MERL Tech DC 2018

We don't doubt that machine learning can help us achieve MEL outcomes more efficiently, accurately, consistently, etc. But taking the plunge—defining a relevant problem, finding a machine learning tech partner, etc.—can be a huge challenge, especially at a resource-constrained NGO. This session will try to make it easier for others in the MERL community to take that plunge.

For the past 6 months, IREX has been collaborating with Lore, an AI startup, to test how machine learning can help us monitor the impact of our media trainings at a scale, efficiency, and consistency that far outpaces those of traditional human "coders". We've trained a computer to examine thousands of news articles in Mozambique, to automatically measure the quality of journalists’ content. We've turning this experience into an explainer for NGOs and global development organizations that provides practical tips about how to (and how not to) use machine learning technology to tackle their challenges.

Join this casual "Q&A" session, where we'll share those lessons and tips and discuss concrete, actionable steps to commons questions, ranging from "What does a TOR for a machine learning MEL project look like?" to "How much time do you need to spend 'teaching' the machine?"

We've heard enough talk about how frontier technology like machine learning can transform our work as MEL practitioners. At this session, let's unpack that black box and walk the talk!