Posted on April 1, 2018 by Rafal Kocielnik
Reporting activites at work is a common practice, either through stand-up meetings, emails, written reports or other means. Unfortunately, most employees consider it difficult, bordersome and borderline useless. Reporting progress is, however, very important for coordination and planning.
Existing tools support management centered reporting: Slack chatbots become increasingly popular for activity reporting at work, but mainly offer management-centered reporting that does support employees’ self-learning.
Modalities of conversation important for behavior change: Text and voice-based conversational agents stand to play an important role in supporting behavior change, wellbeing and productivity in various domains.
We designed and implemented a custom conversational agent called Robota (which stands for “work” in Polish) to support workplace journaling and reflection. Workers interact with Robota through chat and voice, and can explore past interactions through a web dashboard.
Chat module: We implemented Robota’s chat module as a “Slack bot” via the Slack API. The bot has the ability to send and respond to direct messages on Slack (a Slack bot appears just like a person on Slack, appearing in the user’s contact list).
Voice module: We used the Alexa Dash Wand – a handheld cloud-connected device with a built-in speaker and microphone that allows the user to take it to a quiet room and speak to it discreetly. The Dash Wand supports the Alexa Voice Service (AVS) and custom-built apps (called “Skills”). We implemented a custom skill that asks the user a random reflection question and listens to the user’s response.
To understand the potential role of modality (chat vs. voice) for journaling and reflection with a conversational agent in the context of the workplace, we conducted a three-week in-situ deployment of Robota with 10 participants in our lab, which is part of a larger multinational corporation (English, Japanese, Chinese and French accents). We collected user feedback through survey and interviews.
Robota sent a total of 174 reminders for journaling. Robota also sent 98 requests for reflection followed by 59 reminders (34 in the Chat-Reflection condition and 25 in the Voice-Reflection condition). The average length of a daily activity log was 292 characters (SD=239.62). The average length of a response to reflection questions using chat modality was 131 characters, compared to 98 using voice modality.
Work Reflection with RobotaWe found that reflection with Robota helps with management, organization and performance, it also encourages changing perspective and considering new aspects of work. Finally it promotes considering higher-level goals and the "bigger picture" related to ones activities and professional career.
Chat modality feedback“It was easier to read the question and think about it”, “Reading is much faster than listening.”
“It's easier to answer than explaining in a voice. Since my English is not so good, I couldn't answer to a question immediately if I have to speak.”
“Sometimes the answers to the questions are a bit complex, but I write something that is simpler and reductive because I don't want to spend time detailing it out on slack.”
“I also could more easily change my response with the chatbot before submitting.”
“Typing on slack is slightly more formal I guess, it is something that goes into the record”
Voice modality feedback“It's good to have another means to quickly capture some useful points or thoughts.”
“It doesn't take much time to answer, is easier than writing report on Slack.”
“While using voice, it seemed to encourage me to answer right away, which is a bit stressful”
“It is not transcribed and listening to what I said many times is somehow troublesome.”
“It was kind of neat to use the wand and have the voice reveal to me what the mystery reflection question was.”