By Robert Porter for HR Daily Advisor
Since the dawn of 2023, AI has become the dominant topic of discussion in business, technology, medicine, education, engineering, design, the arts, and public policy—nearly everywhere. Corporate eLearning is no exception, with generative AI tools already available that can aid instructional designers and architects of eLearning platforms. Much of the focus this year has been on Large Language Models (LLMs) like ChatGPT, Bard, and Claude. These tools allow anyone to generate text with a prompt.
Even during the early availability of ChatGPT, these capabilities were demonstrated for tasks, including generating scenarios for training courses. We asked the LLM, Claude, to “write a scenario for a corporate training involving an employee making a revenue accounting error that gets the company in financial trouble,” and it generated a story about John, who enters a 2% revenue increase rather than a 2% decrease (and all the mishaps and solutions that follow).
Creating content for a one-hour training module can take 50-60 hours, and today’s LLMs can help instructional designers cut that time down significantly—not by replacing them, but by accelerating the brainstorming and drafting process with content like John’s revenue accounting error.
LLMs and Beyond
It’s likely that everyone reading this has experimented with an LLM and can probably envision an application for the technology in their work. ChatGPT has become old news—what excites us is the vast possibilities AI poses for corporate eLearning in the coming months and years. LLMs are beginning to be integrated directly into Learning Management Systems (LMSs) to help course designers leverage these same content generation capabilities dynamically. For example, powerful chatbots can answer learner questions, clarify and reframe concepts, or reduce or increase the level of detail depending on learner preferences.
Beyond LLMs, generative image, audio, and video technologies enable learners to request that content be presented in a different format, like a chart rather than a block of text, or vice-versa. Other machine learning technologies power functions like adaptive learning, where AI can tailor training to individual learners’ strengths and needs, allowing employees to learn at their own pace. AI can also help identify when skills must be refreshed and suggest refresher courses to prevent skill decay.
While LLMs can help designers create course content today, the ultimate power of these technologies will come when they’re integrated with one another and directly into training modules and LMSs. This integration will come in the very near future, as LMSs, including Adobe Learning Manager, allow for straightforward builds of custom application pods within their existing environments.
AI in the World of Headless
The basic architecture of Headless offers the most incredible possibilities for AI in eLearning. With a Headless LMS, a custom learning environment can be built atop an existing LMS platform, allowing for the power and reliability of a platform like Adobe Learning Manager, but with the look and features tailored to learners’ needs.
With different kinds of AI technologies working together—LLMs, generative image, audio, and video, adaptive learning, and automated feedback—AI will make Headless LMSs to the exact needs of learners. In other Headless architectures like Headless content management systems, a single corpus of content is maintained that can be deployed through various channels like email marketing and web content.
By establishing a library of vetted, raw training content, an organization can create different kinds of training portals to the same content tailored, for example, to different learning styles (visual, auditory, read/write, kinesthetic). AI commentator Matt Wolfe has even demonstrated how ChatGPT can be used to create video games with no coding experience. This function can be used to create gamified features tailored to hands-on, kinesthetic learners.
In the ultimate endgame (as far as we can see it now), individual learners will be able to leverage AI to call upon a library of raw training content and generate the exact training modules they need with just a prompt: “Build me a revenue accounting training module that meets my organization’s annual training requirements. Give me lots of real-world examples and explain any technical terminology in detail. I am a history buff, so please set up all scenarios using historical figures as characters to keep my attention.”
Try entering the prompt above into your LLM of choice to understand how close we are to this technology working in practice.
Will This Really Happen?
Twelve months ago, the current capabilities of LLMs seemed like a pipe dream. It’s challenging to make accurate predictions in such a rapidly growing space. Many concerns about AI remain that could delay or derail some of the future possibilities we’ve discussed. Some corporate eLearning and training are designed to instill standard practice, policy, and procedure. Can those goals be met in a technical environment defined by hyper-customization to individual learners? LLMs are currently limited by so-called hallucinations, in which they dream up facts and present them confidently. Such a problem would be a disaster in a training context.
How exactly AI will revolutionize corporate eLearning in the coming years is impossible to predict. What is certain is that AI will revolutionize corporate eLearning—because it’s starting already.
