Artificial Intelligence: Demystifying and Deploying in Business

As an easy guide to refer to later, here are the key takeaways from his workshop.

• Today's approach to AI relies on computational power. Therefore, it is a tool like a hammer - it can do very little and value creation comes from how the humans use them.
• Not all AI involves machine learning or deep learning. In practice, you would want to minimise the use of these if you can avoid it. This is because 1) rule-based algorithm as AI is still very much in use and 2) it should be used to avoid a) the blackbox problem and b) complications in business setting. 
• For businesses to deploy AI, it is important to remember to be narrow-minded because today's AI can only take on a single task. Ideally, the task is:
  • Standardised and routine data processing
  • Labour-intensive (and therefore costly)
  • Time-consuming (due to labour processing)
  • Rate of error (humans are prone to making mistakes)
• Competitive advantage doesn't come from how good the tool used is. It is all about how the tool is used as part of the more holistic business and strategic thinking. Humans-in-the-loop/expert-in-the-loop that integrates humans and AI is the key to creating business value. This turn requires business people to think through their challenges and solutions.
• Lastly, the key to getting AI to deliver its promised/expected value is not the model itself. It is the AI Ops (AI operations) - CI/CD (continuous integration, continuous deployment/delivery) that delivers the model output all the way to user interface is just as, if not far more, important than the model itself.