In this episode of the Cherryleaf Podcast, we look at how learning and how teaching is changing. We’re going to see a change, and probably a rapid change in what content, staff, and users are given.
Transcript
Hello again. My name is Ellis Pratt. I’m one of the directors at Cherryleaf. In this episode of the Cherryleaf podcast, we’re going to look at how learning or how teaching is changing.
I have recently been involved in an episode of the Visual Lounge Podcast, which will be coming out in September. And on that, we talk about the impact of AI video tools, such as Google Veo 3.
And I mentioned some of the potential uses in e learning in that episode. I thought it would be useful to expand on that. And also, there was an episode on the Scriptorium podcast about e learning, and, again, it thought it would be useful to talk about topics in additions to the ones that they covered.
And it is the case that teaching and e learning is undergoing a quiet revolution.
It’s still the case today that a lot of today’s corporate online training is as it’s always been. But what’s likely to happen is that we’re going to see a change, and probably a rapid change in what content, staff, and users are given.
So the current model works within certain limitations and boundaries.
It’s essentially around providing two main functions the straightforward transmission of facts, and the creation of a safe simulator space where people can practice new skills through different scenarios
And most e learning offerings within the corporate environment follow a familiar formula. A lot of training is provided through videos and in most cases, there’s a presenter who’s speaking and there is audio. You can see the presenter, you can hear the presenter. And to break things up, to keep the audience engaged, there are breaks from just seeing the presenter to seeing something else. Often that is another form of video to illustrate the point that the presenter is talking about, and this tends to be called B roll.
So if you’re talking about, say, on our course, why do you need technical writing, and you say, well, you need it to make sure that people operate in a safe environment, you can have b roll that shows somebody doing something that’s dangerous. On our course, on the foundations of technical writing, we have a piece of b roll of somebody on a boat in the high seas, and they’re putting up the railing where people get in and out of the boat, putting that back up, obviously, they need to do that in a safe way, they need training to understand how they can do it.
You also have images where things are included so that you can see a static image to illustrate a point
And at the end of the course, there’s often multiple choice testing to check that the user has understood what was in the course, and to be frank, often that is very simplistic and not necessarily the best way for somebody to test their knowledge or build up their skills.
The courses tend to be built in English because of the cost of translating into different languages in many situations is prohibitive and often the videos themselves tend to be generic.
So we when we are working for clients, sometimes what we need to do is to be onboarded and understand the organisation’s rules when it comes to things like cybersecurity or reporting any complaints that might come about relating to equality or other issues.
And whilst the course textual content can be specific to a client, the videos that are inserted into those e learning courses, you often get the sense that they are generic to every organisation. That’s whoever’s provided this e learning course for the client has created videos that can be repurposed and reused across multiple clients, and then they are adapting the text for each specific client.
So minimal customisation.
And there’s often single path that every student follows.
And in addition to that, there may be some community forums, there might be some live question and answer sessions or social spaces and it’s a means by which organisations can provide training content at scale, at cost, but with limitations.
Limitations in terms of how much the student can learn, how they can build their skills limitations in how flexible the course can be.
We’re now in a situation where many of those limitations can be resolved and artificial intelligence can help at lots of different points within the e learning journey from the creation of the content, to the assessment, to the ability to personalize each course for each student, or to a narrower group of students.
Let me explain how. Let’s start by looking at content generation. I mentioned B Roll, and in the past if you wanted B Roll to go into your course you would typically go to a site that had stock footage. And there are some sites that provide content free, others where you can subscribe and get access to a wider library of content.
And so for our example of having content in our course around somebody doing something in a situation that could be dangerous.
We had to search on keywords like working or office or danger, and go through a lot of different videos to pinpoint one that might be suitable for the course. And in some situations, there just isn’t a piece of B roll footage in those, sites that meets your need.
With new AI video generation tools like Google Veo 3 you can create video clips. They’re currently limited to eight seconds. You can extend them if you don’t want audio to go with them, but what you can do is you can create video scenes with people talking, and you can describe the scene, you can control the dialogue, you control to an extent the camera movements as well. So you can use a chatbot to say can you suggest a video scene that might imply that somebody or might suggest that somebody is doing something in a dangerous way. It can give you suggestions, you can pick one, and you can, from there, create a prompt you can insert into a tool that generates video.
