How AI Is Pushing the Boundaries of Education: The Use of AI in E-Learning and Innovations in LMS iTutor

Summary: AI is gradually being utilized in various areas of e-learning. The article summarizes commonly cited ways AI is used in e-learning and describes the practical implementation of AI for Just In Time and Micro Learning. In conclusion, it describes the implementation of AI in LMS, demonstrating that AI in corporate education can go beyond merely assisting with human resource training.

Use of AI in E-Learning

Artificial intelligence (AI) is finding increasing use in e-learning. Typically, AI is gradually being adopted in the following areas:

Content Creation

One of the bottlenecks in e-learning is content creation. The original idea of e-learning—that employees would gain access to a large number of engaging, interactive, multimedia e-courses containing the specific know-how of the company—faces long and expensive production cycles with high didactic and technical demands on creators. AI is expected to fundamentally assist with the creation of e-courses, and many authoring tools already have some AI features implemented. In multilingual companies, AI can ensure the translation and localization of materials into other languages.

Virtual Assistants

AI can be used to create chatbots that provide instant answers to students' questions about the subject matter, help with navigation in the LMS, assist with students' technical issues, and are available 24/7, thus increasing support availability and student engagement.

Automated Assessment

AI automatically evaluates open-ended responses in tests or assigned tasks, provides students with instant feedback, performs plagiarism checks, and more.

Data Analysis, Adaptive Learning

AI analyzes data on student behavior stored by the LMS, such as time spent in learning units, the number of lesson or SCO (Shareable Content Objects) launches, and errors in test responses. Based on this, AI offers students supplementary materials, repeats content according to their weaknesses, identifies engaged students and those at risk for instructors, recommends instructor interventions, or, conversely, identifies problematic content for revision.

Virtual Reality (VR) and Augmented Reality (AR)

In combination with AI, it enables the creation of complex intelligent simulations, for example, for medicine, engineering, or language courses, where AI controls, for example, the behavior of characters and environments.

Implementation of AI in LMS iTutor for JIT Learning, Micro-Learning, and Human Resource Substitution

At Kontis, we have been developing e-learning systems for the corporate sector for over 25 years, and throughout this time, we have aimed to set trends in this area. We have long focused on implementing AI in e-learning and learning management systems (LMS) in all the areas mentioned above. Below, you will find possibilities for using AI in areas other than those usually cited—such as JIT, micro-learning, and human resource replacement (HR)—including a description of the concrete implementation of such AI in LMS.

If we imagine an LMS as a platform containing all company training materials and know-how, which the company uses to educate its employees and provide them with the knowledge needed for effective job performance, AI implemented in such an LMS can also significantly help in the following areas.

Just In Time Learning and Micro-Learning

Learning is most effective when the learner has a strong need to master the subject matter. Graduates of prescribed e-learning courses or scheduled classroom training often show a sharply declining retention curve over time simply because, at the time of study, the graduate was overwhelmed with information without feeling the need to know the subject in detail. This need is only felt when they have to apply detailed knowledge in a specific work task. This was the reason for the emergence of Just In Time (JIT) learning—learning exactly when it is needed—combined with micro-learning, i.e., learning in short formats of exactly what is needed. Before the arrival of AI, the main obstacle to effective practical implementation was the difficulty of easily finding exactly what needed to be learned. With AI, an employee can formulate their request in natural language, and AI can find in the LMS, from a large number of courses, manuals, documentation, guidelines, etc., exactly what the student needs to study at that moment, synthesize and summarize the found information into a short block with links to the source courses and materials.

Human Resource Substitution

In a number of specific tasks, AI can even replace some employees. Instead of providing learning materials to an employee to help them complete a work task, AI can solve the task itself. For example, a customer submits a support request to solve a problem with a product. AI can either provide the support worker with a quick synthesis of all information related to the request and the product, allowing the support worker to "micro-learn" Just In Time and solve the customer's request, or AI can directly resolve the customer's request in a support chatbot and forward only requests requiring complex solutions and human contact to support staff. The latter significantly increases the capacity of the support department and streamlines the work of human resources in the support department. Similarly, AI services can be imagined for a number of other job positions in the company, such as AI chatbots helping the sales department, the department serving partners or suppliers, the HR department—handling daily employee requests or recruiting new employees, and so on.

For the implementation of JIT/micro-learning and HR substitution using AI in our delivered LMS iTutor, we have implemented the following steps.

Centralization of Educational Documents

The core of LMS iTutor is integrated with the document management system (DMS) iTutor Documents. Companies usually have training materials processed not only in the form of e-courses but, due to the above-mentioned costs of creating such e-courses, also in a number of simple documents containing guidelines, production procedures, instructions, rules, etc. These documents are in various formats, from text formats such as PDF, HTML, MS Word, Excel, PowerPoint, to graphic formats, and even audio and video files. These files are usually scattered in many places on the company's disk space, on its websites, or in various company repositories. LMS iTutor, like other common LMSs, can work with such scattered materials in different locations via links. However, thanks to iTutor Documents, all these training/know-how materials can also be centralized in a single DMS repository integrated into iTutor. Such centralization in iTutor Documents, in addition to the generally known advantages of document management in DMS, enables their effective processing using AI implemented in the LMS, with unified control of content access rights for students.

