Since the launch of ChatGPT in late 2022, there has been much talk of the impact of generative artificial intelligence (AI) on the way we work and live. There is even discussion on AI’s “existential threat to humanity,” which is either potentially catastrophic or overhyped, depending on who you listen to.

Nonetheless, AI offers exciting benefits across all sectors, including the learning industry. It is already used in digital learning to enhance personalization and provide real-time feedback to learners, among other things. But the sheer speed at which AI is developing means learning designers now have more tools at their disposal, which will ultimately benefit the learner.

Patricia Santos, Cegos’ group chief, corporate offer and solutions, has been experimenting with some of these tools to enhance processes and content at Cegos.

“AI will impact every step of digital learning,” she says. “It is incredibly disruptive, in life and the way we work, and is going to become more advanced very quickly.”

So, how exactly will AI impact digital learning in the short term?

1. Instructional Design

A digital course needs to be designed in a way that learners can access, inspired by the latest pedagogy. And there is software available to do just that!

ChatGPT can easily generate training plans with remarkable sophistication. Other tools auto-generate courses based on prompts from the user. For example, you can take an article on “how to become a better project manager,” upload it to the system and it will generate slides and quizzes to facilitate learning. The algorithm currently produces somewhat superficial results, unreliable content and a very basic user experience (UX). However, it won’t be long before learning providers are using this software to create high quality tutorials, saving hours and hours of work.

ChatGPT is already providing superior answers to some in-app tutorials. Patricia Santos once had an issue with Microsoft Outlook and found the in-app tutorial unhelpful. In contrast, ChatGPT provided a three-step solution in seconds, and it worked. So, we already are seeing advantages to using AI, even if it currently lacks a genuine and authentic edge.

2. Production

Perhaps the greatest impact of AI on digital learning will be on how learning designers curate content and use AI software to create videos, images, quizzes and more.

Currently, for example, the process of producing a professional, polished video is quite cumbersome and expensive; actors need to be hired, studios set up and each video edited.

Some tools allow you to auto-generate videos of human-like avatars presenting your script. This saves enormous amounts of time (and money). However, the results are a little clunky. When Cegos experimented with avatars, the feedback showed that some liked it, but others found it difficult to engage with as it lacked authenticity. However, this is what AI can do now. In the future, developers will deal with these teething troubles and the results will be a lot more impressive.

Similarly, AI can be used to generate specific images based on user prompts. Such tools can produce surprisingly realistic results, so the days of searching for stock images that properly illustrate your messaging could soon be a thing of the past. You simply type in the right prompt — “learners looking excitedly at a computer screen,” for example — and your image is generated in seconds.

Finally, there are several AI powered translation tools that can easily interpret learning content so that it can be offered in several languages. While the result always needs a human eye to ensure accuracy, the process is accelerated with AI’s ability to detect speech and voice patterns in a relatively authentic way.

3. Interactions With Learners 

Large language models, a type of AI algorithm that uses deep learning techniques to understand, summarize, generate and predict new content, have put a rocket under advancements in AI. The result is content that mirrors human writing patterns in a way that is easy to engage with. While live feedback is nothing new in digital learning, the sophistication and range just got a whole lot wider to the point where AI can now be used to coach, albeit in a somewhat rudimentary way.

Some tools can monitor expressions, tone of voice and language to give feedback on presentation skills. Other tools can produce roleplay scenarios where the user interacts with a virtual avatar, programmed to ask the right questions and respond to whatever the user is saying as well as the way they are saying it.

AI won’t replace coaches and/or trainers any time soon. However, these tools give learners the opportunity to practice their soft skills and get personalized feedback, exactly when they need it.

This is all very impressive. But, as with any technological leap, there are limitations and pitfalls ahead.

First, we need to consider the ethical dimensions. Large language models for instance, rely on existing content produced by humans to generate new material. This has implications for intellectual property rights, with authors and actors rightly concerned that others are using their hard work for free and without recognition.

AI tools can also be used to generate and spread misinformation, as large language models sometimes pick up sources that are not reliable. This is why fact-checking is critical when using AI to generate content.

There are also significant limitations. Generative AI is still not able to:

  • Give specific knowledge without the right prompt.
  • Perform physical actions in the real world.
  • Recognize and give feedback to certain behaviors.
  • Recognize, respond and adapt to the personal, emotional and social needs of the learner.
  • Provide personalized advice that considers an individual’s unique context circumstances.
  • Make judgment calls that require moral or ethical reasoning.

In time, these limitations will reduce as AI becomes more sophisticated. And there is discussion on implementing legal limits, so that AI does not become so powerful that it is difficult to control.

However, there is no doubt AI will play a much bigger part in digital learning over the course of the next few years.

“We will always need human trainers,” says Patricia. “Since the pandemic, people have reported digital fatigue which shows the human touch is essential to learning. But we can certainly use AI to improve efficiency, so long as we use it with purpose and validate all content for authenticity.”