Big Data World Singapore 2017 - Moving Towards Digitization & Artificial Intelligence

Editor's Notes

• #2: A very good afternoon ladies and gentleman, how is everyone doing? I hope everyone had enjoyed your lunch so please stay with me for the next 20 minutes as I will be walking you through on how to move towards digitization & artificial intelligence. My name is Garrett and I am a data science professional, have been in the data science industry for more than 13 years, delivered data science projects for companies across various verticals which includes biomedical, government, social services and the financial institutions. I am currently driving Renotalk’s digital transformation journey and exploring artificial intelligence initiatives to help them build their business up to the next level.
• #3: There has been a radical shift towards the commerce industry over the past few decades and this trend keeps on evolving from time to time. I remembered my first experience as a shopper for a pair of jeans started off with the advertisement on magazines and billboards. After viewing the advertisement I head down to one of the retailer’s shop to make the purchase. A few years later after the internet boom era, I am able make purchase online without having the hassle of making a trip down to the store and I can buy anything and anytime I want. Whenever I need anything – from the most trivial item like a mouse, wireless keyboards, business suits when I need to present, dinner or lunch whenever I am hungry and too lazy to go out, or groceries if I feel like cooking. Take a look at the shopping cart, payment method, and the laptop Over the time, these platforms seems so much smarter when they even make recommendations about the items I may need, for example, if I check out barbeque food items, the website will recommend me to get the grill mesh and charcoal wood. Sometimes, they will prompt me for my preferred delivery times and delivery address – whether home or office.
• #4: I will try to keep the agenda simple and easy to digest. First, I will introduce what are the key considerations to go digital in terms of the resources that are required for the transformation – which is the People, Technology and Data. Next, it is very important to know if you are using the right AI for the right problem. I like to give an analogy to this with administering the right medicine to the symptom, if I caught a flu, all I need is a flu medicine, not 101 different pills which could make it worse. Towards the end of my talk, I will present the case studies from the Fintech and Insurtech industry and round it up with a short preview of Renotalk’s digital transformation and AI journey. Last but not least, 5 minutes will be allocated for questions and answers.
• #5: The majority of the costs in moving towards digitization and AI is the people that the business invest in – with the assumption that you are going to build this project in house and not considering COTS. Risks are higher for COTS and I will explain why later. Due to the significant amount of costs invested in people, businesses need to choose and hire the right talent. The first person you want to associate with AI and Deep Learning algorithms are the Data Scientists and this is hyped as the sexiest job of the 21st century by Forbes/Harvard Business Review. Why? For 2 main reasons 1) They are rare talents – someone who can talk in maths, stats, and equations language. They are the ‘mechanic’ for all your machines and they derive patterns from your data. You need someone who can translate insights into business context and tell a story that is easy to digest. Meet your Data Analyst who tells you the big picture. They don’t use fanciful mathematics and algorithms but use simple elementary stats to represent your data and insights. Finally, there will be someone who is able to put your AI machines and visualizations into production and integrate them with your business processes.
• #10: Understanding the different domains of Artificial Intelligence is key for a successful Big Data project. Not knowing the domains and using them inappropriately is one of the most common cause for failing.
• #15: With so much information on our portal, how do home owners deal with finding the right ‘answer’ for their question in the quickest time? The response time in answering the user’s questions is of paramount for retention and it will affects your website’s bounce rate and repeated visitors statistics. How do Renotalk leverage on digitization and AI to solve this problem? Will having a customer service officer to answer questions solve this problem? Yes, it will solve this problem with a caveat: that this CSO is able to handle all type of questions that the users ask and the possibility to simultaneously chat with a few home owners. Can we automate this task by training an chatbot to handle the FAQs? The answer is yes, but it comes with a caveat too: it takes time to train the chat bot. The AI application domain is known as the Natural Language Processing. What is NLP? It is the application of computational and linguistics analysis for computers to understand human language and thus capable to interact with humans. Due to the complexity of language semantics and corpora, deep learning algorithms are fundamentally essential for a chatbot to be robust and appear ‘human’. For the sake of time, I wont go into the details of how NLP and Deep Learning is used in the cognitive process of training the chatbots. We are currently developing RenoBot in the pipeline, you can find our chatbot from our website on the bottom right of the screen. To initiate a conversation from your mobile phone, simply just ask a question and it will respond with a reply to your FB messenger. As this is work in progress, please try this out with a pinch of salt.
• #16: If there are no more questions I would like to wrap up the session with a few key takeaways. To make your move towards digitization and AI, you will need to consider first, The business problem and objectives Whether you have the right data Then bring in the people and infrastructure Know your AI domain and build the right solution How to act upon the insights