Sex chatbot android
Soon I had a Java interpreter ready to run on an Android app.Rather than building Snub, I ended up shipping three different chatbot apps: But, by June 2016, two years later, none of these chatbot projects had really taken off.But after reading Karpathy’s post, I decided to dig back into Deep Learning concepts.The first few times through the article, the concept was so baffling I had to walk through building an LSTM from scratch with matrix multiplications in order to get the proper intuition.Take a look at Sophie Bot in action: Irving is the CEO and Co-Founder of Sophie Bot, where he’s focused on the technical and business challenges of growing his Nairobi-based startup.Irving, in his own words, is a “self-taught Android developer and a self-taught AI developer.” While he’s not scouring arxiv for NLP papers, Irving mentors high school innovators on mobile development. I’ve learned a ton from Irving since we first met on Twitter, and I’m thrilled to share my conversation with Irving for this Humans of Machine Learning (#humansofml) interview.However, this was around the time that I realized how big of a problem sexual health is in Kenya and Africa.I saw a natural opportunity to adapt my earlier chatbot experiments with AIML to something that might be helpful in improving access to proper sexual health information.
In my usual fashion, I purchased the domain before a single line of code was written (it was going to be called Snub).
In this post, we’ll dig into the machine learning and deep learning techniques that power Sophie Bot, as well as Irving’s own journey building an AI startup in Africa. We’re a Kenyan AI startup that answers your questions on sexual health.
Let me give you a bit of background about why we’re building this company.
Definitely not my ideal version of the product, but I thought it could be a useful start. Within a month of launching Sophie Bot, I had recruited a team of five, we had won a ,000 in a sexual health innovation challenge by UNFPA, and was already featured on national press, as you can see in this You Tube video: What he had achieved, though, is automating our earlier rule-based strategy using Python and an SQL database.
My plan was to monitor new questions and add them to our knowledge base over time. He made the bot classify questions and statements, but it couldn’t do much more about delivering answers — and that was our biggest problem. It was at the time we developed an NLP pipeline to gather even more cool insights on the data we had collected.