WorldWideCode is an SMS chatbot to provide programming resources for MENA refugees in their native languages offline. It aims to enable refugees to break into the tech industry and help their local communities with their coding skills. Refugees can message the bot and receive snippets of coding lessons on their phones in their native languages without internet.
According to the UNHCR, 20% of global refugees have no access to the internet. Instructors serving conflict-affected regions report that poor internet connection is the biggest challenge in teaching. Refugee camps do not have stable internet infrastructure, and students can encounter days without internet and electricity at irregular intervals.
We see an opportunity to deliver coding lessons through SMS messages as there is a high penetration rate of mobile phones in the Middle East. In Jordan’s Za’atari refugee camp, the largest Syrian refugee camp in the world, it is reported that over 89% of refugees own a mobile phone.
We find that it is important to translate coding lessons into people’s native languages.
Most coding resources are in English, and coding languages themselves are in English. People can find very few online resources in languages such as Arabic, Kurdish, and Pashto (the language native to Afghanistan.)
MENA refugees have varying levels of English skills, and in our user research, interviewees stated that they prefer bilingual instructors. Bilingual instructors can communicate more content and help students understand the material better.
We use the Twilio API to create an interactive, SMS-based chatbot and Google Translate API to translate programming resources into 100+ languages. In the event that users ask questions about the content, we leverage OpenAI’s GPT-3 API to provide a response in lieu of a real-life tutor.
Taseen Islam, Uche Unanka, Nick Jiang, Wanying
Made at the Impact Fellowship