People are also using AI to create summaries of training courses, and AI is quite good at summarising bodies of text.
Within learning systems, such as the Teachable platform that we use, it can automatically generate multiple choice questions based on the content
I mentioned earlier that there are limitations in trying to reinforce learning solely through multiple choice, but if it’s there in addition to other ways of checking people’s skills, it has its place.
And we’re also seeing tools from Google like NotebookLM.
What that can do is you can give it a collection of content and it can generate from that summaries, it can generate notes, it can generate a podcast that simulates a conversation between multiple people, And what’s now coming within the next few weeks is it can also generate video summaries of the content you’ve provided. And this is aimed at students at school, and at universities, so that and they can also converse, they can chat with the docs, they can ask questions about it.
And potentially, within a corporate setting rather than relying on individuals to do this, you can provide the content or you can provide a platform like NotebookLM and enable students to have e learning content through that means.
Another element is localization
In many situations, it’s not cost effective to create content for multiple languages
Now, there’s the potential to do that, to adapt content for different languages, different regions, even different contexts
And by doing that, you can make content more relevant and engaging for different audiences. So you can take an audio and have it transcribed into a different language, so it can appear that the person is speaking Spanish or German or French and there are also tools and platforms that can take the video and adapt it so it looks like the mouth movements are consistent or correct for that particular language.
Another aspect is AI powered avatars. This is something you can get with tools like Audiate
So you get an avatar, a digital representation of somebody, and you write a script, and you can select a language, you can also select what the avatar looks like and from that it will generate a video where the avatar is speaking the script that you have written.
Avatars offer some benefits, speed to market, you can write the scripts, put the script into the tool, generate the video. You don’t need the time to set up the studio, to book somebody, to do the presentation, to record it, you can do it much more quickly. And with the avatars in tools like Audiate and others, you can pick an avatar and stick with it. You can have character consistency. One of the challenges to do that these days with a platform like Google Veo 3is keeping that character consistency.
It can be a challenge. Avatar videos can run longer than the eight second limitation that you get with platforms like Veo 3 and they can offer providers of content a cheaper way of creating a lot of content. But there are limitations, and those limitations are that it doesn’t have the realism, the engagement of a real live presenter, or, again, the type of really realistic characters that you can create with some of the AI video tools.
Another aspect is the ability to offer personalised learning experiences
So with AI generated content because of its speed and the cost, whereas before it just wouldn’t be practically possible to create lots and lots of video content for many different scenarios, now that option is available. So you have the ability to provide branched training, different narratives. So you can ask
learners to choose between one path and the next, and you can create videos that show the consequences of those choices. You can create learning that’s more personalized, the choose your own adventure type learning experience, which can be more engaging and more memorable than just a single linear path.
We can do more than that with AI video.
In some countries there are limitations due to laws in different countries, but in some countries, what you can do is you can take an image of somebody you can create multiple versions of that person synthetically using AI, and you can take that image or choose from a selection of different images, insert it into a scene
and what the video tool will do is it will synthesise the two images to create a video.
So, for example, if we take onboarding. Somebody is going to join a company. You want them to go to a specific location in the building to ask for a specific person. You can write instructions, you can create images of which building, which door to go through, the picture of the person they’re going to meet
You can do that today. You can also create a video. You can create a video of them walking to the building going to the entrance that they you want them to go to, walking up to the desk and asking for the person they should be asking for and for that person to appear in the video and say hello good to meet you
So they can see the experience of what’s going to happen.
And you can do that for just a few dollars.
Another aspect is real time adaptation and feedback. With chatbots, you have the ability with AI to provide dynamic feedback for people to ask specific questions and to get answers back. So you can help them in a more richer way explain why an answer is right or why it’s wrong, not just whether it is.
What about the downsides? What are the things that we need to consider?
Let’s go back to something that I just mentioned, the ability to create content at speed. Some of the AI video generation tools come with an API.
