A robot contains the puzzle pieces a human is missing.
Talent Development Leader

AI Will Democratize Learning

Tuesday, April 25, 2023

Here’s the good that can come from using artificial intelligence in L&D.

Breakthroughs in artificial intelligence often lead to some degree of public unease. But the recent launch of Open AI’s GPT-4 tool sparked something much more dramatic: a wave of true panic, including experts suggesting a pause in AI development.

Although there is certainly much to figure out, we think there are many promising applications that could use AI for the betterment of humanity, especially in the context of learning. In a world where only 7 percent of the global population has a college degree, and as many as three quarters of workers don’t feel equipped to learn the digital skills their employers will need in the future, this is the conversation people need to have.

High-quality adult learning is a scarce resource. It’s expensive to provide and hard to get right. It involves stark trade-offs between outcomes and scale—something learning leaders think about every day. Personalized tutoring produces outstanding results, but educators can only offer it to a few learners at a time, while open online courses reach millions of learners yet have completion rates in the single digits.

As veterans of both educational technology and AI companies, we see enormous potential for AI to fundamentally reshape those constraints, making professional education affordable, accessible, and effective for anyone, anywhere.

In particular, we’re betting on four trends for AI and L&D.

Rapid content production

Until now, content production has been a major limiting factor in education. Companies invest millions in developing and updating courses for their employees. A single course can take many months to produce. However, with support from AI models like ChatGPT, an instructional designer could create an end-to-end curriculum in a matter of days.

While today’s models are far from perfect, they’re already capable of dramatically accelerating the early phases of content generation—enabling instructors to focus on curating and fine-turning content rather than creating it from scratch. Imagine the productivity increases that could create with your L&D team.

In addition to generation, AI models show increasing promise in converting between content formats—for example, turning text into video or vice versa—making it much easier for a single instructor to serve learners with many different learning preferences. Instructors can likewise translate content between languages, virtually instantaneously, all but eliminating fluency and nationality as barriers that multinational companies often face when producing content.

Personalized content

Learner motivation can break into two buckets: “I can” and “I want.” Learners need to believe they can succeed—they need challenges that push them but still feel attainable. They also need to want to succeed, which means they must see how the material they’re learning links to their personal interests and goals.

Creating a curriculum that meets those criteria for every learner is virtually impossible for a human instructor. Yet just as AI models can generate content in seconds, they can also personalize that content for individual learners in real time.

An AI model may respond to a learner’s behavior by, for instance, adjusting an assignment’s difficulty or dynamically inserting examples and scenarios that are more relevant to the learner’s experience or industry. Ultimately, L&D leaders could see a world in which AI models create fully personalized course versions for each learner, with every content element adapted to maximize motivation and learning impact.

Detailed, continuous feedback

Learning is an active process. It involves an individual trying new things, making mistakes, and—critically—receiving constructive feedback. However, feedback in traditional learning delivery models has always been marginal at best.

That is partly because it’s based on limited information; instructors see learner responses to a handful of formal assessment questions and hear directly from the few learners who are willing to raise their hands in class, but that is just the tip of the iceberg when it comes to what’s really going on with overall learning and understanding.

Further, responding to that limited amount of data takes time, which means feedback is typically sparse and often surfaces days to weeks after a learner misses an exam question. Feedback is so challenging that we often don’t even try to incorporate it as part of our learning modules.

Advanced AI models could change that overnight, particularly in the context of online professional education. Such models can continuously observe each learner in an online course, taking in millions of data points to develop a nuanced picture of the learner’s engagement and understanding. They are already capable of providing rich feedback in real time, even on open-ended assessments.

Ultimately, AI models could turn routine feedback into a true dialogue—delivering personalized critiques, answering clarifying questions, and providing follow-up assessments until the learner has demonstrated content mastery.

Learner-driven exploration

In the real world, learners’ questions and curiosities don’t fit neatly within the confines of a syllabus. That is especially true for adult learners, who approach education through the lens of their specific professional field or personal background.

We’ve seen that firsthand in our work at CoRise, where learners are constantly raising questions that no single human instructor could be expected to answer. In our machine learning courses, for example, learners have asked how individuals use methods such as time series forecasting in an enormous range of industries, from environmental science to supply chain management.

With tools like ChatGPT, learners will be able to move seamlessly between interacting with a curriculum’s core content and exploring applications or tangential content according to their interests.

A new world for learning

Taken together, these trends will change the cost structure of education and give learning practitioners new superpowers. Learners of all backgrounds will be able to access quality content on any topic and receive the ongoing support they need to master new skills. Even small L&D teams will be able to create programs that have both deep and broad impact across their organizations.

That vision isn’t a far-off future. We’re already experimenting with these models in our courses and enterprise partnerships.

That vision isn’t dystopian either—a future in which machines replace instructors. The goal is for AI to amplify humans by freeing them up to focus on the things only they can provide, such as inspiration, connection, and perspective.

In an AI-enabled education model, instructors will spend more time helping learners understand why they’re learning and less time scrambling to coordinate the day-to-day details of how they learn. Peer-to-peer interactions will move away from generic discussion prompts and toward rich, ongoing conversations that create space to bond over shared interests and explore other points of view.

There is no denying the power of AI to revolutionize the world and transform the way learning professionals work.

Read more from Talent Development Leader.

About the Author

Julia Stiglitz is CEO and cofounder of CoRise, an edtech platform helping companies equip their talent with cutting-edge skills to thrive at work. She previously was a partner at GSV Ventures, where she invested and partnered with some of the most innovative entrepreneurs in edtech. Prior to GSV, Stiglitz was one of the first hires at Coursera and led many groundbreaking initiatives. Most notably, as vice president of enterprise, she founded and led Coursera’s enterprise business, growing it from concept to more than 1,400 organizations across the world.

Stiglitz holds a BA from Pomona College, an MBA from Stanford Graduate School of Business, and an MA in education from Stanford University. She is a member of McKinsey’s Consortium on the Advancement of Adult Learning and Development.

About the Author

Sourabh Bajaj is cofounder and chief technology officer of CoRise, an online upskilling platform for individuals and businesses. He has extensive experience in artificial intelligence, building data products, and online education, having worked at Neeva, Google Brain, and Coursera. Bajaj is an alumnus of Georgia Tech and BITS Pilani.

2 Comments
Sign In to Post a Comment
One thing that came to my mind about the potential of AI democratizing learning is how the pricing of AI based tools will affect the process of democratizing of learning. In the end, if you are low on budget, you may not have the financial resources to use these AI based tools (but maybe this is not a valid concern).
That's a great point. In general technology costs have always come down throughout history if you look at air travel, cellphones, internet access so I'm optimistic that over time we'll solve for the this as well. We are already seeing this with research from Meta around models like Llama which can run on the iphone etc.
Sorry! Something went wrong on our end. Please try again later.
Sorry! Something went wrong on our end. Please try again later.
Sorry! Something went wrong on our end. Please try again later.