We build our work on four pillars of responsible AI implementation:
Chatbots
ASAi’s customized AI-powered chatbots help students navigate critical campus services such as advising, billing, and financial aid. Students can text a designated phone number and receive personalized guidance from an AI language model that draws on official university sources. The chatbot can also offer to connect students with traditional human-powered resources.
ASAi’s chatbots connect vulnerable students with information and resources that are traditionally difficult to navigate, improving equity in both access and outcomes. We can also provide high-quality responses across more languages and dialects than typically spoken by staff.
The chats below compare ASAi’s chatbot (left) to two commercially available chatbots for higher ed (center and right). Both comparisons are real chatbots currently in use at four-year American universities. These are real chats, edited only for privacy. Click to enlarge.
As these conversations show, ASAi’s custom chatbots understand student questions more deeply than other chatbots, and they provide richer and more personal answers. Cutting-edge language AI with a focus on student success is what makes our chatbots unique.
Predictive Analytics
Predictive analytics can help institutions anticipate student outcomes to provide proactive support. ASAi’s models help advisors and faculty identify students who may be at risk of a negative outcome so that they can integrate early support into a holistic approach.
Below are some advantages that set our models apart:
- Our models can integrate with our chatbots, learning from conversations to better support each student.
- Our models can be tuned to provide accurate predictions for students of all backgrounds, avoiding harmful bias.
- Our team works closely with institution staff to develop the models while prioritizing transparency and equity.
- We include a range of voices when making key decisions, including student perspectives, and we carefully test our models for algorithmic bias.
Case Study (Predictive Analytics)
ASAi developed a model that predicted next-term retention at a large four-year public university approximately three to five times better than alternatives. The model leveraged 80 variables from student-level administrative data and 33 variables from public data captured over 12 years. Once machine learning algorithms predicted which students were at highest risk of not being retained, our team worked with advisors to provide these students with early advising to help them succeed.
Ethical & Equitable AI
As a core part of each project, ASAi will align solutions with ethics and equity standards that we co-create with you and your stakeholders, and we’ll work together to implement them responsibly to design, test, and fine-tune our AI solutions.
ASA Partner Sung-Woo Cho and ASA Senior AI Analyst Nathan Greenstein authored the chapter “Ethics and Equity in Data Science for Evaluators,” in Artificial Intelligence and Evaluation: Emerging Technologies and Their Implications for Evaluation.
What makes ASAi’s AI solutions unique?
ASA Research is an interdisciplinary team of subject matter experts and technologists. As an SBA-certified Small, Women-Owned Business, our goal is to help you implement the best available tools in a way that aligns with your goals and your values. Our team of established leaders in the field of AI ethics and equity focus not only on building and implementing the best technologies, but on helping you use them to do good.
