How do you envision the role of AI evolving in the development and humanitarian sector in the coming years? What opportunities and challenges do you anticipate happening?
Given how much is changing in this space almost everyday, it’s hard to anticipate what implications AI/ML and particularly Generative AI will have in the development and humanitarian sectors. What is certain is that ‘digital is disrupting development’ and this will continue for many years to come. Like any other general purpose technology, the nature of the disruption and how it affects the least privileged in society depends very much on those who make decisions about how to use it and who has access to it.
We hear much talk about the ‘digital divide’ getting wider and creating even more inequality in society. I have written about this myself and certainly worry about tasks and even roles becoming redundant with potential to dislocate large numbers of people. This is a real risk and it’s critical that we design social safety nets and enable people to be more productive and not less in the face of such massive change.
Being an optimist, I like to think about the potential for AI including Generative AI to be a ‘great equalizer’. We can put incredibly powerful tools and capabilities into the hands of people who are ordinarily left out of certain social and economic ecosystems and include them in ways never before possible.
For example, we can make available ‘virtual teaching assistants’ to underserved students in rural America or rural anywhere in ways that serve world-class content (in virtually any language), design personalized lesson plans, identify ‘gaps in understanding’ and create activities that are best suited to a student’s personal learning style.
We know from research that giving a student 1:1 tutoring has a massive impact on learning gains. Previously, this was only possible for the most privileged students. We can now level the playing field and give nearly every learner access to a world class, 24/7 teaching assistant that can help them master any subject. With advances in voice, image, and instant translation, a virtual tutor could be sitting by your side no matter where you live.
In another example, we can use high resolution satellite imagery collected from space and aided by computer ‘vision’ to spot crop diseases on plants and alert farmers immediately through their mobile phones. Pests are a massive threat to harvests and giving farmers real-time data and suggestions on what pesticides to use and where they are available can dramatically minimize cross loss. Today, more than 30% of global food production is lost due to pests and disease. This could have a huge impact on global food security in a very localized way.
How do you navigate the balance between pursuing innovative approaches in the AI/digital space and ensuring these things work for the betterment of humanity?
Technologies like Large Language Models (LLMs) that power ChatGPT will become commoditized very quickly. Computer processing power (or compute) is growing exponentially and soon almost everyone will have access to a superbrain in their pocket. What will differentiate one model from another is the data that powers it. It’s the rocket fuel that powers the engine that will create winners and losers. So we must be vigilant about who has access to what data, how we protect privacy, mitigate bias, and how we use data to make decisions. Being data-driven is one thing but who’s data is driving whose decisions?
This is not easy to do particularly given how quickly the floor beneath us is shifting. That doesn’t mean we don’t take advantage of super-cars that can take us farther and faster. But it also means we have to develop speed limits quickly, enforce them, put up stop signs, and make sure we have seat belts. We probably need speed traps too but I don’t want to take this metaphor too far! Going too fast without considering consequences would be a mistake.
No general purpose technology is inherently good or evil. We design the systems that include or exclude people. AI is not inclusive by default. In fact, I think if we leave the ‘market’ to its own devices, we will exclude more people than we include. Being deliberate about how AI can be harnessed to create greater opportunities in education, healthcare, water, and access to finance for example must be intentional and carefully designed with inclusion in mind.
We have the power to do remarkable good but we must be deliberate about how we use the tools in our new toolkit . Think of electricity, the advent of the internet and the industrial revolution happening at the same time. It’s breathtaking, exciting but also dangerous.
Revolutions in quantum computing, Gen AI, the internet of things, and biotechnology are happening at once. We’ve never been in this place before. The possibilities are endless but so is the need to be thoughtful, deliberate and inclusive in how we design the systems that govern us. Success in my view is a judicious balancing of promise and peril. Getting this right is incredibly important and where wisdom is more important than raw processing power.