DevLearn Digest 8 Learning to love the robots
Hi everyone,
Happy New Year, and we wish you a happy and successful 2023.
If your New Years’ resolution was to learn more about market systems development or monitoring and evaluation, you’ll be delighted to know that our May 2023 courses are now open for registration. You’ll join over a thousand people who have strengthened their skills and built their careers through our courses, participating in a lively, accessible introduction to the topics. You can find more information on our website, and register before 28th February for an early bird discount.
Now, on with the newsletter…
At a recent conference, I had a discussion with a European NGO about the role of technology in international development. To my surprise, all the other participants were deeply suspicious. They spoke of technology as if it was a slightly mouldy loaf of bread. There might be something useful there, but the emphasis was heavily on the risks, and it might be safer to throw the whole thing away rather than exploring the potential uses.
Now is a bad time to be a technophobe, as I think the next phase of technological disruption, driven by Artificial Intelligence (AI) is within sight. I grew up with science fiction stories of AI plotting to take over the world, so I find it disconcerting that new AI systems – built on artificial neural networks that mimic the brain – are able to write poetry, discuss Shakespeare, give advice, and answer questions. They can even discuss how AI systems might take over the world. If you haven’t already, go to Chat GPT and sign up to find out for yourself.
How will this impact our work? In a narrow sense, improved AI is likely to change the job of an office worker. ChatGPT can already generate funding proposals, reports, and monitoring frameworks faster, more efficiently, and sometimes even better than a human can. I’ve used it myself to write portions of proposals (and to copy-edit this newsletter). ChatGPT can improve and give feedback on written English, opening doors for people who do not speak English as a native language, and reducing the inequity of international organisations staffed with inexperienced English-speaking graduates, whose main selling point is the ability to write well.
More importantly, I believe that AI has the potential to transform the lives of people currently living in poverty. There are many potential uses of AI for agriculture, such as providing farmers with better agricultural advice, providing information for farmers to negotiate a fair price for their crops, and helping smallholder farmers access financial services and credit. Workers will have opportunities to develop training datasets for AI systems, and as AI systems reduce language barriers, they will be better able to take on international remote work if they have the right skills. Fortunately, AI systems can also be used for training. I use ChatGPT to help teach me Excel and programming – it would be relatively simple to repurpose it to offer soft skills, technical skills, entrepreneurship skills, and more.
As my colleagues at the European NGO understood, technological change comes with risks as well as benefits. AI is likely to lead to job losses – possibly of a scale that we have never seen before. It could centralise power in the hands of a few (already powerful) tech companies. People living in the Global South, who often have lower levels of education and computer literacy, may be less well placed to take advantage of new opportunities. Like all revolutions, the timing is unpredictable. Usable AI systems could replace jobs as soon as next year – or it could take decades
But I really think that change is coming, and as development practitioners we need to learn about and prepare for it. Four ways to do so are:
- If you have the opportunity, ensure that someone in your organisation understands the benefits and limitations of AI systems. There are free courses, such as Fast AI, and cheap courses on Udemy which will help you do so.
- Think about the data you are collecting, and how this can be structured in a way that will enable you to take advantage of AI systems in the future. For example, for years I deleted the course forum after each training course. I now regret that oversight – it could have been an invaluable source of data to help train an AI system to give answers to common questions about MSD.
- Think about the ethical implications of AI, considering issues such as bias, transparency, and accountability. This will be particularly important in areas already characterised by high inequality. An AI system that is trained on biased data (for example, a financial institution that provides more loans to men than women) will replicate this bias in their future responses.
- Build partnerships with others working in the area.
That’s all from us this week. Let us know if you have any comments and tips based on this newsletter, what other topics you would like us to cover, and we hope to see many of you on our course soon,
The DevLearn Team