December 7, 2023
Introduction
Advances in Artificial Intelligence (AI) have generated great excitement due to the potential benefits in many areas.
Given the unique security and privacy considerations of the K-12 sector Edsby is exclusively focused on, the company has been taking a measured approach to contemplate where AI can most effectively and safely be applied. When it comes to AI, just because something CAN be done doesn't mean it should be, especially given the unique dynamics of K-12. Ensuring the safety of student data governs much of how Edsby works today.
This document shares Edsby's vision for AI, and describes a wide range of areas being explored. The company expects to begin releasing a series of enhancements related to this work in 2024, however the timing and exact functionality of what may be delivered are subject to commercial and technology factors. Do not consider this document a commitment, but an indication of intent.
Considerations
Privacy
Edsby is entrusted with highly sensitive structured and unstructured information about students, parents and staff. This information goes beyond simple “name, rank and serial number” data and includes assessment data, attendance data, observations, insights, comments, essays, messages, and much more. All of this can provide excellent “fuel” for new AI-based solutions, however there are considerations that temper what is appropriate to do. They include:
- Customers own the data they provide to the Edsby platform, along with the content their users generate within the Edsby platform. Edsby’s Services Privacy Policy ensures customer permission will be explicitly secured before providing data to any outside organization.
- Customers may consider solutions that carefully provide their data to proven, trusted platforms that provide the same sort of data stewardship that Edsby does. Customers should not consider solutions that mix and mingle their data with data from others, which is a shortcoming of many new AI technology platforms.
Any AI-based solution incorporated within the Edsby platform needs to provide the same level of strong data management that Edsby itself provides today.
Human Factors
Change management is always a large challenge in K-12 learning organizations due to their large size, diverse set of stakeholders and small technology teams for the number of users they manage. These organizations have only recently seen widespread adoption of online technology within the classroom and across key stakeholders including students, staff and parents. Careful consideration needs to be given when thinking about introducing new technologies to these groups:
- Students embrace new technologies quickly, however not all new technologies are appropriate for all their use cases. For example, students could attempt to use generative AI solutions to do much of the work assigned to them. However, by doing so, students would miss out on key aspects of the learning process that lay the foundation for future learning.
- Staff are keen to have solutions that save them time so they can focus more on teaching and working with students. They aren’t expected to entertain solutions that take away their ability as professionals to manage how they teach and assess. Nor are they, or any applicable labor organization, apt to embrace solutions they view as threatening to their job function.
- Parents are keen to be more engaged in their child’s learning journey. But they want to do so in authentic ways with quality input from professional teachers. Parents are rejecting solutions they view as mechanical and lacking appropriate personalized feedback from teachers. Such mistrust has arisen even from simple sources such as the use of comment banks by teachers for report cards.
Teacher AI Applications Under Exploration by Edsby
Teachers are the “quarterbacks” of engagement in Edsby. They have the widest range of tools in the Edsby platform, and, not surprisingly, Edsby's research to date has identified the broadest set of potential applications for AI tech for them. These are summarized below. Work has started on several of these.
Content Authoring Assistant: Edsby provides teachers with ways to author a range of content including lessons, assessments, class posts, journal notes, observations, and more. Our early experiments show the viability of adding an option that enables the teacher to provide content requests to an integrated generative AI engine. Examples would include “create a 200-word summary of the history of the country of New Zealand,” or “provide an overview of the krebs cycle.” Such an integrated approach could simplify and speed content authoring. This capability could also be tied into the way curriculum standards are stored in Edsby so the teacher could instead select portions of the curriculum (for example “A2.1” and “A2.4”) and request content that explains the standards selected in a student-friendly way. We have been experimenting with various tunings and AI engines that maximize the pedagogical value of the content.
Online Quiz Assistant: Edsby has a flexible facility for building online quizzes. It supports a wide range of question types. These are used for both summative and formative learning. It can take a teacher some time to build out an online quiz as they need to specify the question, one or more answer alternatives, and designate the correct answer. Early experimentation has confirmed the viability of auto generating Edsby question sheets through generative AI tools. This can be done by selecting learning material already in the Edsby platform, for example by selecting a Unit in the Content Builder, and simply providing instructions on the desired online question sheet composition, such as the number of questions of each type desired. The generative AI tool auto-creates an Edsby question sheet with an appropriate set of questions and answers in place. The teacher can then choose to edit, extend, or modify the question sheet. Edsby currently has an early version of this in development.
