What modern online education looks like with Seturon
Source: Element5 Digital on Unsplash
The market for online education and learning platforms has received a strong boost in the CIS. Unfortunately, not all services in Europe can offer not only a similar quality of training, but also a similar visual design. As a result, Seturon already looks like a strong competitor to European platforms, despite its relatively recent entry into the market.
What is Seturon
Seturon is an LMS (Learning management system — editor's note), but it has a few features that are either not on the market at all or are very rare. One of the things that not all companies have is a full mobile interface. We put a lot of effort into this because we thought that the majority of people would be trained from a smartphone screen. This goes against the conventional wisdom that people learn mostly in front of a computer.
We also already have the tools to develop non-linear learning. Let me explain a little bit more about this tool. Most of the time when someone makes a course they put one of two methods of progression into it. The first is where you do the tasks in a strict order and you're tested. For example, you have tasks one, two, three and that's how you do them. The second is where you can go through the tasks and units in any order. That's what Coursera does.
But there are situations where you want people to go through the course in an orderly way, but with a focus on the knowledge and skills they already have. This is how it works: for example, a person has completed Task A, where they have learnt a new concept, say the 'right-hand rule'. They then move on to Task B, which is a test. If he has learnt the rule and passed the test, he goes on to Task C. If he fails, we explain the same 'rule of the ant' in a different way. That is, we do not send him to the C text but to the A2 text.
If we think the explanation is enough, the student moves on to the next task. If the concept was quite difficult and we need to check that the material has been learnt, we give another test on the same topic.
In this way it can happen that a person taking the same course will come across questions and tasks that he or she has not seen before, but which will help him or her to learn the material better.
Of course, all these forks are not necessary and we can do the usual linear learning. It all depends on the specific task. There are also situations where people need to be trained on an annual basis.
And so they don't have to go through the same material over and over again, we just give them a test right away. If the student has done well, great. We just give them a certificate and say goodbye until next year. If not, then we look at which subjects they are falling behind on and give them extra study to cover those weaknesses.
What Seturon teaches
Seturon teaches what the customer needs. We are not experts in physics, maths or cybersecurity. But we are experts in methodology. If a client comes to us and says, for example, 'I need people to learn how to work like this', we will ask an expert to explain everything in detail, and then we will make a course that conveys all the necessary information in a very clear way and, most importantly, for the right target audience.
In fact, we don't care what kind of course they want from us. We need a request, we need an expert from whom we can take all the important knowledge and make a course out of it. Sometimes there is no expert, but there is already existing training material that they want to translate into an online format. This also happens.
One of the most memorable courses we did was for a client who wanted to enter the English-speaking market, but his English was poor. He didn't need to speak directly to the audience, but he had a very good understanding of how to do financial modelling. He spoke English, but he had an absolutely horrible accent, it's called butchering.
How we solved the problem. First, he gave us all the text, all the knowledge that should be in the course. Then we found a good AI tool, recorded the client in a studio with good lighting, with a good background. We also recorded his consent to use his appearance and voice as an AI avatar. We then recorded a few more minutes of him speaking with a good accent, even worked on his pronunciation, and from that we made 6 hours of AI video with the avatar.
It turned out well, and we decided to offer this idea to people who don't have the time to record hours of content. Because 10 hours of tutorials is not 10 hours of recording, it is many times more. Especially since this tool is now actively developed, we learned how to make cool avatars there. After this case study, we used avatars four more times for other projects.
It's also very useful if you want to add something to your course. For example, there's a subject that's very complicated and the students don't understand it. And the teacher has flown somewhere and won't be back for a while, so we can't record it in person. We go back to the AI, feed it the necessary text and add that part to the existing course.
Who makes the courses
When Seturon entered the international market, we already had a team of really great and very cool specialists who are now exploring a new market segment with us. Most of them have 7 years or more of experience, which means they have caught the waves of popularity of gamification, avatars and other trends.
And since we're talking about course refinement, it's worth talking more about this aspect. We have a section called 'Analytics' on our platform where we regularly check the health of the entire platform and individual courses. You can easily track the number of participants in a course. If 50 people have applied and 48 have been accepted, it means everything is fine, they have received the emails and are interested. And vice versa - if only 15 out of 50 showed up, it means there are some problems, we need to understand what the difficulties are, where people have gone.
