Solely focused on their work. Not communicating with anyone and had headphones on for the majority of the time.

Continuously talked to the other person. Often reviewing concepts and information. Explained or clarified confusing concepts to one another, accompanied by feedbacks to one’s understanding of the concept ( “Good”, “Okay, it makes sense”).

Barely interacted with one another, except for when they ask each other questions. Enjoying each other’s company.

Studying in groups of more than two people, they reviewed the content together and shared study materials. Facilitated discussions in which they built on each other’s ideas.

One key finding from the interview was the absence of voice interaction in the user’s mental model of studying, which we learned from the interviews. Yet, there is an on-going research on conversational learning.

One research showed that the learners used distinct perspectives, interests, and styles to shaped their integrative effort to make sense of the idea while learning through conversations[7]. To add, another paper highlighted how the learning process can be influenced by the use of conversational tools as learning catalysts [9].

These viewpoints were relatable for college students who often need to study and internalize different types of information from multiple subjects. We wished to explore these advantages of conversational learning for the college student’s learning experiences.

The future of voice interaction lies in providing a desirable emotional experience to its users. A study [5] has revealed that the tone and language of a voice can impact the user’s perception and how perceptible the information they are about to listen to. Listening to cues signaling pressure and control leads to preferential and more in-depth processing mechanisms. The incorporation of nudge theory[6] and the usage of positive feedback in the dialog can also be beneficial to help users be motivated to embarking a task.

From here, we wish to utilize the tone and language of the voice assistant to change the student’s perception of studying subjects they are not motivated for. As confirmed by our research, academic studying is often a stressful, but necessary experience. We wish to improve the desirability of the study experience by making it positive and encouraging for students.

Through voice assistants, users can quickly and directly get information. A survey on Alexa[4] usage showed that the top feature users executed most were Alexa to check and search for various facts.

With access to various study materials granted by the students, the voice assistant would save student time’s on finding the materials they need to study. This will be advantageous for students who seek immediate feedback and on-demand access to class materials. It would lead to students’ effective use of time and attention, two crucial factors that distinguished the students’ good studying experience to an unpleasant one.

To add, time management skills have been identified as one indication of higher performance and lower stress and anxiety [7]. Accentuating the voice assistant’s capability on this aspect for the students may help reduce their stress level.

For the students, searching for the right study materials and organizing them is part of their studying process. However, the accessibility of the studying material influenced the student’s motivation to start internalizing the material.

Students found it very difficult to spend time and effort in studying subjects they lacked interest in. This was common for students’ worst studying experiences - a negative cycle of lacking motivation, not studying for the subject, getting a bad grade, and having to study more for the subject they don’t like due to their bad grade.

Upon ideating the voice assistant, we decided to focus on the Amy Persona out of our three personas. This was due to the fact that Amy had the lowest motivation level out of the three personas and a majority of the data was collected from this user type. Thus, we reframed the studying experience to students who do not study regularly, face challenges staying motivated to study, and study alone.

Interviews and surveys with students showed that they wanted to be more efficient and productive with their time for or while studying. Inspired by this finding, we ideated interactions that would support users allocating specific times to study. This scenario illustrates how the user would schedule a study session at specific time blocks of a day with the voice assistant.

A common challenge for students was combatting low motivation and high stress due to it. This scenario shows how voice assistant can approach a user with low motivation and help the user to start and finish a study session.

One way students combatted low motivation was by studying at environments that they found inspiring and could focus best in. For instance, they sought out environments with some white background noise. This scenario demonstrates how the voice assistant suggests the users change their studying environment. It also highlights the voice assistant’s knowledge of both the user’s preference and the ability to suggest the best locations to study.

Users can allocate specific time periods to set up study sessions with Esbie in hourly intervals. Based on the information on the user’s calendar, Esbie can suggest a time period at times the user is not occupied for the user could schedule a study session.

Esbie can remind the users of their study goals. Esbie records and saves the goal the user verbally stated when asked: “What is your study goal?”. This question is asked to the user when they schedule up a schedule session with Esbie. It can also reiterate the goals when asked to remind the study goal.

Esbie can suggest users places to study based on the user’s previous study location patterns by relating the time of the day and the subject the user studied.

During WOZ test, we found some participants were more likely to give rapid-fire prompts. A user said that “I didn’t know if I just could say “hey Esbie” or more” versus “I thought I could say everything at once”. This was due to a mismatch in our concept model and the user’s mental model of how to provide information to schedule a session. Our interpretation was to receive information components of scheduling - time, subject name, and goal - one at a time. Thus, the interactions were planned as an array of questions. However, some participants gave all the three components of the information at once, and they were not presented with a way to exit the interaction or speed it up.

Our experience often limited the user’s control and freedom to end the interaction. The participants felt like they were only able to end the session when Esbie asked “Is there anything else I can do for you?”, which was only at the end of the conversation. Adding on to that, users were also unable to go back to review their responses or undo their responses when they made a mistake. This gave them little control or freedom, which we wish to take into account in the future iteration.

To combat this, we added a command “Bye Esbie”, which the user can say to end the conversation at any given time. We also added a command “Esbie repeat” for the user to be able to repeat what Esbie has said.