Grading Assistant: Edsby's powerful “Grader” facility in the Edsby Gradebook enables teachers to view, mark up, annotate, and score submitted student work in one integrated user experience. Today, Edsby integrates with external tools such as Turnitin that provide a higher level of analysis, such as plagiarism detection. However, such tools are expensive and can create a complicated user experience. An integrated AI solution could potentially provide initial analysis and feedback, including identifying grammatical and spelling issues, and even initial assessment evaluation information for the teacher to review and refine. Edsby believes this could reduce teacher time spent grading and simplify the teacher user experience when compared with using external tools, and is experimenting to validate this hypothesis.
Reporting Assistant: An important and time-consuming aspect of a teacher’s job is creating input for report cards. This usually includes both structured information (such as a grade) and unstructured information (such as a comment). For teachers to do this effectively, best practices suggest evaluating a wide range of considerations beyond a simple weighted average calculation from a gradebook. Teachers should consider the most recent performance trends, the level of consistency in performance, and so on. Edsby believes an appropriate integration of AI tools could help teachers analyse these trends for each student and help create the most accurate input. An AI-powered reporting assistant is expected to be able to generate sample comments, potentially by leveraging an existing teacher comment library and the student’s performance. This reporting assistant is also envisioned as being capable of sentiment analysis along with basic spelling and grammar analysis to assist the teacher in generating appropriate report card comments. Edsby intends to investigate the value of AI in this area.
Student AI Applications Under Exploration by Edsby
Students are envisioned to able to benefit from a number of integrated AI solutions within Edsby. These are currently expected to include:
Content Coach: Edsby believes it could be beneficial to students to have an integrated “content coach” that could analyze content the student has generated for spelling and grammatical issues, sentiment analysis, and overall clarity and coherence. This could apply to a broad range of areas, including class and group posts, submitted work, and even messages to teachers and other students.
Ask Edsby: Edsby recently introduced a new AI-powered chatbot experience for unauthenticated users on its public www.edsby.com website. This chatbot has learned how to respond to a large number of end user questions regarding how to get things done in Edsby. Our early work in this area suggests it would be relatively easy to add a role-based version of this into the Edsby product. The existing public website bot is restricted to Edsby-published help files as the corpus of knowledge from which it draws, and any similar system in the Edsby product is envisioned as being similarly restricted.
Data/Analytics AI Applications Under Exploration by Edsby
Edsby provides a powerful analytics facility that provides teachers, school administrators, and district administrators with insights into trends in many different areas including enrollment, attendance, reporting, student performance, and so on. The user experience for Edsby analytics today leverages modern business intelligence software and includes standard graphical features for filtering, drill-down, and so on.
There are a few challenges with this approach:
- The data views available are often constrained by what the dashboard developer felt would be useful, and
- The end user may not be familiar with business intelligence software and may be reluctant to explore and use power features like filters and drilldowns.
Business intelligence tools are now being extended to have built-in AI-based facilities for both generative AI and machine learning to provide for new and more flexible ways to leverage the underlying data feeding the tool. We expect to use these new AI-based facilitates to:
- Create more natural user experiences for users who are not experts in using business intelligence software. These user experiences are imagined as supporting both text and spoken requests. An example of such a request is “show me how many students in Grade 3 are currently averaging below 70% in mathematics and how that is trending over the last 3 years.” If a school administrator asks such a question, Edsby is exploring showing them the results for their school. If a district superintendent asks such a question, it is imagined as being able to show them the results district-wide.
- Create more powerful user experiences that leverage machine learning to provide predictive capabilities. Such a request could be similar to “show me the students in Grade 9 math that are likely to fail this year based on their performance so far.”
Conclusion
Edsby is encouraged by the promise of AI. Watch for a gradual introduced of AI-powered capabilities as we take a careful approach to evaluating where the application of this technology can provide the biggest returns with the least risk to our K-12 customers.