Can chatbots help the 76 million people who can’t read this?

Illiteracy is a major problem plaguing populations across the globe. For 76 million people, being unable to read or write creates an imbalance in life opportunities and denies access to the services we take for granted and assume as basic human rights.

On top of this, modern digital technology that empowers us to be productive, healthy and safe becomes elusive. You can’t ask for government services on a website if you don’t understand the letters on the keyboard in front of you. You can’t read a health warning from an NGO if the letters are nothing more than inky symbols.

However, the same chat technology that is becoming ubiquitous in developed countries offers potential in the fight against illiteracy. Just as my 6 year-old son turns to our Amazon Echo to understand what the weather will be today and when the Golden State Warriors next play, so a farmer in rural India that never had the opportunity to complete her schooling could converse with a chatbot to understand why her crops are dying.

Or at least that’s the theory.

Students at UC Berkeley are looking to test this in practice, through a Smarter Villages project currently underway with the Andhra Pradesh Government in India. Teams from Berkeley’s Machine Learning Club and DiversaTech (both student-run bodies) are working with IBM to explore how chatbots can deliver useful services to those previously disenfranchised.

The seeds for this project started last summer when Professor Solomon Darwin from Berkeley’s Garwood Center for Corporate Innovation took myself and the leaders of the two Berkeley student organizations into his office and explained a particular issue that crops up in rural parts of his native India. A low caste member of society may be abused by someone at the top of the chain, like a landlord. Reporting the situation at a local police station proves futile if you are unable to complete the requisite forms. The professor then posed a question: could chat technology and cloud computing help address this issue?

From there the #chat4good project took wings.

Other potential areas for investigation emerged. India leads the world in infant mortality. In rural areas, health workers with minimal education deliver babies, often times not knowing how to address common issues like cholera and other pandemics. Now India also has almost ubiquitous cell phone coverage (more so than the US). What if simple medical advice and education could be provided in conversational format on mobile devices?

Over the course of the last semester the students created a prototype of a chat-based app that can be used to address these (and other) issues. I’ll let the students share their own work in a later post… they are currently completing their Finals and wrapping up the year.

One surprise for me this semester was seeing how quickly the students went from concept to prototype. They used IBM Watson developer services in the cloud to build the conversation flow and coded an environment around this that could allow this chat to take place on a smartphone. A combination of speech-to-text, natural language processing and translation services bring the app to life.

The summer gives us the opportunity to plan ahead on how we can conduct more research on the ground in India and establish a pilot by the end of 2017.

You can follow along on the project at #chat4good and if you wish to get involved, either reach out to me or contact one of the student groups.