My Friend Fizbit: a Chatbot to Teach Emotions
One of my first jobs in tech was at an assistive technology startup focused on helping parents and caregivers of autistic children communicate with their kids through special interests. If a parent wasn't available, we wanted a way to engage the child while we contacted the parent to chat. Thus, Fizbit, our chatbot was born.
I lead the content team in creating thousands of media clips paired with short dialogues to engage kids around topics like identifying emotions. I analyzed user behavior as kids interacted with our first version of Fizbit, and created personas to focus our efforts in the next iteration. Users generally got bored of the pre-made conversations, but had to sit through all 6-8 prompts to get to another topic.
In response to this, I spearheaded the initiative to integrate NLP (natural language processing) into Fizbit for a later version of the chatbot. This gave us the capability to create more complex dialogue trees based on kids' responses, allowing us to follow the topic they wanted to talk about more closely rather than following a rote script.
This change doubled the time users spent with the chatbot, and even inspired a marriage proposal. (Don;'t worry, Fizbit let them know he's too young to get married, but they can be best friends!)