Jack McQuibban is the Advocacy & Networks Coordinator in the Global Policy & Practice team at Restless Development. In this post, he takes a look at the positive impact that Artificial Intelligence and Machine Learning could have on international development.
A few months ago, a comment from my sister over dinner sparked something within me. A primary school teacher, she told me the children in her class were learning to code during I.T. lessons. Kids aged 9 and 10 were learning what it meant to code and create digital resources, whilst I didn’t have any idea how to begin to describe what coding is. I’m going to get left behind I thought.
Ever since, I’ve been reading up on tech issues and listening to relevant podcasts (except for a very belated reliving of Serial). But it’s been the future of Artificial Intelligence (AI) and Machine Learning (ML) in particular that has grabbed my interest the most. I am by no means an expert in this area, rather a curious newbie keen to learn more, but here are my thoughts on the positive impact AI and ML can have on the international development sector.
So what are AI and ML?
Artificial Intelligence (AI) is extremely complex, but simply put it is the aim of “creating computers and machines that can think like humans do”. It’s the science and engineering of making intelligent machines, especially intelligent computer programs. Narrow AI is already all around us, and has been for years, whether it’s Facebook suggesting friends or Siri’s voice recognition.
There are two classifications of AI mostly used: applied and generalised. Applied AI is more common, machines doing tasks better and quicker than humans can – such as trading stocks or manufacturing cars. Generalised AI is more complex, encompassing machines that can in theory do any task. This is where the most excitement is at the moment, specifically on Machine Learning (ML).
ML is a subset of AI and is really what’s driving AI forward at such speed at the moment. ML is the creation of computer programs that can learn and grow themselves, when they are exposed to new data, without the necessary oversight of a human. With the internet revolution and the incredible amount of digital data being generated every day, algorithms are being created to allow machines to analyse this data faster and more precise than any human can. A simple example of this is Google tailoring what you see when you search for something, immediately tailoring results to what it believes best matches your interest by using the data it collects from your previous browsing history.
So how can AI and ML transform the systems, processes and ultimately impact of international development for the better? I don’t think it’s sensible to call for a total transformation towards machine-led development. But here are three ways in which the sector can better utilise the technology already available:
High Quality, High Speed policy making
Firstly, advances in artificial intelligence and the ability of machines to digest incredible amounts of data mean that we are now in a position where the consequences of our future actions are much more predictable. Machines can analyse the data from millions of previous similar actions taken and then precisely predict the impact of a replicated idea.
If we, as development practitioners, utilised this technology at a policy creation level, we would be able to simulate the impact of a whole range of different actions. This includes measuring their effect, whilst analysing and compensating for extremely detailed and local contextual factors, such as weather patterns or market statistics. Machine Learning provides us with the ability to deliver high quality, high speed policies where the outcomes would have a level of certainty never possible before.
We could accurately predict the impact a new HIV drug would have if made available on the market at various levels within national and local contexts, by cross-examining factors such as the mean incomes of the population and the accessibility of health services in the region.
Or, we would be able to state, accurately and quickly, the impact on national gross domestic product and employer’s incomes if Governments provided more skills based employability trainings for young people, as well as identifying the towns and villages that most need it.
With the right access to the data already available, machine learning technology provides us with an incredible opportunity to make effective, rapid and accurate policies to help us better tackle poverty and inequality across the globe.
Predicting the next global gamechanger.
Building on the point above, deep AI that can analyse vast amounts of data to forecast the impact of future policies can also provide us with the answers to development questions too big for humans to comprehend effectively or quickly enough.
Algorithms can be learnt and then developed, with the addition of new data, to accurately predict the big global trends that will change the we way implement development programmes and policies, such as on early disease detection or specific crop yields in a certain year.
Look for example at the emerging field of “Culturonomics” that analyzes huge digital texts and media archives to provide precise predictions of future trends and events. According to Culturonomics, machine-led analysis of a 30 year global news archive has resulted in the ability to forecast the revolutions in Tunisia, Egypt and Libya, as well as the 200km radius for Osama Bin Laden’s hiding place in Northern Pakistan. What if we applied this technology to development? We could empower development practitioners with the resources to predict, with authority and confidence, when and where the next big medical crisis will emerge – meaning, in turn, we would be ready to halt and diffuse future emergencies like Ebola.
Knowledge is power – Getting to know those you are working with and for
Deep machine learning, combined with the vast increase in digital data being generated and collected on individuals, will mean that by 2050 most scientists predict artificial intelligence will surpass human intelligence in the majority of societal tasks. Machines will know us better than we know ourselves.
By tracking our online movements, and increasingly our offline activities, with smart home support devices such as Amazon’s Echo (also known as Alexa) or Microsoft’s Cortana, machines are learning about every aspect of our lives. They will learn what makes us happy or sad, our interests and passions, our everyday routines. What if this technology was utilised by those of us working towards creating a sustainable future for people and planet? To allow us to better know the people we are working for and with?
For us at Restless Development, this technology, if used in the right way, could help us better empower the young men and women our work is aimed at. Today’s 1.8 billion young people globally are brilliantly diverse but, as such, they require and seek different approaches to tackle the different issues that are most important to them. If we had better data on the daily lives of these young people and the communities they live in, including their activities and interests, we would be able to run more targeted programming and campaigns that greatly improve our impact on each specific community we worked in.
The potential impact this would have globally on development is huge. The management of vast sums of data of course comes with a number of moral and logistical issues. But the possibility of organisations and institutions using machine learnt data to specifically target communities, helping them effectively empower individuals to lead change and progress themselves is a tantalising prospect that is now excitingly within our reach.
Artificial Intelligence is moving at an incredible speed. With it comes unprecedented opportunity, but also unprecedented risk as machine learning takes power out of the hands of engineers and designers. There is a growing amount of evidence to suggest humans in the future will be unable to understand the reasoning or process behind the decisions made by computers than have learnt their programming themselves.
With machine learning, we can deliver more impactful and evidence based policies at a speed never seen before, we can predict both the next big crises and future opportunities, and we can get to know the changemakers we are working with much more than we ever have done previously.
If done well and with the right partners, this provides an unprecedented opportunity for millions of people to lead themselves out of poverty, tackle the growing health issues facing many societies and better mitigate the risks of the incoming climate change crises.