Support Automation: Creating a Level 1 Support Bot; An Exploration into the Benefits and Risks of Emotion Recognition

In my technical project, my team and I created a level 1 support bot that can be installed into Slack channels to automate service support. To create a bot that was useful, we employed different machine learning techniques such as SVM, which allowed the bot to learn from user responses. These same techniques are applied to emotion recognition, my topic for the STS research paper. This helped me have context during my research, allowing me to understand both the technical side and the societal implications of the technology.
This past summer, my team and I created a level 1 support bot that answers user questions as they post them, using information provided by previous developer and community answers. My team supports the CIO pipelines, which are used whenever a developer pushes code to ensure that the code is up to company regulations and that no secrets are in the code. There is a company-wide movement to have all applications employ these pipelines, however, setting them up can be a confusing process, leading to the users posting in the CIO support channels. Previously, there would be a developer from my team who was on-call to answer these questions, however, questions can be repetitive, and developers could be spending that time on higher-priority tasks. The project was to create a bot that could be attached to the channel which listens for posted questions and answers them. The bot is trained on previous channel history, ensuring that the responses are accurate. Installing this bot cut down on developer support time by 50% immediately and the eventual goal is to fully automate the support, optimizing developer time.
In my STS paper I examined the emotion recognition, a fast-growing technical field that deals with the part of us that makes us human; how we feel. There are real benefits that can come from the technology, with my paper focusing on the education, medical, and virtual communication fields. In the education field, instructors can utilize emotion recognition to understand how the students are feeling about the material to adjust each student’s material, ensuring comprehensive understanding. In the medical field the technology has had a significant impact in the mental health space, where diagnoses are closely tied with the patient’s emotional state. By being able to accurately understand and analyze emotions, practitioners can provide precise diagnosis for patients. Lastly, in the virtual communication space, emotion recognition can be used to gauge the audience’s emotions, allowing the presenter to dynamically adjust the presentation to provide a more engaging experience. While there are benefits to emotion recognition, the sensitive nature of emotions requires us to handle the development of the field cautiously. Currently, there are some laws surrounding data privacy, however, there are none explicitly surrounding emotion recognition. While this is a good start, there will have to be more laws set as the field progresses to ensure ethical boundaries. People’s emotions can be used to manipulate them so having strict laws around how the data is collected and what it can be used for are of the utmost importance.
I have learned a lot from my technical project and my STS research. From my technical project I was able to understand how machine learning techniques work and apply it. This allowed me to have context when I started writing my STS paper and understand the technical parts of the papers I was reading.

School of Engineering and Applied Science

Bachelor of Science in Computer Science

Technical Advisor: Briana Morrison

STS Advisor: Richard Jacques