Posted by Isaac Asimov needs no introduction to lovers of science fiction. One of the most famous writers of modern science fiction, his works have been adapted for film, television, and radio by various personalities. His greatest contribution perhaps remains the ‘Three Laws of Robotics,’ which serves as
the foundation of robot-design and Artificial Intelligence.
One of Asimov’s popular short stories is “Cal.” Cal is a robot that works for an author. As part of a thought-experiment, the author asks a technician to program the robot with a full dictionary and eventually Cal writes a short story that is full of humor. The author is shocked to discover that the quality of writing is much better than his own works. The author calls the technician and asks him to reprogram Cal back to its old self. Cal overhears the conversation and plans to kill the author.
Nearly three decades later, science and technology are at a juncture wherein fiction from the past is slowly turning into reality. Heliograf is an Artificial Intelligence-powered tool developed by The Washington Post’s tech team and came into prominence during the Rio Olympics for writing over 300 short articles. Since then Heliograf has produced several articles that have been published regularly. Forbes uses Bertie an Artificial Intelligence tool to help reporters with data, first drafts, and templates. Tools like AI-Writer offer a paid service that allow users to key in a headline or caption and generate a relevant article with cited resource links at the click of a button. It is now evident that Artificial Intelligence and specially programmed tools can create content on par with what a trained human can write; and it is just a matter of time and the costs involved, that’s preventing wide-scale adoption of Artificial Intelligence for content. In this blog post, we shall look at the challenges, benefits, opportunities, and implications of using Artificial Intelligence to create content for eLearning programs.
Table of Contents
Can Artificial Intelligence Think On its Own?
In mid-2017 there was a lot of hue and cry made by mainstream media publications about Facebook shutting down an AI program because the AI program developed its own language. As always, one should treat such news with a pinch of salt and do a thorough research of the facts. Unfortunately, we live in an era of clickbait articles and sensational news. Here’s a detailed post by the Facebook Engineering Group on what happened.
Quoting these lines from the Towards Data Science blog on the incident.
“Nothing scary, shocking or even note-worthy happened. Just a regular day with a regular scientific experiment. And then all hell broke loose as wanna-be journalists posted one doomsday article after another. Shutting down a chat bot when it stops showing a reasonable outcome is about as ominous as changing a faulty light bulb.”
Looking at the data available in the public domain, we are still some distance away from creating an AI solution that is completely independent and can make logical decisions on its own.
Artificial Intelligence & Content Creation
As stated at the beginning of the article several news agencies are already using automated tools to create a pool of content, which is further refined by human editors before publishing. In fact, we are seeing instances of several pieces of writing being regularly published without moderation. Chatbots in some form or the other are being used by various online service providers and telecom giants to moderate first-level complaints on websites and social media accounts. Online learning aggregators also integrate chatbots to guide learners to pick a relevant course or remind learners to complete a program on time.
In this section, we look at how AI can empower content creators to create content from scratch. The challenge for content creators in the world of eLearning is the vast variety of topics on which they work. In a mid-sized eLearning firm, the typical instructional designer or content writer would create content on topics as diverse as heavy engineering and micro-computing to gender sensitivity and financial literacy. The sheer depth of research and reading to be done on a vast variety of topics can be quite intimidating for a campus-hire. Unless the individual has the capacity to work on different topics and finds joy in writing on them it becomes a big challenge for the project manager to keep the individual motivated and engaged.
Opportunities and Possibilities of Using Artificial Intelligence to Create eLearning
Locomotives are an essential factor in the growth of a nation’s economy. Railway tracks are the key factor for the growth and prosperity of the United States of America. Careers were built, fortunes made, and new lands acquired as railway tracks were built across the length and breadth of America. From the early days of coal and steam powered locomotives, today we have trains running on electric power and solar energy and the usual fossil fuel-powered engines. At Origin, we work with one of America’s largest locomotive manufacturer and most of the training content that we create is for their trainee engineers.
AI and Virtual Reality can be used in partnership to determine scenarios that can lead to the damage and wear and tear of locomotives and engines. Simulations can be created that replicate the complex machinery that makes up the locomotive and the VR-experience can showcase how these engines and locomotives can be repaired.
Other applications of AI in eLearning could replicate advanced technology used in luxury cars when the automobile determines that the driver is under the influence of alcohol or is feeling sleepy and alerts them to stop the car or automatically shuts down the engine. The learning platform or the eLearning course itself could be designed to feature an engagement meter that measures the level of activity or engagement of the learner and based on reducing engagement levels come up with an interactivity / quiz / animation that would help regain the attention of learners.
