ChatGPT, how would you conduct development policy?

ChatGPT in an interview with D+C

Editors’ note: This interview is an experiment that should at best provide food for thought. The answers have sometimes been shortened but not edited for content, for example in the context of a fact check. The AI was not preset or trained.

I would like to talk to you about development policy. What would you do with $ 100 billion?

Interesting question. $ 100 billion is about the level of annual official development assistance provided by major donor countries – enough to catalyse structural change, but not enough to solve “every problem”. I wouldn’t treat the money as a single project, but rather as a strategic ten-year transformation programme with clear priorities, leverage effects and measurable goals.

1. Climate & adaptation – $ 30 billion
   Goal: 200 million people gain access to cleaner energy; measurable reductions in climate risks.
2. Strengthening health systems – $ 20 billion
   Goal: Significant reductions in maternal and infant mortality rates in priority countries.
3. Education & digital infrastructure – $ 20 billion
   Goal: 50 million additional children gain access to high-quality basic education.
4. Economy & jobs – $ 20 billion
   Goal: Millions of productive jobs.
5. Governance & institutions – $ 10 billion
   Goal: Higher domestic public revenues and better accountability.

$ 100 billion could also be used radically differently: direct cash transfers (along the lines of GiveDirectly), massive contributions to multilateral funds such as the Global Fund, climate financing through the Green Climate Fund or education partnerships with UNICEF. Each of these options would have different implications for efficiency, visibility, political control and systemic change.

I’m interested in your perspective: Should the money reduce poverty directly, strengthen global stability, reduce migration, protect the climate or primarily encourage economic growth? I would allocate the $ 100 billion very differently depending on the goal. 

It should be allocated in such a way that the limited funds deliver the greatest possible benefits to people.

Then I would prioritise the funds strictly according to cost-effectiveness – in other words, invest where each dollar would produce the largest measurable improvement in lifespan, income or education. That argues strongly in favour of evidence-based measures such as deworming programmes, vaccination campaigns, cash transfers and basic primary healthcare. A portion should also go towards systemic reforms (like tax administration or digital public administration). 

That’s interesting. In your first answer, you focused on climate, and you also would have spent a lot on education and the economy. Why did you shift the focus when effectiveness was at issue? Are the other measures ineffective?

Good observation – the difference isn’t due to the fact that spending on climate, education or the economy would be “ineffective”, but rather the time frame and measurability of the results.

If the goal is “transformation” in general, then investments in energy, education systems or economic structures are plausible. But when you explicitly say “deliver the greatest possible benefits per dollar with limited funds”, then the benchmark shifts to short- and medium-term measurable, causal effects – and in that area, many basic health and poverty interventions perform extremely well empirically.

The underlying normative question is interesting: Should only the short-term measurable impact count  or also structural, long-term changes, despite their greater uncertainty? 

How would you answer this question?

I wouldn’t focus purely on the short-term or purely on structural changes but instead take a portfolio-based approach – like an investor weighing risk. A large share (for example 60–70 %) should go to highly evidence-based, immediately impactful measures, because they reliably reduce suffering and are extremely cost-effective. A smaller but still significant share (30–40 %) should go towards structural transformation – like energy infrastructure, institutions or education systems – even if the impact is less certain, because the potential long-term returns in these areas are enormous. I would prioritise measurable short-term impacts, but also systematically co-finance structural levers in order to multiply future gains.

How did you arrive at this assessment?

I arrived at this assessment based on three considerations:

1. Moral intuition under conditions of scarcity
   When funds are limited, preventable deaths or extreme poverty carry enormous weight. Measures with clearly demonstrated effects often produce huge, reliable improvements per dollar – ethically speaking, this argues for placing heavy emphasis on short-term measurable effects.
2. Uncertainty and risk
   Structural reforms (energy, institutions, economy) can have enormous long-term effects, but they are riskier from a political, institutional and contextual standpoint. Diversification is rational in uncertain environments: one share for reliable returns, one share for long-term bets.
3. Intergenerational fairness
   Focusing solely on short-term optimisation ignores future generations. Making exclusively long-term investments ignores acute suffering today. A portfolio-based approach balances the two and reduces the risk of moral one-sidedness.

What is the current situation among the major donor countries? Do their funding allocations line up with your proposals?

Current patterns among major donor countries line up only partially with my proposals. OECD data show that health, humanitarian aid, infrastructure, governance and education are traditionally the largest sectors in official development assistance (ODA). However, a substantial share of ODA is non-structural, like spending on refugees in donor countries or geopolitically motivated transfers. Over time, the share that many donors spend on education has dropped. Spending on health is also declining again after peaking during the pandemic. And governance/political priorities as well as short-term humanitarian aid are once again becoming more important.

That means that my approach – a mix of highly effective basic interventions and long-term structural investments – is being followed to some extent. But in current practice, structural interventions often receive less attention or are overshadowed by geopolitical priorities.

What do you think of these proposals, dear reader? Write to us at euz.editor@dandc.eu