‘Bazbot’ was one of 11 chatbots developed by students over two days at Melbourne Polytechnic.
Marc Blanks, executive director of both tech schools, says that the chatbots were created in response to a genuine need.
“We found that we had a problem in developing our tech schools’ websites, where we had lots of different people from different backgrounds asking questions and needing different answers, and it would have ended up creating mountains and mountains of FAQ pages, which people wouldn’t have looked at.
“So we have an industry partner – Microsoft – and we were chatting with them about our problem and Microsoft suggested that potentially a chatbot could be a solution,” Blanks says.
“It was a great opportunity to get students involved with developing the tech school itself by making the chatbot, so that it would be an authentic task for them where what they did would have a real purpose and a real life,” he says.
“Quite a bit of [the design process] involved actually identifying the sorts of questions that people might ask, so we did some crowd sourcing activities, where the students grabbed as many questions as they possibly could ... and combined them all together and we had this huge library of questions.
“So after that we had to come up with answers for those questions and basically program them into the chatbot.”
The next step was training the chatbot.
“Training involves looking at different ways that people might ask the question, or different spelling mistakes they might make, or different language conventions, where they might say ‘g’day’ instead of ‘hello’, so the chatbot would understand and give the most appropriate answer,” Blanks says.
Students also gave their chatbots a personality, with ultra-ocker Bazbot being among the strongest.
“The fantastic thing was that out of all the student groups, every student group was successful and created a chatbot that would have been fit for purpose.”
Blanks says that the event made a strong impression on students.
“As soon as the event finished there was a spike in the data about an hour afterwards, when all the students went home, and they were obviously showing their friends and family, the data went through the roof,” he says.
Click here to see students reflecting on their chatbot learning exercise.