As technology and science advance, education must keep up to meet the growing needs of learners and the market. Digital learning has altered perspective, ultimately changing how educators teach and students learn. We’re seeing e-learning become a key part of education, from workplace training to university programs. What’s driving this change? Increasingly, it’s the use of artificial intelligence, particularly AI-generated synthetic videos. But can they replace traditional video production, or is it just a new trend that will fade away?
Our colleague, Daniel, with his fellows, conducted a study in order to answer this question.
In this article, we’ll explore evidence-based research and answer the most intriguing questions about synthetic videos. 🧐
Summary
- What are AI avatars, and how do they work in education
- The potential of generative AI for learning: a case comparison
- How we see the future of AI-generated content
What are AI avatars, and how do they work in education
The first questions to answer are: what are synthetic videos and AI avatars, and how does producing videos with them differ from traditional methods? Artificial avatars are digital characters that look and act like real people. They can mimic speech, gestures, and facial expressions, making them an alternative to traditional video instruction.
The process of producing such videos is the following:
When you convert a traditionally filmed video into a synthetic format, it significantly simplifies the creation and editing process. Instructors can change the texts as needed, modify the script, cut or add a scene (which does not require additional human hours to produce new content). AI avatars, like the ones created with Synthesia or Heygen, offer a way to streamline video production, making it faster and more cost-effective. Being able to create or update a course video in hours, with minimal human involvement, is what synthetic videos bring to the table.
Once the primary video build is complete, translating it into other languages becomes quicker and more streamlined, as the avatar can speak different languages.
Why educational institutions and businesses are adopting this technology
The EDUCAUSE Horizon Report: Teaching and Learning Edition highlights ongoing financial challenges in higher education. Declining enrollments, rising costs, and reduced state funding create long-term instability. At the same time, the demand for online courses continues to grow, making it crucial for institutions to rethink their budgets and find cost-effective solutions.
(Excerpt from the synthetic video created for our client)
AI avatars can scale up video production in ways traditional methods simply can’t. Imagine you need to quickly create 2,000 videos for five different audiences across five languages. How much time and money will it take to use traditional filming methods? Years? Synthetic video is the tool that will not only make this task easier but also make it feasible quickly. This makes synthetic videos a scalable and affordable option for schools, universities, and businesses looking to enhance online education.
We know this is achievable because we’ve successfully used AI for online course creation in the localization of educational courses for the sustainable energy workforce, with content produced at unprecedented speed using generative AI. Within a span of two months, we successfully produced over 70 hours of learning content in four different languages. See an example of a video we localized into multiple languages:
The potential of generative AI for learning: a case comparison
Ultimately, the key question to consider is whether learning outcomes differ when using a video created through traditional methods compared to a synthetic video featuring an AI avatar, both covering the same material. And we have the answer.
Let’s analyze the study, “Generative AI for learning: Investigating the potential of synthetic learning videos,” conducted by our colleague, Daniel Leiker, and his fellows. The paper explores how AI-generated synthetic videos compare to traditional instructor-led videos in an online learning environment. Using a mixed methods approach, Daniel’s group analyzed both learning outcomes and learner perceptions to assess the potential of generative AI in education.
Excerpt from the synthetic video created for the client.
What Daniel’s group did
The team developed a series of synthetic videos for our client as part of an initial research study to explore the potential of AI-generated synthetic videos.
The team sought insights into how these videos could impact learning outcomes and learner satisfaction.
👉 The experiment conducted
In an experiment with 83 adult learners, aged 18 to 64 (average age 41.5), two types of instructional videos were tested: traditional instructor-led videos and AI-generated synthetic videos. Participants, mainly from a global professional learning community, were shown content from a microlearning course. One group watched a video featuring a live instructor, while the other saw a synthetic video created with Synthesia’s platform, where a photo-realistic AI clone of the instructor delivered the same content.
(“Generative AI for learning: Investigating the potential of synthetic learning videos”)
Learning outcomes: effectiveness of AI vs. traditional videos
While the study found no significant difference in learning outcomes between the two video types, there was an interesting distinction in how learners perceived the AI-generated videos.
Participants who watched the AI-generated content were more likely to strongly agree that the video met their expectations than those who viewed traditional instructor videos.
(“Generative AI for learning: Investigating the potential of synthetic learning videos”)
This suggests that, at least in the context of this study, AI-generated synthetic videos were able to satisfy learners’ needs for explicit and engaging content.
Constructive feedback and areas for improvement
However, when it came to open-ended feedback, learners wanted more detailed examples and clearer explanations of concepts. While 47 percent of responses showed negative sentiment, it’s important to recognize that the feedback was largely constructive, with many suggestions focused on improving the content, not the medium.
This underlines that while AI-generated videos are effective, the quality of the instructional content remains crucial. Learners want to understand the material thoroughly, regardless of how it’s presented.
While AI-generated videos show promise, challenges remain, particularly in ensuring content accuracy and reliability. In this study, researchers mitigated these concerns by using expert-curated materials.
How we see the future of AI-generated content
The conclusions from this experiment are impactful. Not only does it reinforce the growing confidence in the effectiveness of synthetic videos, but it also paves the way for further innovation in generative AI within education. As an e-learning solution provider, we recognize that transitioning to AI-generated content can raise key questions for clients:
- Will this approach deliver the same impact as traditional methods?
- Can we stay within our budget and time constraints?
While embracing AI technology may seem daunting, our hands-on experience proves that its benefits outweigh the challenges. With the proper guidance, educational institutions and businesses can seamlessly integrate AI avatars into their learning programs, delivering both value and innovation.
A strong example from our portfolio is a project for a global organization that needed to localize extensive e-learning content efficiently, without sacrificing accuracy or learner engagement.
Following ISO 17100:2015 standards, we streamlined the translation process by combining our team of seasoned linguists with a video editing department that used AI-generated synthetic videos. This AI-powered approach allowed us to produce all the content without compromising quality, making multilingual e-learning easier, smoother, and cost-effective. For more information about the project, follow the link.
We encourage educators and businesses alike to explore the possibilities of AI avatars in their course creation processes, and we look forward to seeing how this technology will continue to shape the future of education.
