In my previous post I explored the subtle, yet significant, difference between brute force computing and ANN (Artificial Neural Networks), and how that distinction is slowly producing what one would call “true” AI products in the classroom.
According to a 2018 research report, AI-based technologies are primed to grow within the edtech market by just under 47%. ANN-based AI programs have enormous potential to redefine what we understand to be personalized learning, and while in the early stages, it is well worth looking at the innovations that are at the cutting-edge of Learning Management System design.
Read more: 4 Ways AI will be a great teaching assistant
AI Assess
Developed by University College London’s Knowledge Lab, the software is primarily designed to assess as students learn, thus removing the high-pressure, high-stakes exam format that many academics feel is an unsuitable way to assess student knowledge. The system is not only designed with high content inputs, swiftly able to assess the degree of “correctness” of answers and completed tasks, but “intelligently” and continuously assesses a student’s knowledge of the subject being studied throughout the teaching/assessment process.
Not only are the various tasks and steps framed in a way that builds conceptual understanding, but the longer students interact with the software, the more the AI component can learn about the student.
The program is able to assess not only wrong and right answers, but steps taken and hints used and then map that over previous tasks and assessments completed. In this way the program is also assessing meta-cognitive elements (what students think they know), how they interact with the program, as well as the student’s previous work.
The system can then predict how a particular student will do on certain assessments, comparing that with actual results give a far more in-depth and immediate picture of the student's current status of comprehension and knowledge.
Smart content
One of the earliest applications of “true AI” in education was the development of software that could rapidly and automatically disseminate and repackage existing educational content into smaller, more accessible and more personalized e-learning modules.
Software today leverages deep learning to assemble custom textbooks, that adapt to teaching methods, class-types and learning outcomes. Once teachers have uploaded their syllabus and material into the software, the system analyzes the content to create books and core concept material that is perfectly tailored.
Chat bots
Many chatbots are burdened with stiff linguistic style as well as rote and repetitive responses. AI is changing how some edtech companies are approaching the goal of creating intuitive, intelligent teaching bots.
Edtech Foundry’s chatbot Differ, is deployed to address standard administrative and enrollment queries, alongside more technical questions involving course work. Over time the chatbot teaches itself how to incorporate student responses and engagement into refining its answers and guidance.
Robotic Process Automation (RPA)
RPA is a way to design automated processing using AI and a number of projects have been specifically started to calibrate RPA specifically to help educators find and automate mundane or repetitive tasks, such as:
- Monitoring Attendance where RFID tags allocated to each student automate the roll-call and attendance process, where teachers and school administrators can simply download attendance figures from the automated system.
- Registration and enrollment is massively time-consuming - AI enabled computers and bots can learn to identify eligibility criteria in application forms, and produce short lists of eligible candidates.
- Administration of payments, student files, records, infrastructure, facility and resource usage, staff records etc can all be automated using AI enabled RPA software.
Grading
Computers have been used to grade and produce multiple choice tests and assessments for a while, but AI technology has advanced those tools to now being able to assess and grade essays. Not without its detractors, the software has gained traction in a number of states and districts, with 34 million student essays on state and national high-stakes tests being graded by software in 2017 alone.
Additionally, a range of high-quality grammar, proof checking and plagiarism apps enable students to load their papers prior to submitting them, to benefit from the software’s ongoing learning (from scanning increasingly more documents). Students gain instant feedback on everything from punctuation, grammar, poor spelling and plagiarism — tools that will invariably help catch language errors before the paper is even submitted.
Finally…
It is clear that while computers and digital communication technologies are making great strides in enabling classrooms and other learning environments, “true” AI has not yet hit the edtech mainstream. This is no doubt because of the long development cycles and fast-changing understanding and application of nuanced AI concepts; but somewhere soon, true AI software and programs will emerge — and as those systems learn how we learn, we will no doubt be able to teach more students, with better outcomes.
