Girl Effect’s Springster team has showcased a pioneering chatbot that answers girls’ questions about sexual health, to experts at the Global Digital Health Forum in Washington DC. The forum is the leading networking event for global public and digital health experts from more than 100 countries.
Here Lorya Roblin, Acting Director of Springster, explains the thinking behind the development of the chatbot and what the team has learnt so far.
In many countries in which Girl Effect works, girls aren’t allowed to discuss sexual health and contraception, are shamed for their curiosity, or given answers that are unhelpful and incorrect.
While moderating Springster, Girl Effect’s global mobile platform, we noticed girls wanted specific advice on taboo topics around sexual health content, and a private and non-judgemental space in which they could safely get information.
As Springster reaches 2m girls annually it would be impossible for moderators to respond to each and every one, so we turned to chat app technology creating a chatbot called Big Sis which currently operates through Facebook Messenger.
We ran workshops with girls to co-create the chatbot and make sure we were prioritising girls’ safety. Equipped with a solid plan for the role it could play in girls’ lives we embarked on an iterative test & learn approach, during which we built, tested and optimised the chatbot.
Girls using Big Sis are guided through entertaining, informative content with quizzes to test their knowledge on sex, sexuality, contraception and relationships. Privacy, trigger words for disclosure, and the risks girls potentially face even for looking at sexual health content, were factored in at development stage.
At the moment Big Sis is a prototype but we are already seeing results. 85,000 girls clicked through from Facebook ads showing girls’ eagerness to chat.
Of those who used it, 54% said they couldn’t get the same information anywhere else – this rose to 70% for some topics – and 82% said they would share it with friends. On average, girls improved understanding of topics by 36% after reading the content.
We have completed our first phase of development and are now exploring rolling out the chatbot to other platforms, such as WhatsApp. We are also exploring building up AI capacities with Machine Learning to enable girls to ask questions and be directed to content that most closely answers their questions.
Through this initial phase we’ve established that a chatbot can work in multiple diverse countries. Our next step is to focus in on one country before launching a new and improved version of Big Sis at scale, so that we can equip girls with the knowledge they need to stay healthy, get an education, earn their living and make better decisions about their own lives.
If you want to know more about how Girl Effect built the chatbot, download the paper below: