The Future of Learning: What to Expect in Digital Learning in 2025

Our latest webinar was a timely look at the key trends in digital learning on the horizon in 2025. And it follows neatly from last week's blog, which explored 2024's significant innovations and developments.

Joining the Digital Learning Institute's Tom O'Sullivan, the moderator for the discussion, was John Kilroy, DLI Founder and Dr Luke Hobson, Assistant Director of Instructional Design MIT, author, blogger, and podcaster.

The discussion was wide-ranging and insightful, and you can catch the full webinar here.

Here, we share some of those insights organised around the following three main themes to emerge from the discussion:

• AI tools, especially generative AI

• Personalised and adaptative learning

• Universal design for learning and accessibility

We also examine the skills, key tools, and technologies needed for tomorrow's workforce.

Tom kickstarted discussions by saying that even though we've barely scratched the surface of AI's potential in our everyday work, we've all perhaps become fatigued by talk of it. He challenged John and Luke to share real-world examples of AI in digital learning to motivate educators and L&D professionals.

John advised that the DLI recently launched an AI for Learning Certification program, where students identify a use case. He said there have been several innovative use cases, including one from a university with a foundation course with low engagement and a dropout rate 25% higher than other courses. The university wanted to understand why and leveraged AI analytics tools on student performance and behaviour to do so. According to John, it turned out that a lack of support around assessment was the main issue for students. The university has now been able to address that and has reduced the dropout rate by 5% in just one year.

'In 2023, there was a lot of experimentation going on,' declared John. By contrast, he advised that this year had seen plenty of good examples of use cases with tangible results. 'I suspect that will catapult us into 2025 as well,' John said.

Luke agreed, saying that higher education incorporates AI into daily practice in many exciting ways. At MIT, Luke and his team helped a colleague create a virtual avatar using D-ID to deliver course announcements. 'We were able to clone his voice, have him write a script and create a visual representation of him to deliver his messages. That way, he's not recording videos of himself all the time,' he advised.

Luke said many universities also use AI tools like Google's Dialogflow for virtual TAs and support systems within courses and learning experiences.

John said he had recently come across research from Harvard University about the impact of AI tutors, showing that it has helped double the learning gains compared to classroom learning.

He said the DLI is trailling using AI to provide initial feedback to students on their portfolios and assessments, and so far, students have been enthusiastic. However, he added that some have queried to what extent the data will be used in their formal assessments. 'So, it's important to be really clear about where the role of that AI assistant and tutor fits in and where it doesn't,' cautioned John.

Luke pointed out that AI is also having a massive impact on personalised and adaptative learning.

'We don't all start in the same place on our learning journeys,' he advised. 'Some of us have far more prior knowledge, and some of us are going into a subject matter that's completely foreign and brand new to us,' Luke continued.

He said that thanks to generative AI, we can adjust the level of difficulty for course content based on user data. 'We can change that based on having that personalised journey on the back end, while still giving learners the flexibility and the freedom of control,' advised Luke. According to him, it's vital that students retain control of their learning.

He went on to say that personalised learning is more challenging in higher education settings than corporate ones. 'Making all that work in regard to [tertiary education] standards, regulations and accreditation is difficult,' he pointed out.

John agreed and said it's perhaps easier to implement personalised learning in the corporate sector. 'Personalised learning has been the holy grail for so long. With generative AI, there’s definitely an opportunity to do personalised learning at scale and with 24/7 access,' he said.

Tom moved the discussion along to modes of learning. 'There's a growing feeling that the [post-pandemic] compromise solution of blended and hybrid delivery modes is not actually a compromise at all,' he said. 'It's potentially the best possible solution for outcomes,' he added.

John agreed and advised that as learning designers, it's essential that we get comfortable with various modalities and modes of delivery, including immersive experiences. However, he suggested that we need to have clear standards and a stronger sense of what good looks like in different delivery modes than we do now.

Luke emphasised the value of student feedback. Addressing student and learner needs, wants, goals, and individual situations is essential.

Tom advised that one of the best ways to improve learning for everyone is to implement the universal design for learning (UDL) principles.

According to Luke, the framework is widely accepted in the US. He also said he had created an open-source UDL Pal on ChatGPT. You can ask questions about UDL and get an answer on whether you are on track. 'And now, with the capabilities of GPT4, you can upload your own files and ask it to review that from a UDL lens,’ Luke advised.

He also said instructional designers should follow the example of big companies like Microsoft and Google. Before releasing a new product, they pilot it with potential users with accommodations and impairments and get their feedback.

John agreed that there has been a push for UDL to be recognised as a pedagogical framework. He said the UDL Guidelines were updated and reissued in July. 'It's generally based on the how, the why and the what of learning. But they've also introduced the who of learning,' said John. 'This goes back to what Luke was saying earlier about making sure that all decisions we make are learner and student-focused.'

The arrival of generative AI has eliminated one of the main excuses for not implementing accessibility: the amount of time it takes. John said there are many tools, like Ludia, that can take some of the load off designers when it comes to accessibility.

The panel agreed that micro-credentials had gained momentum on both sides of the Atlantic. 'For the first time, US universities are adopting this with intentionality and purpose, which is great,' declared Luke. He added that 22 states in the US have removed the job requirement to have a degree, which opens the door to more credentialling.

Micro-credentials offer the short, flexible, more accessible learning that people want, especially if they are stackable. 'What we're seeing with micro-credentials is that they should be industry-focused,' said John. Furthermore, students want to understand how they fit together potentially into a larger qualification. So far, according to John, there's been significant investment into micro-credentials, but the uptake has been low, suggesting there's work to be done on how micros are marketed to students.

All panel members agreed that we're unlikely to see a reduction in the digital learning job market. Nor will there be a requirement to have a background in, for example, machine learning or natural language processing.

John advised that most tools already have AI features, or they soon will. 'We won't spend as much time in Articulate or any of these authoring or video production tools because AI will assist us and make it a much quicker process,' he said.

Instead, instructional designers will have more time to spend on other critical tasks like collaborating with stakeholders and subject matter experts, mapping out learning experiences, or meeting accessibility standards. 'We're going to have more time for creativity, innovation and talking to our students than we had before,' declared John.

Luke agreed and said that the emphasis will be on learning to use AI tools and functions to get the most out of the platforms. 'We don't want a person who doesn't know what they're doing and is now designing awful learning experiences at the speed of light because it's a click of a button,' warned Luke.

According to John, learning sciences can help designers and educators make the right decisions about implementing AI from a pedagogical and learning experience perspective.

The discussion finished with the panel sharing their top picks of the latest tools and technologies.

Luke's top tools are as follows:

• GPT4, Claude or Gemini – generative AI

• Adobe Premier, D-ID or Runway – video creation tools

• HeyGen – video translation tool

• Suno or Google's AI Test Kitchen – music creation platforms

• Anthropic and Claude for learning analytics

John's selection of new technologies included the following:

• 7taps – microlearning creation platform

• Padlet – digital collaboration tool for creating lesson plans and rubrics

• LessonUp – front-of-class teaching tool

• ID-Assist – AI-enabled instructional design tool

• Canva – graphic design platform

One clear thing from this insightful and engaging discussion on digital learning in 2025 is that the pace of change won't slow down any time soon. So, hold on tight. It's going to be an exciting ride.

And if you want to keep up with emerging trends, discover how you can join our CPD program to continue learning and sharpening your skills with new short courses and practical toolkits released regularly.