Some of the AI video generation tools have APIs or will have APIs
so you can start to automate production of videos. You can start to have a site where you can drag and drop the image of the person that’s joining the company, drag and drop the image of the sites that they’re going to, enter the names of the new sales, enter the names of the people they’re meeting. That gets added to a prompt, and it generates the video. It’s there for it to be unloaded, and again, within your workflow that you can automate, it can be then stored locally.
One of the downsides of doing that is that you’re probably, at the moment, getting a 50% success rate in the videos that are generated. So half the videos that are generated will be wrong, will not be of a suitable quality. So we need a human in the loop. We need to take that source content and improve it.
You can’t have a fully automated workflow. You still need a platform like Camtasia, like Audiate, some of the learning management tools to take the content, bring it together, and to produce something of suitable quality. Now the AI reduces the time, reduces the cost of creating the source content, but you still need that
One big challenge is assessment.
And this is a big challenge for schools and universities and could be a challenge for corporations as well
You may well have seen reports about schools and universities worrying about essay writing. Today, a student can use a chatbot to generate an essay or find an answer very easily.
So how can we effectively test their knowledge.
We can decide to fight the technology, try to prevent them, try to detect it, or we can adapt the method of evaluation.
Some organisations are going back to live timed tests. The downside of that is the cost and implication of organizing those types of tests, it also is a challenge for people who are neurodivergent people like those who have dyslexia.
So a more modern and inclusive approach that some are adapting is to shift from checking if people can recall knowledge to critical thinking
Some teachers are asking their students to answer a question using AI and then to be assessed not on the answer, but to be assessed on their analysis of what the response was, how good it was, what the AI system got correct, and what it got wrong, what it missed.
And that approach is assessing a student’s true understanding of a subject. It’s making them think critically about the information they’re given, which is an important skill to have today within the work environment. What about the impact on careers in e learning and organisations that are offering services in the e learning space?
If your role, if your organisation, is all about providing the content, then you may have
some concerns, because the ability to do that type of work is at risk that it can be done cheaper, better, faster through AI.
If your role is to think more strategically about what it is that you want learners to learn, what experience do you want them to have, think at that more strategic, holistic level about the ultimate goal, then probably you’re in a situation where things were a lot more brighter because you can do more. You can offer better solutions than there are today, much more relevant to the students that are out there
We are still in early days. AI avatars are still limited. Tools like Google Veo 3 are very much in the early days at the moment in terms of length and quality and what can be done, but it’s obvious the direction of travel. And it is likely that the quality will improve within months, possibly even weeks.
And for any organisation, the adoption of new technology comes down to three questions.
Does it save time? Does it save money? Does it improve quality? The trifecta, as it’s called
So does using AI in e learning save time.
And the answer is it can do by automating repetitive tasks like generating quiz questions and summaries
creating initial video drafts.
It can dramatically cut down on the production time.
Does it save money? Again, it can do. It can reduce the need for time consuming lengthy video shoots in a studio or outside, the need for a video production team to record it or to the amount of people needed to edit it afterwards and the manual effort of localising.
Those are the main areas where you can see significant cost savings.
And can it improve the quality? Which is probably the most crucial question.
And if the will is there to improve the quality of the e learning that’s provided, then yes, it can. It can improve the quality by creating more interesting, engaging, high production content.
And also enabling experiences that were just impossible beforehand from a practical, from a technical, from a cost effective, from a time perspective.
So the use of AI in a e learning environment isn’t only about efficiency. It’s about the opportunity to do things that just weren’t possible before. Creating deeply personalised, engaging, effective learning experiences.
Content that can adapt to each individual, content that’s more has a greater impact than before.
So it’s a very interesting time. How do we balance that automation with the authenticity, with the quality that’s neededto build something that’s smarter, but also more relevant to each, more humane, more human to each student.
So, what do you think
Some of you, I’m sure, will have thoughts on this as a consumer of e learning content or as a creator of it. And we’d love to hear your thoughts
You can add your comments to our blog posts. You can email us at info at cheerleaders dot com. And if you’re interested in our e learning courses or getting us to help you create e learning content, then again we’d love to hear from you and you can contact us probably again best by email info at cherryleaf dot com
Thank you for listening.
Leave a Reply