Conversion of E-Courses to Documents

Information in e-learning courses is usually presented in an engaging and interactive way for the student. Unfortunately, in this form, the information is less understandable for AI. The team e-course development tool integrated in iTutor, iTutor Publisher, has an interface that also transfers the content of e-courses created in it to iTutor Documents. During the transfer, it automatically converts interactive e-courses into a more AI-readable format in the form of PDF. For example, information that is not displayed in the initial form in the course and requires the student to click through or perform some more complex interaction is converted into a PDF with all texts already visible, which AI can then easily read and understand, without having to try to interact with the course itself to reveal such information.

iTutor AI for Handling Student Requests

The previous two steps ensure that all educational and know-how materials of the company are available in the central iTutor Documents repository in the form of documents, with unified management of specific users' rights to specific information. The iTutor AI module has been implemented in iTutor to address JIT Learning and micro-learning as follows:

Automatically converts information in newly published versions of documents into plain text, a format that current large language models (LLMs) can best process for information analysis and synthesis.

Splits the converted texts into smaller parts and creates so-called embeddings from them. An embedding is a numerical representation of the semantic meaning of text—what the text is about—represented by multidimensional numerical vectors with thousands of dimensions, which numerically represent the distance (similarity) between individual texts.

Stores the created embeddings in a vector database. This is a repository of multidimensional vectors, where it is possible to search for the closest vectors using various mathematical methods.

Processes user requests for study formulated in natural language by converting the request into an embedding and searching the vector database for embeddings closest to the request. This provides links to information likely to address the topic of the request. It checks whether the requester has reading rights to the texts corresponding to the found embeddings. If not, it iteratively searches for other close embeddings, with the user having rights to the content of the texts from which the found embeddings are made. The texts corresponding to the found embeddings, together with the user's request in natural language, are processed by iTutor AI using the selected most suitable LLM and provide the user with an answer in the requester's natural language, including links to the documents from which the answer was constructed. The user can study the answer—it is a short synthesis of educational materials, i.e., micro-learning studied using the JIT method. Alternatively, the user can open the provided links to the sources in the answer. If the source is a document, they can open or download the document; if the source is a web page, the user can go to the website; if the source is an e-course, the user can launch this e-course in the LMS and study the material interactively and multimedia-wise.

The above description is simplified for the purposes of this article. The described process also involves means of maintaining the context of the conversation, setting similarity scores for embeddings, determining the appropriate LLM, automatic translation of source materials into the requester's language, and many other aspects that go beyond the scope of this article.

iTutor AI Chatbots

iTutor AI enables the creation of automated AI chatbots that can be placed on any HTML page. To create an iTutor AI chatbot, it is only necessary to create a virtual user in the iTutor LMS and assign them the appropriate permissions to documents in iTutor Documents. The iTutor AI module generates the required chatbot, whose JS code can be placed on any company page in the intranet or internet.

For example, in LMS iTutor, a virtual user can be created and assigned similar permissions to documents as employees in the "Salesperson" position in LMS iTutor. The chatbot generated for this user can be placed, for example, on the company's website, where it will answer questions about the company's products and services to potential visitors, similarly to salespeople. Thanks to the central rights management in iTutor Documents, the chatbot can be precisely tuned to what it should know. This chatbot probably will not have access to the company's product price lists, which real salespeople do. This would result in the chatbot creating specific price offers for customers, which it probably should not be entrusted with—that is the job of a live salesperson. However, it can be granted access to documents describing the basic principles of price construction, and additionally, it can be given access to a document with instructions on which salespeople to contact and how for a specific product or service price offer. When asked about a price, the chatbot will only describe to the requester what the price is usually made up of and provide a link to a specific person in the company to obtain an exact price offer.

iTutor AI has its own API; in addition to the automatically created chatbots described above, it is possible to program custom complex chatbots, supporting, for example, voice input/output, video, or an animated avatar, and any other imaginable functions used in chatbots. iTutor AI shows administrators records of user questions and chatbot answers, with feedback ratings from questioners. This allows for the gradual improvement of the materials on which the chatbot's answers are based, thereby improving the chatbot's responses.

Practical Example

Finally, here is an example of a simple chatbot that answers basic questions about our company and its products and services based on information gathered from the website where the chatbot is placed. The chatbot also has access to a few additional supporting documents uploaded to iTutor Documents for its operation. Creating such a chatbot in LMS iTutor takes just a few minutes.

https://www.kontis.net/