Corporate Training

Creating Online Courses with Neural Networks: Practical Examples

Creating Online Courses with Neural Networks: Practical Examples / Skillbox Media

A recent analysis of the corporate training market, carried out by the Digital Learning community, revealed that over the past year the percentage of organizations using artificial intelligence tools to create educational content increased from 3% to 13.5%. Among these companies, the insurance company SOGAZ-Med stands out; exactly one year ago, its training center began integrating neural networks into the full cycle of developing online courses for its employees.

There are organizations that use artificial intelligence technologies to develop educational products according to individual requests. One such company is the WINbd Management Academy, which is actively exploring the latest solutions, including digital speakers.

In this article, we present fascinating examples shared by representatives of various companies at numerous online events.

Optimizing the process of generating visual content using digital avatars

On June 24, a webinar was held, organized by the WINbd Management Academy in collaboration with the Digital Learning community, on the topic of "Preparing for the New Reality: AI vs. L&D." During the event, Danila Drapeza, head of the AI ​​and Neural Networks Competency Center at WINbd, shared the company's experience using generative neural networks to create customized educational programs.

First and foremost, this involves automating processes related to the creation of educational materials, which methodologists previously performed traditionally, without the use of technology.

  • Working with data arrays. For example, artificial intelligence studies questionnaires with program participant feedback, generating general conclusions and recommendations for potential improvements.
  • In ChatGPT, it is possible to upload a course presentation, from which the artificial intelligence generates a glossary containing important terms and their explanations. Furthermore, the neural network facilitates the rapid adaptation of content created for one group of students to the needs of another. For example, if an expert needs to prepare a presentation for schoolchildren, the AI ​​will rework educational materials originally intended for professionals, making their style more accessible and simplifying complex concepts.
  • Creating landing pages based on templates. As the speaker explained, the organization participates in the nationwide project "Active Measures to Promote Employment," which regularly implements educational initiatives that require standard landing pages. To optimize time costs, WINbd experts developed a template with a predefined landing page structure. This template is loaded into ChatGPT along with the new program's data, after which the neural network organizes the information according to the established pattern.

However, the most interesting applications of artificial intelligence are those that Danila Drapeza himself described as experimental.

In 2022, WINbd launched a free online course on building a career in the 21st century, developed entirely using neural network technologies. Ideas for the modules were proposed by ChatGPT, which also created the text content, but all material was reviewed by experts. In addition, other language generation models created illustrations for the course, as well as digital speakers who voiced the lessons.

Screenshot: "WINbd Management Academy" / Rutube

Student survey results showed that the majority of participants found the content engaging. However, opinions regarding the level of engagement provided by the digital speakers were mixed.

Danila Drapeza noted that LLM technology has advanced significantly since then. While in 2022, virtual speakers seemed "rather mechanical" and the quality of static image generation left much to be desired, today the situation has changed dramatically. As an illustration, he cited a modern project management course created for Sechenov University. Two methodologists and one expert in this field participated in the development of this course, and teachers were not involved in recording the video lectures.

Screenshot: "WINbd Management Academy" / Rutube

Using ChatGPT, the materials provided by the expert were reworked to ensure they were understandable to any audience. The visual design, a course map, was created using Midjourney in the school's corporate style and colors. Additionally, all resources were complemented with an audio version recorded with a synthesizer voice, as well as a video featuring a virtual "talking face." This allowed participants to choose the most convenient learning format. The speaker noted that this approach proved highly successful—the course demonstrated excellent completion rates.

In a project for Sequoia School, a business education provider in the tourism industry, WINbd is taking an innovative approach. As part of this collaboration, an educational blog dedicated to the application of artificial intelligence in business was developed. Danila Drapeza served as the expert, transcribing his lectures on this topic, thus creating an extensive text database. After this, the school's founder, Daria Loseva, was recorded on video, and a digital avatar was created based on her voice and appearance. This avatar was synchronized with text materials, which made it possible to obtain realistic video lectures that create the impression that they are being given by a real live speaker.

Screenshot: "WINbd Management Academy" / Rutube
Screenshot: "WINbd Management Academy" / Rutube

Later, a more "cartoonish" image of Daria was developed for Sequoia, which the school could use in short videos for social networks.

Danila also cited another interesting example. A large corporation with tens of thousands of employees approached WINbd to create a digital avatar of its CEO. However, the goal of this project wasn't to record video lectures. The virtual director was supposed to greet each employee in a corporate messenger, informing them of important company news.

Danila Drapeza shared examples of situations where digitizing a real person to create an avatar proves difficult or even impossible for various reasons. In such cases, fictional characters, or more precisely, completely virtual images, come to the rescue.

