Insights & Teachers' Network by FeedbackFruits

Harnessing the power of AI to help educators design and share innovative, data-validated, and effective pedagogy.

The growth of digital learning technologies over the past years has done much to make education more flexible and widely available than ever before. However, it also poses new challenges for educators. So much learning takes place outside of the classroom, beyond educators’ field of observation, that it can be difficult for educators to monitor and guide their students’ progress. A further challenge for educators lies in finding innovative, effective, and validated ways of incorporating digital technologies in their education.

            We propose to harness the power of AI to aid educators with answering both these challenges. Insight, our educational AI agent, will perform quantitative analyses of students’ efforts in digital learning applications. It complements this with semantic analysis of students’ contributions to give educators insights into individual student’s contributions, suggesting areas of improvement and potential interventions to guide students’ learning. Automating parts of the diagnostic process frees up time for educators to give more in-depth feedback and instruction to students who need it. Insight's analysis can also be used to give students insight into their own progress, which can be highly motivating for students. Furthermore, Insight’s quantitative and qualitative analysis of student contributions can help teachers see which elements of their course are particularly effective and what needs to be adjusted.

           This brings us to the second element of our solution—namely a Teachers’ Network. In this network, educators can share their most successful learning activities, their best practices, and the lessons learned with colleagues. In the future, Edubot will be able to suggest colleagues who faced similar pedagogical challenges in the past, pointing the way to validated, effective solutions.

           Suggesting solutions from colleagues at educators’ own institution has several advantages: Familiarity with colleagues increases the trust educators will have in the proposed solution; it lowers the barrier to adoption of innovative pedagogy; and it creates a more unified experience for students. This way, the wheel doesn’t have to be constantly reinvented, leading to time savings and higher quality pedagogy.

          Using AI to give insight into students’ strengths and challenges and suggest methods for improvement will aid educators in offering personalized, effective, and data-validated pedagogy, while demanding less of their time. Personalized learning has always been highly labor intensive and therefore a privilege of the wealthy. AI can help put personalized learning within the reach of students and educators from all layers of global society.

Using AI to qualitatively analyze students’ work and automatically suggest materials and methods to personalize learning, however, has (to our knowledge) never been done before. The gains in efficiency and scalability that this approach entails can truly change education as we know it. A digital, easy to use platform for sharing these insights and innovations with colleagues will help the spread of this approach within and among institutions. By lowering the barrier to adopting new innovations and making sure recommendations are backed up by data, we can accelerate the rate of adoption of innovative pedagogy.

Our solution uses AI to aid educators personalize, refine, and share their blended learning courses. As such, AI and digital learning tools are indispensable to our solution. That said, the key drivers of educational success have always been and will continue to be students and educators. The technology we offer will considerably aid their efforts of personal and societal growth.

In September of 2018 we will start a pilot of our AI agent with one of our partners, to help them gain insight at scale into their students’ progress. We will continue our research into links between specific online student behaviors and learning outcomes, thereby training our AI agent. Based on the results of the pilot and our research, we hope to start more pilots with one or several of FeedbackFruits’ partnered institutions of higher learning. Furthermore, we hope to start beta testing of the Teachers’ Network (albeit without an AI element) at the University of Amsterdam in early 2019.

Over the next three to five years we will refine Insight’s descriptive, predictive, and prescriptive functionality and make it available to all current and future institutions that partner with FeedbackFruits, thereby reaching tens of thousands of students. The Teachers’ Network will likewise be rolled out to all these institutions. Insight will be integrated into the Teachers’ Network to automate the process of finding the innovative pedagogical methods shared by colleagues.

Since we are a part of FeedbackFruits, we do not need to build and retain our own customer base. FeedbackFruits partners with institutions of higher learning for a minimum of four years. Once Insight and the Teachers’ Network are added to FeedbackFruits’ suite of tools it will be available to all the educators and students of those institutions. The strength of this suite of tools facilitating blended learning, coupled with the analytical and sharing capabilities of Insight and the Teachers’ Network should convince the institutions to extend the contract beyond those four years. 

FeedbackFruits’ suite of tools is currently used by almost a thousand teachers at twelve Dutch institutions of higher learning and these numbers are growing. All the tools in this suite offer analytics, showing teachers such data as which students have finished an assignment and how many comments they have placed. When our research has shown us how we can distill specific insights from this raw student behavior data we will start offering AI analysis to our users. We are likewise iterating designs for the Teachers’ Network, where user friendliness will be an important factor in adoption amongst our users.

In 12 months we expect finish several Insight pilots, testing our hypotheses for the link between specific online behaviors and learning outcomes. We can then refine our solution and conduct broader beta tests. The qualitative component of our AI agent is currently being considered as a potential search engine for MIT OCW, potentially reaching millions. The Teachers’ Network will be rolled out first at the University of Amsterdam, and within 3 years at all our partnered institutions. We expect our solution to serve about 1,000 users in 12 months, and in the order of tens of thousands in three years. 

Our teams combined backgrounds in AI development, UX design, classroom experience, and deep connections with a host of teachers at a broad range of Dutch and international institutions of higher learning put us in a position to propose solutions to real-life didactic challenges that real-life teachers and students will be able to adopt. We are in constant communications with teachers in the field, in cooperation with whom we design and build practical, technological solutions. 

FeedbackFruits' charges yearly licensing fees to institutions of higher learning for the use of its suite of tools. These fees cover maintenance, feature development, and support costs. Additionally, institutions are charged fee for co-creative partnerships through the EdTech DoTank. Through these partnerships, institutions can propose educational challenges in answer of which FeedbackFruits then designs and builds technological solutions. The partnership fees cover the development costs of new features and functionality.

Licensing agreements between FeedbackFruits and institutions remain valid for 4 years, guaranteeing a steady source of income. 

When released, the Teachers' Network will be available for all institutions/users of FeedbackFruits' suite of tools to use. Information shared on the network will be available to all, which will attract new institutions to FeedbackFruits. For a fee, institutions will be able to white-label the Teachers' Network, allowing educators at the institution to share their insights exclusively with colleagues at the same institution.

We believe that the Solve community can be very helpful in validating or research into the connection between online student behavior and learning outcomes, both through academic research and partnerships for pilot programs. Furthermore, we hope to add more innovative educators to the Teachers' Network through the Solve community.

Finally, we believe that the exposure that comes with participation in the Solve community can help drive FeedbackFruits' international expansion, which would benefit our solution as well.

The main barrier is pinpointing which online student behaviors lead to which learning outcomes. Solve's community of academics and educators can help us tremendously in this regard.