October 31, 2022
Posted by: DevDigital
Artificial intelligence and related developments in data processing and learning have changed a variety of industries, including higher education, medical research, and medical care. This article focuses on how a principal element of artificial intelligence, machine learning (ML), has been changing how online college courses work, along with changing most other aspects of higher education. Education companies and educators have produced a wide range of ML applications for grading, testing, evaluation, and mentoring. All of those functions are just as critical in online higher education.
A Few Notes on AI (Artificial Intelligence) and ML
Artificial intelligence is the computer science discipline concerned with creating software and machines that can learn from experience, solve problems, invent, and so on, tasks associated with artificial intelligence. Machine learning is all about training programs to recognize patterns. An ML application will use an algorithm to “train” software using data created just for that purpose, or real-world data of some sort.
Supervised, unsupervised, and reinforcement learning are the three standard approaches to working with an ML algorithm. In supervised learning, the system has a set of inputs (data points, images for example) and a set of desired outputs, all labeled so the software can find a suitable path to the desired outputs. In unsupervised learning, the software works with unlabeled input data to create rules. Reinforcement learning uses feedback to correct the software as it searches for the best way to navigate a maze, classify essay answers, and so on.
Higher education, online or off, brings some challenges that are common to education and some that are unique. Artificial intelligence backed by machine learning is proving useful in streamlining administrative work, grading, and assessment activity associated with online college courses.
Chatbots for Questions and Answers
Fielding students’ questions about tests, assignments, readings, and administrative trivia can become time-consuming. Imagine a college with 2,000 online students taking one or two courses each. A chatbot is a partial answer to the flood of queries. That functionality is going to be limited. A piece of software that picks out key phrases in a question and suggests a link or a document to review is helpful, to some students. A better chatbot takes that generic advice and “learns” how well the answers satisfy people. The software can then adapt so it offers new and better answers.
Some people will save time by copying more of their source material than they should have. If they do not give credit to their sources, they are guilty of plagiarism. Spotting this kind of cheating is not easy but it is important. A machine-learning algorithm can be “trained” to find text that copies something else, particularly something readily available online. These AI-driven tools are helping professors and authors, while also creating a slightly fairer online learning experience for college students.
For evaluating teaching materials, understanding of the course material, and more. Researchers at the University of Michigan created M-Write, an automated tool for analyzing text. This tool speeds up the laborious process of grading essays, something that is so time-consuming it is not practical for a large class. Therefore, many instructors rely on multiple choice and true-false tests; Those instruments do the job even if they aren’t too instructive for the professor. M-Write allows instructors to achieve more insight into how well students really understand the material, without having to study each essay at length.
Helping Students Succeed
A logical extension of that text analysis example is using machine learning to look at other types of learning, such as multiple-choice tests, matching exercises, and case studies. Researchers at UC-San Diego developed an intelligent tutoring system for the EdX online learning platform. This system uses machine learning to help students navigate a path to success using quiz questions and exercises. The intelligent tutoring system then evaluates student performance and adjusts their path through tutoring. A similar system would be useful in any kind of online learning pathway, a series of online courses with practical exercises and frequent tests.
The “micro evaluation” tactic this tutoring system uses can help students and teachers in other ways. Students who take quizzes or complete simple exercises can get customized help. Teachers can learn where the course materials or the course structure could be modified to improve their online college courses or their in-person teaching. Doing the same work manually might be impractical because of the time demands. An “intelligent” piece of software can cut out most of the human labor.
Students often spend an excessive amount of time trying to find information or resources and may not be fully aware of what they need. Some education companies have found ways to support student learners in other ways, for example, by helping them find the resources they need. This is what online learning giant Course Hero has done, so learners can more easily navigate their vast database of resources. In effect, they created a smart search engine similar to the “recommendation engines” that social media companies and online retailers use to help people find fun videos or products they might like.
Machine Learning Offers Many Opportunities to Improve Online Education
Machine learning algorithms have improved testing, grading, and evaluation in education but there has been less attention given to higher education. With so many online college courses, machine learning offers new opportunities. Using those machine-learning insights to create or upgrade software is a complex undertaking though. If you have a difficult software development challenge, we invite you to contact us at DevDigital and schedule a short conversation about your project.