Kerk Kee and his associates at the University of Missouri are developing a new chatbot to help neuroscience researchers and educators with the hopes of making it available to the masses.
Chatbots have been around since the 1960s, but have become more prevalent with the advent of the internet. Looking for something specific on a website? Simply click the "Let's Chat" button in the bottom right-hand corner of your screen and an artificial intelligence (AI) chatbot can help. However, there are limitations to what they can do and how much they can help.
Texas Tech University's Kerk Kee, an associate professor of professional communication in the College of Media & Communication, received a grant from the National Science Foundation's (NSF) Office of Advanced Cyberinfrastructure to design a chatbot to provide user support for neuroscience researchers, educators and students who use the CyNeuro portal.
"Big data requires technical skills – such as programming, using cyberinfrastructure (CI), high-performance computing, etc. – to fully harness," Kee said. "Otherwise, hidden insights remain buried in big data, which does no one any good. The barrier to entry is high, including in science, technology, engineering and math (STEM) fields. So, a group of people started the science gateways (SG) movement to create a point-and-click, user-friendly interface approach to big data, to reduce technical requirements before users can access and harness big data.
"However, even with the creation of SG, challenges remain, especially due to a lack of user support of SG and CI in general. The high learning curve and lack of user support deter user adoption of SG and CI. When users adopt them, they struggle with meaningful implementation, and some discontinue using. The big-picture problem then becomes, STEM researchers are not able to maximize the potential of big data for groundbreaking research 'to promote the progress of science; to advance the national health, prosperity and welfare; and to secure the national defense,' which is NSF's mission."
Kee is working with Prasad Calyam, an associate professor of electrical engineering and computer science, and Satish Nair, a professor of electrical engineering and computer science, from the University of Missouri.
Kee is the social scientist helping with the social/human side of the project. Calyam is the computer scientist providing expertise in designing chatbot/conversation agents with AI. Nair is the computer scientist who created CyNeuro, an SG for neuroscience research and big data. CyNeuro is the platform where the chatbot will be tested for the project. The group named its chatbot Vidura for a wise adviser in Indian mythology.
Kee will conduct three studies over the course of the three-year project. First, he will design an approach to assess a CyNeuro user's competency in cyberinfrastructure proficiency and neuroscience proficiency.
"Crossing these two dimensions, we will create a two-by-two matrix with four quadrants (high-high, high-low, low-high and low-low)," Kee said. "Every CyNeuro user initially will be placed in one of the four quadrants, and the chatbot will provide user support based on where they are on the matrix. So, if two users ask the same question to the chatbot, the chatbot will provide user-centered and proficiency-matching responses, depending on where the users are located on the matrix. For instance, even when Vidura receives the same question, if the users are from different quadrants, let's say a user who is low in cyberinfrastructure proficiency and high in neuroscience proficiency and another user is low in cyberinfrastructure proficiency and low in neuroscience proficiency, the chatbot will respond differently."
Also, as CyNeuro users advance and learn over time, their placement on the matrix can shift, allowing the chatbot's responses to remain relevant to them.
In the second study, Kee will conduct a validity test by asking users to complete a few tasks on CyNeuro. The tasks will vary between CI and neuroscience proficiencies where the tasks assess the users on cyberinfrastructure proficiency and neuroscience proficiency.
"We hope the users who completed the assessment measure in Study 1 and were placed in one of the four quadrants can successfully complete the tasks matching the appropriate level of proficiency on each of the dimensions during Study 2," Kee said. "If so, we know our assessment measurement from Study 1 works. If not, we will go back to tweak the assessment in Study 1, because we really want to be able to put users in the appropriate quadrants so the chatbot can provide customized, user-centered support."
The third study will test the chatbot's usability. The group will study how the chatbot can best interact with the users, provide answers to users' questions, ask users follow-up questions, etc. They also will pay attention to the overall design of the chatbot and observe and interview the users during Study 3 to get design feedback.
"I study the adoption, implementation and the ultimate diffusion of new technologies, new behaviors and innovations in social systems," Kee said. "In this project, the innovation is SG/CI, and through an interdisciplinary and sociotechnical team, we are designing a user-centered chatbot to support the individual adoption and systemic spread of SG/CI technologies by users in STEM. This project also will contribute to an emerging field called human-machine communication in the greater discipline of communication."
Starting with CyNeuro in neuroscience research with big data, Vidura eventually may spread across different SGs and different STEM fields.
"Even if Vidura ends up only helping a small group of SG/CI users in STEM, we hope it makes a difference in their world in some way," Kee said. "In the end, we believe Vidura can help STEM researchers conduct research with big data that can help promote NSF's mission."