1) Content curation is dying.
2) Learning content is auto generated.
3) Generative AI search rocks.
The three reasons why engagement, learning and gaining expertise has gotten easier.
Up until the emergence of OpenAI’s ChatGPT, we used to think that besides search and building learning content we needed to either have enough people to mentor everyone, have meetings to impart instructions or curate and build long form learning content to train everyone on our company’s products or job skills. The ideal format of this is a capability academy: having equal measure on-demand courses, content, live coaches and mentors along with cohort based learning programs. These approaches work well but they require both resources to setup and maintain along with a matching ratio of personnel to learners which is effective but expensive. Enter ChatGPT. Now we can move to automate the coaches and mentors while even enabling ChatGPT to generate the courses. This is why we now more then ever believe there are some great reasons why engaging, learning and imparting expertise to everyone has and will be getting much easier, Here are our top 3:
Traditional long-form content development (concept, design, interviews of subject matter experts, execution and testing ) requires a great deal of thought in terms of organization, determining learning objectives, goals, design – to make it engaging as well as the time spent for interviews/ knowledge transfer sessions between subject matter experts and instructional designers. This is why today, very little training is produced for the unique job skills of employees in companies. Typically, it’s well produced, more generic and sold as third party content based on a given industry such as by Linkedin Learning.
Modern authoring tools available to us such as from Articulate or Adobe makes us wonder: What’s the ratio of time we have to spend to build every resulting minute of learning content? Is it 5:1? – 5 hours of effort for 1 hour of content? Bryan Chapman a number of years ago attempted to answer this question and had even more alarming ratios of 49:1. Then, what happens when this digital learning content needs to be updated? This is why typically only fairly static content like security awareness, harassment or other key HR topical courses which don’t change much are produced.
However, now there are emerging companies producing automated generative AI authoring approaches which can instead of taking hours or days to produce instructional content, can be accomplished in minutes. One such company, Nolej.io which recently was one of the top 20 finalists for the 2023 top 20 ASU/ GSV Cup which chose from 200 digital learning and workforce skills companies. Nolej.io is remarkable from our experience to produce a multi-page course with assessment and activities automatically generated in a few minutes. All we had to do was decide how many assessments, type of activities and drop in an existing video, pdf or other content. It did the rest. The entry cost: $19.99 per month. We envision instructional designers and authoring tool experts becoming more like conductors in an orchestra, instantly calling for different content to be produced by these tools, automatically pulling from worthy company videos, pdfs and documents which are then automatically updated. Even on the fly digital learning could be produced based on user’s questions or previous experiences. Tools emerging like Adobe’s Firefly point to this future.
2) Curation…will die a quick death. ChatGPT powered Sharepoint will take care of this for you.
The process of content curation, deciding the order and relevancy of content we believe will move to systems like Microsoft Sharepoint and it has already started by Microsoft combining ChatGPT to assist in content creation. You’ll then be able to hold a conversation with Sharepoint and have the relevant response, tutorial, generated example shown to you and pulled from the collective content in your Sharepoint. This is the case for most any emerging generative AI/ ChatGPT empowered content store as well. We won’t be so concerned about organizing content as much as having cross checking systems in place to ensure answers and responses generated are accurate. The challenge with current authored learning content is they are a collection of thoughtfully organized pages and chapters which serve to teach on multiple learning objectives. By design they are built manually or as above via generative AI as a process completed by sound instructional design principles to maximize the ability to engage and teach a learner. These long/ multi-chapter courses themselves are an act of curation because at the time it was necessary to ensure learners received the right combination of content to achieve set learning objectives. As said, it will be easier for the generative AI systems to themselves produce this content based on the guidance of instructional designers and then be automatically updated or changed based on the original set goals of the course.
3) The above AI systems will enable a level of understanding of context such that when a learner needs to know something the system will mimic a human coach and effectively tutor on any topic or answer any question.
Educational Psychologist Benjamin Bloom in 1984 posed a challenge right in the title of his research. He contended that one on one coaching produces two standard deviations – 2 sigma – of improvement to students learning a given subject versus those in a classroom. That means if they got a “C” previously while in a classroom setting they would get an “A” after learning the same subject from one-on-one coaching approach. Put another way, 60% more students achieve mastery with one-on-one tutoring in the same amount of time as learning the same material in a classroom. ChatGPT powered systems are definitely leading the charge for automated one-on-one tutoring. Time will tell if they achieve these same results.
Feathercap is your team’s answer platform. We leverage ChatGPT and our own context tracking to automatically generate and tailor the right answers and responses to every employee, vendor, partner or customer query. Current customer support, customer success and enterprise search relies on hardcoded workflows and content links to answer questions. Our customers tell us we save them time, money and energy managing their flow of knowledge and actions by getting everyone answers that they need.
Copyright © 2023 Feathercap Inc.