WINbd developed images of fictional characters, creating unique portraits and various voices for both male and female characters. After these images were approved by the company's management, they began to be used to record internal video courses, voice materials in the knowledge base and in marketing activities.

Screenshot: "WINbd Management Academy" / Rutube
Screenshot: "WINbd Management Academy" / Rutube

According to the speaker, modern capabilities of artificial intelligence significantly simplify the work of professionals, allowing you to save both time and financial resources of the organization. The virtual teacher is immune to illness and won't cancel a shoot, eliminating the need to rent a studio each time to record multiple takes, which then require lengthy editing. This is especially useful when there is a need to regularly update educational materials.

Furthermore, after beginning to integrate AI into its development, WINbd ended its collaboration with the design agency that previously created its visual materials. Now, the internal content department, using neural networks, independently prepares all visual content, which has significantly accelerated the completion of tasks.

Nevertheless, Danila Drapeza emphasizes that "a neural network is just a digital tool that allows a person to delve deeper into research," and any content it creates, be it text, images, audio, or video, must be carefully reviewed by specialists.

Artificial Intelligence as a Tool for the Full Cycle of Personnel Training

The HYPER / DIR-2025 conference was held on June 25 and 26, focusing on artificial intelligence and its role in the educational process. During the event, Natalia Kubrak, a business coach and methodologist at SOGAZ-Med JSC, shared her experience using AI in creating online courses, describing how she and her team integrate these technologies into their work. Natalia reported that six specialists work in the corporate training center. The company has 4,500 employees, for whom a variety of programs are developed—from technical skills to general training, as well as courses on occupational safety and training for a succession of qualified personnel. Mandatory courses typically have high attendance rates, while employees show significantly less interest in optional classes and rarely initiate their own learning. However, with the implementation of artificial intelligence tools, they have managed to increase the number of participants in optional courses. Natalia Kubrak and her team use neural networks to solve a variety of problems related to course creation. Artificial intelligence creates a profile of the target audience for a course, for example, on teamwork. It highlights the key skills and competencies required for successful mastery of the material and identifies common business problems that can be addressed with this course. Furthermore, the neural network analyzes similar programs and current trends in online education, suggesting optimal formats for presenting educational content.

While developing a course on teamwork, Natalia came across an illustration depicting ancient Greek gods. This inspired her to the bold idea of ​​combining the images of gods such as Zeus, Athena, and Ares with Belbin's typology of team roles. ChatGPT helped her implement this concept by creating a course structure consisting of various modules and lessons, describing the characters' personalities, and offering case studies and assignments for students.

Screenshot: Direct Academy channel on VK Video"

Neural network technologies have also proven effective in situations where concept refinement is needed. In this context, Natalia offered a helpful tip: to improve the quality of the result, include in your request a mention of one of the creative techniques that you consider appropriate for the current task. An example of such a request might sound like this: "I need help with brainstorming. I want to work out the details of the course and develop a storytelling theme about the interaction of gods. What gamification elements can be introduced: achievements, Easter eggs, or something else? What types of practical tasks will help develop adult students' skills in teamwork? Help me refine this idea. Use the SCAMPER technique."

The speaker recommends experimenting with a variety of creative approaches to achieve the desired results. Such techniques include, for example, the Five Whys method, Six Thinking Hats, the Focal Object Method, lateral thinking, as well as the Jobs to Be Done concept and many others. Natalia Kubrak was particularly interested in the expert panel method implemented with the help of ChatGPT: if necessary, the bot can be used in the role of a marketer, methodologist, creative director or other specialist, and the neural network will offer ideas or comments based on the selected role.

Read See also:

Five Ways to Develop Creative Concepts for Corporate Training.

Once the concept took shape, the neural network began supporting the SOGAZ-Med methodological group at various stages of course creation. For example:

  • Unfortunately, I cannot create a table in text format. However, I can help you create a course syllabus in the form of text describing the content, topics, and activities. Please specify what kind of course syllabus you would like to develop, and I will be happy to help!
  • To develop a slide presentation plan based on the course scenario, you can follow the following structure:

    1. **Title Slide**
    — Course Title
    — Name of the Author or Instructor
    — Date and Venue

    2. **Introduction**
    — Course goals and objectives
    — Brief overview of the content
    — Importance of the topic

    3. **Course structure**
    — Main modules
    — Topics of each module
    — Course duration

    4. **Teaching methods**
    — Class formats (lectures, seminars, practical classes)
    — Resources used (books, articles, online materials)

    5. **Expected results**
    — Knowledge and skills that participants will acquire
    — Application of the studied material in practice

    6. **Assessment and Certification**
    — Performance Assessment Criteria
    — Certification Requirements

    7. **Conclusion**
    — Summary
    — Recommendations for Further Study

    8. **Questions and Answers**
    — Discussion Time
    — Answers to Participants' Questions

    9. **Contact Information**
    — Ways to Contact the Course Author
    — Additional Resources and Links

    This outline will help you create a presentation that is clearly structured and easy for your audience to understand.

