimal demand for innovation, including among donors like DFAT who are legitimately twitchy about cybersecurity and still haunted by innovationXchange-era beanbags, and you’ve got a sector that’s not exactly sprinting into the AI era.

But here’s the problem: much of the current discourse treats AI as a tool for development. We tend to talk about AI as a new technology for education, health, or service delivery. But AI is also a force influencing development trajectories and reshaping economies, governance, power, and trust.

The second group, both countries and individuals who currently lack control of AI resources, face the bulk of the risk. These risks include job losses in sectors like manufacturing, AI-driven misinformation, and centralisation of elite power. These harms are real for the already marginalised, while the potential benefits of AI, such as improved education, remote health diagnostics, and better service delivery, remain concentrated in wealthier communities.  

Without deliberate engagement from the development sector, AI will entrench and deepen existing inequalities. It’s easy to disengage. I didn’t want to think about AI until recently when I discovered ChatGPT could write decent haikus. But if you care about people, you should care about AI.  

We who could shape it
wait too long, unsure, afraid—
but harm won’t wait too.