Then you can look at behavioural indicators in the course itself. The speed at which people are progressing, the percentage of tasks that are being completed. For example, if you see that up to a certain question or task everyone was progressing normally, and then there was some sort of huge dropout, it means that there is something wrong with that block. You have to go into it, look at it, ask the students themselves what happened. And based on the feedback and the study of the material, we have to come up with improvements and corrections.
Our job as methodologists is to make the final course understandable to the target group and to really help them absorb the knowledge. Therefore, we can come to the client and say that this concept is not working and suggest changes to the course, for example, adding more graphics, more visual material, and so on.
What is the methodology
Methodology isn't just a word, it's a science based on several principles. You need to know your target group, you need to know the material and you need to know the ground on which you're teaching. All these things are necessary for a successful educational process. And more often than not, these are things that are dictated to you by the client. You can't say 'let's teach like this', they have a clear request.
Based on that, we choose and develop a training methodology. It all comes down to these three aspects: who we teach, what we teach and where we teach.
We look at the age of these people. Because there are pedagogical methods when you teach children and andragogical methods when you teach adults. We don't teach children yet, we teach adults.
There are some important differences. An adult is no longer a sponge, absorbing a lot of useless knowledge, but has a lot of experience. And when you explain new concepts to them, they are likely to already have some baggage of knowledge and experience on which to 'plant' those concepts. Here's the thing: adults don't learn things just because you tell them to. They need to understand the value of the knowledge and ideally be given an area of practical application right away.
One of the key principles of effective learning is scaffolding, or the zone of proximal development. This is a rather old concept developed by the Soviet methodologist Lev Vygotsky. Its essence is that you need to understand what a person can do now, and in effect identify their zone of proximal development.
This is exactly how it works for us in the way described above. That is, we give a person a concept and offer them a task to do. If he or she can do it, great, we teach him or her more. If not, we explain it in a different way and offer to apply the knowledge again. So we have a lot of tests and checks to help people learn and apply the concepts we've given them.
We have auto-checks, which are tests. The tests range from the standard one-choice, multiple-choice and true-false to linking concepts and typing the answer into blank boxes.
There's also a concept for automatic checking, which hasn't been implemented yet, but we want to get there. It's the validation of some test items using AI. This will use a generative model that we will train to check the work of our students based on specific knowledge that we will put into this AI.
There is a block of assignments that are not yet automatically checked, for example essays or detailed answers to some questions. This is usually checked by a person who understands the concept and has the right knowledge. We have assignments where you have to give a presentation and write a separate document to accompany it.
This is the case for individual work, but there are many more options. We have group work where we sometimes ask students to test each other against a set of criteria. Of course, they only get the list of criteria and a template of what can be considered a good answer after they have posted their own solution to the problem on the platform.
Another type of group task is simple teamwork, where all participants in a particular group are assessed at the same time. This helps to test not only knowledge, but also how people interact in general. We recently received a request to create such tasks and we have developed them.
Why we need AI
We know the areas where AI is pretty good, and we know exactly where it's bad. For example, AI doesn't understand the meaning of the words you give it. It only analyses how often certain words are used in relation to each other. So it will be perfectly capable of doing a test based on some text, but it will not understand the meaning of that text, what its value is.
We are very fond of AI and actively use it, with rare but very important exceptions. We NEVER ask AI to write a course. That is, the structure of the course, the structure of the longreads, the selection of the knowledge we want to pass on to people - no.
That's why the entire core is always built and tested by humans, and only then can we connect the AI to perform tasks for which it is very well suited. For example, you can record a video, you can draw graphs, you can create tests based on the knowledge that is 'fed' to the generative system, some simple illustrations if you need to add them.
The AI is great at making tests. This is an amazing option in general, especially if you have 20 pages of conditional text: you'll get tired of composing questions yourself. The AI does it in a few seconds, all you have to do is check that everything matches the text. But even so, you can make it use only the knowledge contained in the material, and no other.
Thanks to the skilful and, above all, appropriate combination of AI assistants and the work of highly qualified specialists, we have been able to create a competitive and very high quality product that has already helped many companies to train their employees and increase team competence.