  • Unfortunately, you did not provide the source text or topic from which I could base the text for the slides and ideas for the illustrations. Please specify what information exactly needs to be processed or what the presentation should be about.
  • prepare sample assignments for students, as well as testing questions;
  • create graphic materials;
  • 1. Defining Content Goals:
    — What is the main goal of creating content?
    — What result do you want to achieve?

    2. Audience Research:
    — Who is your target audience?
    — What are their interests and needs?

    3. Competitor Analysis:
    — What approaches do competitors use?
    — How can you stand out from the crowd?

    4. Idea Generation:
    — What topics are relevant to your audience?
    — How can you diversify your content?

    5. Choosing a Format:
    — What format will the content be presented in? (article, video, infographic, etc.)
    — What style would be most appropriate?

    6. Outlining:
    — What is the structure of the content?
    — What key points should be included?

    7. Content Creation:
    — Who is responsible for writing and editing?
    — How will the creation process be organized?

    8. Optimization:
    — How will you optimize the content for search engines?
    — Are you using keywords and meta tags?

    9. Fact Checking:
    — Have all the facts and data been checked for accuracy?
    — Are the sources of information referenced?

    10. Editing and Proofreading:
    — Who will be responsible for the final check?
    — Have all aspects, such as grammar and style, been examined?

    11. Publishing:
    — When and where will the content be published?
    — Are there any scheduled dates for publication?

    12. Promotion:
    — How will you promote the content?
    — What channels and platforms will be used?

    13. Evaluation of Results:
    — How will you measure the success of the content?
    — What metrics will be used for analysis?

    14. Feedback:
    — How will you collect feedback from the audience?
    — How will the information obtained be taken into account for future projects?

The expert recommends providing the neural network with as much detailed information as possible and, similar to creative methods, clarifying the desired approach to completing a specific task. For example, before asking ChatGPT to create a module scenario, it would be wise to first download an article on the Kolb cycle and specify in the request that the result be based specifically on this model.

Natalia Kubrak also emphasized that the neural network copes quite successfully with creating test questions that require multiple-choice answers. However, the development of practical tasks is still the responsibility of a methodologist on their team. Artificial intelligence can offer the methodologist some ideas for such tasks, but at the moment it is not able to independently develop them so that they effectively contribute to the development of the necessary skills, are feasible within the framework of the course and meet the technical requirements of the LMS.

Read Also:

Five skills that educational designers are beginning to reconsider in light of the use of neural network technologies.

Nevertheless, as the speaker notes, content created using a neural network requires careful verification and analysis. It is necessary to identify valuable elements that correspond to the interests of a specific audience and help solve the stated educational objectives, and then combine these elements into a coherent whole. Manual work is also involved in the process of creating a course in the designer.

The neural network allows for the creation of a variety of creative options for course titles, slogans, and annotations that will help in its promotion among employees. In addition, it can be used to prepare the texts of emails with training invitations.

The experience gained by the training center indicates that SOGAZ-Med employees highly appreciated the use of AI storytelling in the course's promotional materials. For example, to create an announcement for a course on multitasking, Natalia prepared the following prompt:

Friends, hello! We have amazing news for you! We're launching a new course that will help you become true multitasking masters at work—especially in the insurance industry.

As you know, working in an insurance company requires the ability to handle many tasks simultaneously, and our course is exactly about that! It consists of three exciting modules:

1. **Planning**—learn to prioritize and organize your time so you get everything done.

2. **Organizing your media space**—we'll create a comfortable and productive work environment where you can focus on what matters most.

3. **Anti-stress** – we'll master techniques that will help you cope with stress and stay calm even on the busiest days.

Each module includes practical tasks and fun games to make the learning not only useful but also interesting! At the end of the course, you'll take a final test to reinforce all your knowledge and skills.

Don't miss the chance to improve your multitasking skills and make your work more enjoyable and efficient. Join us and become a master of your craft!

  • What is this course?
  • Who is this course for?
  • How is the course structured?
  • In a small town, there lived a man who was famous for his ability to handle several things at once. His name was Alexey, and he was a master of multitasking. Every day, he woke up early to get everything he had planned done. He had a lot to do: working at the store, caring for his family, playing sports, and even finding time to read. One day, an elder known in the area for his wisdom came to see Alexei. He watched Alexei fussing, trying to do everything at once. Seeing this, the elder approached him and suggested an experiment.

