Tristan Gardner

People often throw around the term “AI” to mean “the omniscient thing that will make whatever I want automatically and at a quality better than I could on my own.” This is short-sighted and gives the human brain a severe lack of credit, at least for now. Before diving into machine learning over the past year, I admit I was definitely one of those people. Let me break down from a high-level what “AI” means as it relates to enabling continuous measurement and individualized learning in the workplace.

Simply put, AI is about programming machines to think and learn like humans. LLMs are a specific structure of AI modeled and trained on the complex relationships between words, their subcomponents, and how they combine to describe the world around us. They have the ability to process and generate text in a way that mimics human understanding. This opens up new possibilities for content delivery and management in video education, such as improving the search-ability and accuracy of transcriptions and translations, making learning materials more globally accessible.

‍

How AI and Large Language Models Are Shaping the Future of Learning

‍

As technology evolves, it's clear that AI and Large Language Models (LLMs) have a significant role to play in the field of learning and development. LLMs offer a way to make complex data, language and relationships more understandable to computers which in turn provides more meaningful output to us users. Internal learning design teams will be enabled to facilitate personalized and engaging learning programs without additional hires or longer hours, neither of which do companies consistently have free cash or time for.

‍

Making Video Learning Accessible

‍

Video modules are already a key component of digital education, and transcribing them is key to making learning accessible to everyone. LLMs can improve the speed and accuracy of video transcriptions and translations, and even tackle the challenges of localizing content for different cultures and languages. A difficulty in localization arises in handling cultural references or linguistic structures of speech in one language that don’t translate to another. For example, the idioms below do not have any direct translation to an idiom common to native Spanish speakers.

‍

“Let the cat out of the bag” - Used to describe revealing a secret, this idiom doesn't have a direct Spanish counterpart with the same imagery.

“Kick the bucket” - This is a euphemism for dying. In Spanish, there's no idiom that involves "kicking" and "bucket" to mean the same thing.

‍

You often only find content transcribed in a few languages after core content is made in English. Now, we can serve any language because transcription costs, even with AI, are so low and human editorial time is decreased.

‍

Localization quality is increased by transferring meanings/concepts between languages, not just literal translations. With image-recognition models added to the mix, we can also derive the meaning from on-screen visuals like graphics and BRoll which can be added to our transcribed data. This is something that we’ve only been able to do with human reasoning until today, and it’s incredibly beneficial for learners who aren’t sighted. 

‍

Further, models can be tailored specifically to education content on topics like leadership, human capital, environmental responsibility to increase performance further in a certain field. AI-powered transcribers can pick up on nuances like proper nouns and report names that typical processors often misspell. These capabilities make learning more inclusive and accessible for everyone, regardless of any knowledge domain, a learner’s location or native language.

‍

Boosting Interactivity and Searchability

‍

One of the exciting developments in learning platforms is the potential for increased interactivity and searchability, thanks to LLMs. This technology could allow users to find specific video snippets that answer their questions without having to watch entire lessons, making learning more efficient and tailored to individual needs for more immediate cases or diving deeper into a concept. 

‍

When an LLM is fed the rest of educational media on an LMS, the model can make relationships between all of the content on the platform. Connecting information in the right ways increases its value to a probing user, just like the internet did for the world. Now, consider that this web of content can be connected to the historical data of a learner’s progress and their engagement. The model can draw meaningful conclusions about their current level of understanding and, because it also generates language, suggest the next best path forward for that individual learner. Filling personal knowledge gaps rather than those of a company or department becomes a realistic possibility for under-resourced management teams.

‍

Personalizing Learning Journeys

‍

Perhaps the most significant impact of LLMs is their potential to create personalized learning paths. With the ability to analyze a learner's engagement and progress, LLMs can suggest customized content that addresses individual strengths and weaknesses, making learning more effective and engaging.

‍

I recently met with Judy Lieu, a young innovative learning designer who’s had roles in researching learner engagement and comprehension. High-touch programming, which refers to personalized and interactive learning experiences, often requires significant resources to execute effectively. Judy's innovations demonstrate how authoring tools combined with ChatGPT can revolutionize this area, making personalization scalable.

‍

She points out that one of the key challenges in virtual high-touch programming is maintaining the level of engagement and personalization at scale. By integrating LLM capabilities as I’ve described, such as advanced search-ability and tailored content delivery, LMS/LXP platforms can offer personalized learning experiences that were once only possible through hands-on management of individual learners.

‍

As Suora encounters engagements that suit these innovations, we are excited to introduce our clients to new possibilities that advance the project as a whole and result in better learning outcomes. It is core to who we are to add these new use-cases to the array of our capabilities, to continually advance them and to share our experiences with the broader community in a new age of product innovation.