We’re often asked to make decisions based off limited information. Take buying a house for example. For a fairly large decision, it doesn’t involve the most crucial step: living in the house. Now imagine being given the opportunity to live in your potential new home for six months before actually purchasing it — like taking it for a test drive. That’s exactly what Senior Business Analyst Len Zhou says the Chatbot Exploration project resembles through its new approach. “How much better would it be if you could live in the house before making a decision?” Zhou says. “You can go to other people’s houses to gather ideas. You might see things you’ve never considered, and I think this is the same case with technology.”
Driven by a curiosity to learn more about how chatbots currently on the market compared to what we have already built with Eve, IST embarked on a project with the Registrar’s Office (RO) and the Faculty of Extension to explore how different chatbots might be useful to solve real university problems. Simon Collier, Manager of Solutions & Standards, has designed the project in a fashion uncommon to how we’ve carried out projects in the past: experimentation. “We have selected two vendors, two parts of the university and two use cases to solve in six months,” Collier says.
The RO is testing Ivy, a chatbot service that specializes in higher education institutions to answer applicant questions during the admission process. With the typical influx of questions becoming more sustained and lasting outside of just the application season, much of the RO’s time is being consumed by answering routine questions that don’t require a human-to-human interaction. Also needed for the purpose of alleviating workloads, conversational agent Yana is assisting the Faculty of Extension with questions surrounding courses and availability.
Currently in the proof of concept stage, the two chatbots will go live next month and operate for six months before being “killed off,” Collier explains. “We’ll be using what we’ve learned from the experiments to put together an RFP, and ultimately that RFP will give us a product or service that we can provide to the rest of the University.” The experimentation process brings a new perspective allowing for trial and error. In fact, it’s an incredibly responsible way to carry out a project, both financially and in terms of soliciting more comprehensive findings. “This knowledge means that when we go to RFP and we spend a bunch of money, we’re doing so more responsibly,” Collier says. “We’ll better understand what we’re asking for.”
As the Technical Lead for this project, Identity and Application Support Analyst Johannes Pauw can't say enough positive things about being given the opportunity to test, and willingness to fail. “The important thing is to find out whether or not the technology is a good fit for our goals as fast as possible,” Pauw says. He explains that long-term, the experiments actually speed up the process of enacting a new service as you can move quicker, without the fear of taking each step with such care. “For us to be able to implement it, we have to understand it and for us to be able to understand it, we have to be willing to experiment and take risks,” he says, “because otherwise, how can we know whether or not it fits with what we're trying to do?”
But why not continue to use and expand on Eve? “When we first built Eve, we built it the way we thought the standard process would be,” Zhou explains, “but when we went out to the market, and explored what was out there, we found that the market has leaped and bound past that.” Companies designing chatbots like Ivy and Yana are single discipline, with all the time and money to focus their efforts on creating the best product or service. That’s not to say that Eve couldn’t get there, Collier explains, but it would have taken IST a significant investment to catch up. “There just isn't really much point in us trying to replicate an already existing commodity that's out there,” Pauw says.
Ivy and Yana will take much of the work off IST’s plate, self-sufficiently building or finding their content, answering questions, and working towards continuously improving. Taking chatbots at UAlberta to the next level is really about what they can do with future integrations, Pauw says, which includes linking into PeopleSoft, Slate and other university services like bookings. Above all else, user experience will drastically increase, with these chatbots being more personalized, immediate and interactive. “The chatbot is able to be where the customers are,” Zhou says. “It's interactive, it's where you are, and it's made for you.”
Pauw unequivocally supports carrying out all future projects through experimenting, saying it's the next step for IST. “As an organization we’re searching for ways in which we can provide value to the university that doesn’t turn us into a commodity.” As an organization it’s our duty to find unique issues that the rest of the University isn’t dealing with and spearhead the mission to cause transformational change, Pauw says. “I think this kind of experimentation style is crucial for being able to do that.” Collier echoes the sentiment, pointing out that it proves especially beneficial for projects “where we don’t know what we don’t know.” Experimenting increases cost and time commitment, but decreases uncertainty and the level of risk.
The chatbots will experience the beginning of two semesters in order for the team to measure and evaluate their value. “If it goes well, and we anticipate that it will, we will use this information to come up with a university-wide chatbot strategy,” Collier explains. “What we learn will guide our strategy and guide our architecture. I think we're coming into it with a better understanding of the impact on the student experience.” Ultimately, the experimentation process sheds light on the project’s overall goals, which Collier summarizes as a blend between the student and employee experience and improving efficiencies for the RO and Extension.
This is still very much the beginning for our new friends Ivy and Yana, but the project team is excited for what’s next. “I'm personally confident that we should definitely see a real impact on the percentage of simple questions that people get answered via the chat bot versus having to go through a human,” Pauw says. Either way, being given the opportunity to experiment first has led to a lot of lessons learned. “You don’t know what you don’t know,” Zhou says. “Sometimes we have ideas as to how something should work, but really what you should be focused more on is what you’re trying to achieve and then let the technology show you what it can do.”