    "Let's do this," he said. "You take two buckets of water, one cold and one hot. Try pouring them into one bowl at the same time."

    Alexey, surprised, agreed. He filled both buckets and began pouring water into the bowl. However, as soon as the cold and hot water met, they began to mix, and as a result, Alexey received only lukewarm water, failing to maintain either of the original temperatures.

    The elder, observing his efforts, said: "You see, Alexey, when you try to do too many things at once, you lose the essence of each one. Like cold and hot water, they mix, and you get neither in full."

    These words got Alexey thinking. He realized that while multitasking may seem attractive, sometimes it is worth focusing on one task to achieve the best results. From that day on, he began to plan his tasks more thoughtfully, devoting sufficient time and attention to each of them.

    Thus, Alexey learned to value the quality of his work more than the number of tasks he could complete at one time. And with each passing day, his life became more and more harmonious.

  • "Multitasking is a myth we create to justify our inability to focus on one task."
  • A success story serves as a powerful source of inspiration.
  • Studying on the course requires a thoughtful approach and organization. First, it is important to carefully review the syllabus to understand what topics will be covered and what skills can be acquired. Then, it is worth setting aside regular time for study to ensure consistency and in-depth study of the material.

    The next step is to actively participate in class: ask questions, participate in discussions, and share your thoughts with other participants. This will help you not only better absorb the information but also broaden your horizons.

    Don't forget about practice: try to apply the knowledge you gain in real situations by completing assignments and projects. This will help you consolidate the material and see its application in practice.

    It is also helpful to take notes and summaries to organize what you have learned. Return to them regularly to refresh your memory and test your knowledge.

    Finally, communicating with instructors and fellow students can greatly enrich your experience. Exchanging opinions and advice will help you better understand complex topics and establish useful contacts.

  • In this course, participants will acquire new knowledge and skills that will help them in the future. They will be introduced to key concepts and practical aspects, allowing them to confidently apply what they have learned in real-world situations. Each module is aimed at deepening their understanding of the topic and developing the necessary competencies.

One of the stories created by ChatGPT tells the story of a glassblower who is faced with the problem of choosing between several significant orders. However, true success comes to him after he realizes that constant rushing does not bring results. By easing his hustle and bustle and learning to manage his time effectively, he found a path to prosperity. This narrative form resonated with employees, and subsequently, the use of storytelling in announcements attracted significantly more students to the optional courses than before. Natalia Kubrak notes that the works generated by neural networks have many similarities, often repeating the same plot motifs and phrases. Therefore, at present, one shouldn't expect astounding creativity from neural networks. The SOGAZ-Meda Training Center uses ChatGPT to collect feedback from participants after completing the courses. Students complete questionnaires, and the most valuable part of these documents are the open-ended questions that provide important ideas and insights. The neural network can generate a variety of question types that stimulate thought and encourage genuine feedback.

When a course is completed by multiple participants, their responses to open-ended questions can be uploaded to ChatGPT for further analysis, allowing the neural network to draw overall conclusions. However, Natalia noted that the neural network does not always accurately perceive the emotional aspect of reviews and tends to average out ratings. For example, if one extremely negative review out of 500 positive and neutral reviews is found, this could serve as a significant indicator of any shortcomings in the course content, the learning process, or student support. However, the neural network may overlook this review, treating it in the same category as the others.

According to Natalia Kubrak, even the free version of ChatGPT has extensive capabilities, and the quality of the result largely depends on the formulation of the request and the information the user provides to the neural network, taking into account information security aspects within a corporate environment. At the same time, the process of developing training aimed at achieving a given goal remains a task that requires manual intervention from a methodologist.

The speaker also noted that certain unresolved issues exist. For example, who exactly has the right to be considered the creator of the generated content, and who owns this intellectual property? Furthermore, how will employees taking the courses perceive the material if they learn it was developed using neural network technology? Natalia expressed her sincere opinion, admitting that she would not want to take a course created entirely by artificial intelligence.

The speaker raises an important question about the degree of trust in the analytical conclusions drawn using artificial intelligence. Natalia Kubrak notes that neural networks can be useful for analyzing employee feedback, but it is important to independently verify the results. At the same time, relying on AI analytics when making serious business decisions is not recommended.

Read also:

  • One Russian company has developed digital avatars for teachers, which will be used in the training of employees.
  • Research demonstrated which artificial intelligence models are capable of performing the functions of teachers.
  • Five common mistakes when integrating artificial intelligence tools into the process of training and professional development of employees.
  • More recently, new agents based on artificial intelligence have appeared that are capable of transforming online training in the business environment into something completely different than it was before.