EXPLORATION PROJET - CHATBOT
At 20, I left the house for the first time to do an exchange program on the other side of the world. I realized how difficult it is to plan it. For my graduation project, I partnered with 2 other designers to conceive Nanook, a chatbot who acts as a coach for French students who want to make international mobility. I was the project lead and I mostly worked on user research and identity.
Even if studying abroad is becoming more and more frequent, around 50% of students interested to do an international experience don't do it. And for those who make the step, it is a real struggle to plan it.
Study abroad expanding rapidly, there are a lot of pre-existing research studies about it. I started to gather this information, mostly from sources such as Campus France or Unesco. Then, I created a survey to understand the motivations, struggles and demography of people who actually studied abroad and those who decided or had to stay. From it, I analyzed the major obstacles to do an international experience.
The level of language is a big obstacle to expatriation. Students only agree to a modest 6.8 / 10 to assess their level of English. Most of students don’t feel comfortable with the idea of having to take an English course in their discipline when 1 out of 6 thinks they are totally incapable.
of French students are not comfortable with English course in their discipline
With an average of 6 075 euros for 6 months stay, studying abroad is very expensive. 32% of students who decided not to study abroad are convinced that they can not finance the stay. Indeed, the aid public authorities cover less than half of the living expenses of beneficiaries. But on another hand, there is a lack of information as to the availability of these aids.
of students with scholarship leave without a mobility grant
Life abroad is sometimes very different from what one knows at home. Getting used to the culture and standards of the host country is not easy, especially in countries with a huge culture gap. The habits, rules, perceptions, humour, tolerance, references, etc. specific to another culture ask for new learning and adaptation. 42% of students would have like to have language classes on the spot while 20% admit having had trouble integrating the country's habits.
of students miss their culture and country during their stay
I was struck by the number and diversity of stakeholders involved to study abroad. Most of the time, it takes more than 6 months to prepare for it. Information are scattered on many levels, is not always up-to-date or doesn't match the situation. Moreover, it is ofter complicated administrative procedures that students face for the first time.
of students who did an international experience would have liked to be guided
Inside our target core, we have two different targets:
- Students who are already counting abroad
- Students that we can encourage to leave
The only thing differentiating these two targets is a greater need for reinsurance at the second. We will therefore treat the two targets as one.
The research phase made it easier to define a scope for the project.
Nanook is a friendly chatbot that has all the answers. The aim is to offer French students a support to guide them in all stages of their international stay. We chose to create a chatbot to fit with usage habits and values of our target.
Here is an example of the use of Nanook by a young person who was looking for housing in Montreal. She asks for a place in the city center, cheaper than 700 CAD and that is available from February. Nanook allows having answers both fast and personalized thanks to:
Send and receive money
Make a budget
Find a place
Go out suggestions
Before developing anything, we decided to create a MVP with a chatbot creator. Obviously, it was impossible to include every topic of expatriation. The user journey helped us to decide to focus on three subjects: visa, housing and banking. We also focussed on 3 countries: Canada, Australia and England.
We created scenarios with clicking options to find out if users found answers easily. However, the study of these scenarios did not seem to bring us much information. Each time, users arrived at the end of the course without difficulty. But, these scenarios only answer a very specific question. Nanook is supposed to be able to act like a human and answers all the questions that a user would have.
So we asked twelve students to write down any questions they might have about visa, housing and banking to find out how they would use the chatbot and what was their expectations.
The question of the price of housing comes back a lot as well as its ease to be found. Students want special offers for them with reduced prices and ease of organization.
Students wonder if they really have to open an account. If they need to, they want details about the opening (duration, exchange rate, closing, bank card)
Everything about Visa are quite stressful. Students need to be reassured with all the current informations about administrative details.
Replies to the questions depend on the country and even on the city inside the country. We realised that we will have to with only one city. And even for one city, the work was considerable to meet expectations. Indeed, despite the directive to focus on visas, housing and banks, some issues are out of the box. These topics therefore seem important to users and need to be supported quickly:
The chatbot will have to be able to understand the questions in spite of misspellings and grammars as well as the absence of accents and punctuation. Indeed, students tend to write very quickly and make typos or spelling mistakes. However, we do not want to put these mistakes forward but learn to understand them.
Most users ask the questions in a very formal way. We want Nanook to be like the good friend who can give you advice. To see how to make users feel comfortable talking to Nanook as a friend.
After working on Nanook, it is obvious that there is a need for a guide through international experiences. Schools want to push it onto students but are often overwhelmed to guide them through the process. And students are facing so much information coming from various sources. It is hard to quickly find info that is relevant and up to date.
However, creating artificial intelligence is not possible for now. A deep personalization comes with a human understanding of the person needs and can't be automatized for now. Or at least, not with accessible technology. Moreover, we noticed that a written-chatbot doesn't help to save time. It would probably require an oral exchange to be efficient.
But even if building the project as it is now is not possible, it doesn't mean that it should be thrown away. It would be interesting to start by following manually few students to help them and figure out what could be automatized and what can't. Even if Nanook will not substitute entirely the human help, it could remove all the easy questions to let people focus on the rest.