L&D + Gen-AI: The Next Chapter

Modern-day AI can create, critique, and adapt. With the right level of human involvement, we can save time and achieve better outcomes. The question is how. The answers are limitless.

In my last article, we covered the immense and largely unrealized potential of Generative AI that business leaders everywhere are recognizing and investing towards. L&D leaders can’t afford to ignore this opportunity.

But what does embracing Gen-AI look like in concrete terms? What exactly do you get when you cross large language models with traditional approaches to learning?

Let’s see if we can take a stab.

The bigger picture

The KPMG survey (“Generative AI: From buzz to business value”) that I referenced heavily in the first article contains little in the way of predicting what form Generative AI solutions will take.

That’s understandable, as currently, business leaders are working on their strategies and not yet implementing them. It’s the job of all respective industries to figure out precisely what form Gen-AI solutions will take in their field.

One question in the KPMG survey, however, gives an idea of where expectations lie.

What form will the impact of generative AI on the workforce take?

Source: KPMG Generative AI survey, March 2023

From this, we can see that the main impact will be on ways of working (rather than eliminating whole swathes of the workforce - as we have already established).

By enabling more efficient workflows that cut out time-consuming, low-value-added tasks, humans will have more time to create and collaborate.

That gives us a place to start.

The impact on L&D

Formal research is underway at institutions such as Stanford’s Accelerator for Learning and the Institute for Human-Centered Artificial Intelligence as to how Gen-AI can be used in educational settings.

Furthermore, a whole raft of apps is emerging to support educators - The Thinking Effect curates a list of AI Tools that are relevant for educators, from course creators to apps that will convert any website into a quiz.

The likely truth is that there is no part of the L&D value chain that cannot be improved in some way by the power of this new technology.

Let’s go through some of the main verticals.

#1 Instructor-Level: Creating educational content

Let’s start with the most obvious - learning materials.

Crucially, Gen-AI is not limited to creating text, but can increasingly generate audio-visual content as well. Future iterations - paired with AR/VR - are likely to create fully immersive environments where individuals cannot only chat but interact in more holistic ways.

Instructors can also leverage the power of AI outside of the seminar room - to generate lesson plans, co-create teaching notes, as well as provide first drafts for quizzes and assessments.

AI does not simply create, but can also modify tone and language for existing content, and even the level of complexity with which ideas are expressed. The same content can be adjusted for different settings and students with greater ease and speed.

#2 Learner-Level: Enabling individualized learning paths

Just as Gen-AI applications can be trained on specific datasets, they can also learn about and adapt to individuals.

This is useful in a self-learning context where there is no teacher. But it can also ease the strain on live instructors, who can now more easily keep track of and meet the needs of a disparate group.

Natural Language Processing (NLP) and sentiment analysis are supplemental tools that can process a user’s responses in order to assess not only their progress, but also their level of comfort with various learning methods (e.g. text vs. audio), and so suggest an optimal pace and format.

#3 Company-Level: Optimizing the human resource portfolio

Zooming out to the organizational level, AI can combine the data gained from individual interactions with a company-wide analysis of where the total employee base stands with respect to the desired portfolio of skills.

If a deficit is found in a certain area, this can reveal a need to add more modules to compensate and reskill, before going out to seek external hires.

Sophisticated analyses will consider not only the current state, but also project future needs based on the trajectory of the current resource pool and talent pipeline, enabling L&D to be perceived for what it is - a strategic rather than a support function.

The role of the human expert

As we have emphasized before, while Generative AI may be here to stay, the human expert is also going nowhere.

Marc Booker, Ph.D., Vice Provost for strategy at the University of Phoenix, has expressed his experience with AI in a learning context as follows.

“So far, our experience using AI in curriculum development has been more akin to having another partner to review your work to give you suggestions or different ideas to ponder, rather than a replacement for human capital.”

The most crucial question of all when exploring this topic is to determine as precisely as possible where the role of humans stops and that of machines begins. It is not likely to be a clear boundary, but more akin to a jigsaw or intricately woven tapestry.

However, a human expert should without doubt be involved before - and not just after - publishing or granting access to an intended audience.

Encouraging students to use uncurated Generative AI for knowledge discovery (cutting out the human intermediary entirely) can seem like an excellent time-saver until the AI generates false, incomplete information. We’ll get into this more in our next article on risks.

What this might look like in practice

As stated above, there is no single answer to the question of how to leverage Gen-AI in L&D. Even if there were - no one would have it yet. Now is the time not for answers but smart hypotheses.

In a provocative article written for Forbes, David James of the technology firm 360Learning describes how a new paradigm could work. Here is a brief summary:

■?????? Step 1: Macro-analysis: based on the data within the firm and the trends outside, who in the organization needs what in the way of training?

■?????? Step 2: Generate a “1st pass” solution: Use the findings of Step 1 to prompt AI to create and populate frameworks for solving the gaps.

■?????? Step 3: Bring in SMEs: human experts can now critique and curate the 1st pass to create a refined, final version before launching it to the community.

Going back to the earlier point about productivity, do you notice how much time is saved by going about the process in this way? Giving SMEs something to critique, rather than build from scratch, means they can now focus only on the highest value activities.

It’s also clear that the end result is likely to be more effective as it is based on a proven, targeted need. This is clearly preferable to a catch-all, one-size-fits-all course that is aimed at everyone but meets the full needs of no one.

Another important point, following on from this, is that a more targeted, more efficient process will allow us to create more interactions - perhaps on a daily basis - with our target users.

This is not just a nice to have. It is a necessary development in response to the increasing pace of change and the need for just-in-time education.

It also indicates just how far from the current model we might eventually end up.

Conclusion

The promise of Generative AI is indeed transformational. Finally, we have a learning tool that can:

• Create at scale
• Critique individually
• Adapt continually (both to the user and to the world outside)
• Move seamlessly across languages and cultures.
• Follow the guidelines that we set for it.

While remaining wary of the danger of trusting too much to artificial intelligence and failing to deploy human intelligence where it is genuinely valuable, we would guess that - if fully unlocked - the potential for GAI to fundamentally change the face and effectiveness of L&D solutions is limited only by our willingness to change.

The precise nature of this future can’t be predicted. As the saying goes, it can only be created.

The question is - by us, or someone else?