Artificial Intelligence, LXPs, Netflix, and the Content Recommendation Engine
Netflix started out as a DVD rental and sales company. Look where it stands today? Netflix produces its own original entertainment content, and houses entertainment licensed from different studios that it streams to millions of users globally. What sets apart Netflix from other similar streaming services like Amazon Prime Video or Hulu is the amount of thought that has gone into the design and the search and recommendation engine. To simplify it further, the vast video library that exists within Netflix has been appropriately tagged and segmented under different genres. When any user on the platform streams a specific movie or television series and watches it in its entirety. The recommendation engine built into Netflix recommends similar programs or movies for the user to watch next.
Now would it not be wonderful to have a learning platform that engineers this logic to offer a ‘Netflix-like learning experience’? Remember this is different from just showcasing a tiled structure of programs with interesting thumbnails. AI and powerful algorithms could be used to create a recommendation engine that analyzes the learner information (designation, education qualification, skills, programs recommended by line manager, time taken to complete programs, etc.). Based on this data the LXP can be made to recommend the ideal program for learners. Though several learning products in the market claim to offer this feature, it would be interesting to examine if this really works with a large subset of data on learners, and lots of varied programs within the system.
AI should facilitate ‘smarter learning’ by examining all the data fed into the platform to provide a unique, personalized, and result-driven learning experience for the learner. With ‘adaptive learning’ we have already put this into practice and assess learners based on the correct or incorrect answers that they offer and make the next question in the series progressively easier or tougher. The learner should be able to derive a palpable benefit by completing a course recommended by the system.
Challenges in Using Artificial Intelligence to Create Content
Pricing of the AI-powered tool – Smaller organizations work on tight budgets and with skeletal staffing. To invest in a tool and then again dedicate a person to go through the content and then correct it as required, could prove expensive and cumbersome.
Resistance to change – A mindset that is difficult to change – the fear that Artificial Intelligence and robots will take away the jobs of humans is quite real. Across the world, especially in core engineering and manufacturing we are seeing technology evolving and cutting down actual human intervention. It will still take some time for a full-fledged Artificial Intelligence solution to analyze and create a fool-proof eLearning storyboard, which will not require any human intervention.
Empathy – The quality that separates a human from a robot – the ability to understand and share the feelings of another. Writing involves the content creator to think and feel like the reader / learner and put oneself into their shoes. No amount of programming or coding will help create a solution that will be able to feel the sentiments or care for the learner. This is the single biggest quality that helps humans excel at what they do. Bias in AI is another big challenge that tech-firms need to address.
Benefits & Implications of Using Artificial Intelligence to Create Content
Costs & Time – The primary benefits would be linked to cost and time. The AI-tool could be used to correct the errors in a basic storyboard or the tool could be used to create an outline or a mind-map that visualizes the program, which then the ID takes and develops. By investing in technology, even a small-sized eLearning vendor would be able to compete with larger players with more employees by creating cutting edge eLearning content primarily built with a suite of Artificial Intelligence and Machine Learning tools.
In a capitalist society, we look at profits. But we are humans too and believe in the innate goodness of people. The concept of ‘karma’ helps us stay grounded and create opportunities for people to learn and grow. The ever-lasting ambiguity and conflict between doing what is right and what is good impacts the deployment of AI as well. At some point, we would need to strike the right balance. As stated earlier an end-to-end development and handover of an eLearning project fully designed using Artificial Intelligence tools is still a dream or rather a work-in-progress.
What Does the Future Hold?
A significant amount of time and money is being spent by both governments and private organizations in developing AI-based tools. These tools cover a broad spectrum of work across industries. In eLearning – Artificial Intelligence and Machine Learning is helping graphic designers select and create better images. The tagging of images with relevant labels is a time-consuming process. An Artificial Intelligence-based solution like the Adobe Sensei framework can automate this task. We already use machine voice to test out the first version of the eLearning course to see how well it works. Solutions that curate and store voice clips in a digital vault that can be used for in appropriate courses can help eLearning designers to a great extent. At the rate at which technology is developing, there are strong possibilities of some big eLearning provider making a big-ticket announcement of their full-fledged eLearning program conceptualized and designed by an Artificial Intelligence tool.