Create a chatbot that helps your day to day

The initial objective should be to provide students with a solid foundation on how machines learn through examples provided by humans. The second objective is to help students demystify the internals of artificial intelligence by delving into building a chatbot. Lastly, it's also important to talk about the bad parts of machine learning, for example ethics, and how that plays a role in who should train the chatbot, what dataset to use and the implications of bias.

This course aims to demonstrate how artificial intelligence is used in chatbots. With a heavy focus on how to train a machine to understand common utterances, intents and slots. The goal of this course is to understand the difference between artificial intelligence and basic decision trees through a variety of exercises such as controlling virtual IoT devices with voice and text, implement Speech-to-Text (STT) and Text-to-Speech (TTS), integrate other deep learning modules and finally, for students to solidify their understanding, bundle it up into building a chatbot.

  • Speech-enabled Smart Assistant Device
  • PC/ laptop with internet connection
  • Common programming languages like Scratch / Python / JavaScript
  • IDEs
Cloud Services
  • Natural Language Processing (NLP) Platform
  • AI vision cloud service platform
  • Platform for building IoT integration and common deep learning modules
Major Activities
  1. Aims to introduce bias and understanding machine learning into 4 different activities.
    1. Build a chatbot to understand text input to control a variety of virtual IoT devices.
    2. Expand on the first exercise by introducing utterances and slots.
    3. Build your own chatbot that understands speech and talks using Text-to-speech.
    4. Build a simple study buddy that helps test the student on a specific topic.
  2. Showcase (final session)
  3. Students are tasked to create a chatbot for the school's webpage for handling public enquiries, or students can create a chatbot for the school open day for answering enquiries.

Learning Objective(s)
  • Understand the benefits and limitations of a chatbot powered with artificial intelligence
  • Applications of artificial intelligence
  • Where do we stand with artificial intelligence today
  • Problems that artificial intelligence can solve now
  • Understand how computers can be trained to recognise the intent behind sentences
  • Confidence thresholds indicate when the machine cannot recognise meaning
  • Understand the ethics on why who trains a specific model is important and the ethical implications from training set bias
  • Understand basic accessibility features that enable persons with disabilities to access the chatbot (e.g. keyboard accessible, proper labels, sufficient colour contrast, etc.)
  • How virtual assistants today work
  • Theoretical session: 4-6 hours
  • Practical session: 20-25 hours
  • Medium
Target Level S1 - S6
Target No. of Students About 20-30 students per class