Man in refugee camp sitting outside looking at his phone

And every new innovation offers the same seductive bargain: more development, less politics (Rohingya Creative Production/Unsplash)

Artificial development

From Facebook to artificial intelligence, the development sector keeps mistaking technical fixes for political problems.

Simon Finley

16 September 20264 min read

Facebook was going to democratise the world.

At least, that was the promise.

Somewhere between connecting humanity and making the world more open, Silicon Valley convinced itself that giving everyone a voice would inevitably make societies more democratic.

Development agencies were equally enthusiastic. Digital technologies would leapfrog weak institutions, amplify citizen participation and make governments more accountable.

Then came Myanmar.

For millions of people, Facebook wasn’t simply a social media platform. It was the internet. (Opens in new window) Facebook effectively became the gateway to news, political debate and public information. Democracy appeared possible. Elections were held. The country was opening after decades of military rule.

But instead of strengthening democracy, Facebook entered a country with weak institutions, little media literacy, deep ethnic divisions, limited independent journalism and a military accustomed to operating with impunity.

The technology didn’t transform the politics. The politics transformed the technology. Rather than becoming a platform for democratic participation, Facebook became an accelerant for hate speech, misinformation and ethnic violence. (Opens in new window) It exposed a lesson development had spent decades trying to avoid.

Technology doesn’t replace institutions. It amplifies their absence. Facebook wasn’t democratising the world. It was expanding a business (Opens in new window).

And for development agencies, this resulted in confusing commercial expansion with democratic transformation.

A 2019 photo of Hakimpara camp, Bangladesh, where Rohingya fled and still remain in the years after massacres in neighbouring Myanmar (UN Women Asia and Pacific)

promising a way to skip the awkward bits.

First, it was computers. Then mobile phones. Then social media, open data, digital identity, blockchain, big data.

Each was going to change everything. Some changed quite a lot.

Digital cash transfers (Opens in new window) transformed social protection, reaching more than a billion people during Covid-19 alone, allowing governments to put assistance directly into people’s hands. Satellite data (Opens in new window) made it possible to track rainfall, vegetation and drought across places where ground monitoring was limited, giving governments and aid agencies earlier warning of emerging food crises.

These are not small achievements. Technology can be extraordinarily effective when the problem is genuinely technical.

But none changed the fact that governments are run by people, institutions are shaped by power, and politics remains stubbornly resistant to being disrupted.

There is something deeply seductive about turning a political problem into a technical one. Technical problems have answers. They can be measured, funded, piloted, scaled and placed neatly into a results framework.

Politics is considerably less cooperative. It involves interests, institutions, history, patronage, identity and people who may benefit enormously from precisely the problem you are trying to solve.

There is something deeply seductive about turning a political problem into a technical one.

This is the less glamorous end of development. Reforming bureaucracies. Changing incentives. Strengthening accountability. Negotiating political settlements. Making institutions actually work. Occasionally persuading powerful people that reform might involve surrendering some of their power.

It is slow, difficult and deeply unsexy work. Which helps explain the enduring appeal of technology.

If technology can solve the problem, perhaps the hard work can be avoided.

And every new innovation offers the same seductive bargain: more development, less politics.

Now it is artificial intelligence’s turn. And this time, the promise is considerably bigger. So is the ambiguity.

Artificial intelligence is a technology, a collection of technologies, a general-purpose technology, an industry, a revolution, an existential threat and, depending on who is selling it, the solution to almost everything.

That ambiguity allows AI to carry both extraordinary fear and extraordinary promise. Everyone gets to imagine the version they expect.

Ask 10 people what artificial intelligence means and you are likely to get 11 answers. Whatever definition people choose, its impact on development is already visible (Opens in new window). It can translate languages, diagnose disease, support teachers, analyse crops, detect disasters and process quantities of information that would overwhelm any government department.

For countries with limited resources and enormous needs, that is an extraordinary proposition.

But there are limits to what intelligence, artificial or otherwise, can solve.

An algorithm can identify which children are most likely to leave school (Opens in new window). It cannot make a government care whether they do.

It can tell governments where poverty is concentrated (Opens in new window), where violence is likely to erupt (Opens in new window) or which communities are being left behind. It cannot decide whose interests matter.

That isn’t a computing problem but a political choice.

Put AI in capable, accountable institutions (Opens in new window) and it can make them considerably better at what they do. Put it in institutions that are corrupt, exclusionary or unaccountable and it may make them considerably better at that too.

The experience with Facebook in Myanmar offered a warning.

Technology doesn’t arrive in a political vacuum. It inherits the institutions, interests and divisions already there.

Artificial intelligence will be no different.

About the author

Simon Finley

Simon Finley is an Australia-based strategic adviser with more than two decades of on-the-ground experience across more than 30 countries in Asia and the Pacific.

Topics

Aid & development Artificial intelligence Myanmar Cyber & technology