With an unmatched passion, knowledge, and optimism related to advances in learntech, Donald Clark has over three decades of experience at the intersection of technology and learning. He is CEO of WildFire, an AI content creation company, and a professor, researcher, speaker, and learntech blogger (he’s written a series of posts on 100 learning theorists who have shaped the world of learning). He’s also author of the book Artificial Intelligence for Learning. (Listen to the audio at the end of the episode for a special discount offer on the book for Leading Learning Podcast listeners.)
In episode two of our seven-part series exploring the frontiers of learning technology, Celisa talks with Donald about the important role of artificial intelligence and data in shaping the learntech landscape. They also discuss related trends, what’s on the horizon, and key ideas learning businesses and society need to consider to ensure effective and positive applications of learntech, now and in the future.
To tune in, just click below. To make sure you catch all future episodes, be sure to subscribe via RSS, Apple Podcasts, Spotify, Stitcher Radio, iHeartRadio, PodBean, or any podcatcher service you may use (e.g., Overcast). And, if you like the podcast, be sure to give it a tweet.
Listen to the Show
Access the Transcript
Read the Show Notes
[00:24] – Intro and background info about Donald Clark
The Frontiers of Learning Technology
[03:18] – When you think of the phrase “frontiers of learning technology,” what comes to mind?
Donald shares that because he’s tended to always to work on the leading edge of technology, and, for him, “frontiers of learning technology” means two things: artificial intelligence and data.
Really everything we do online is mediated by AI and data.Donald Clark
The whole online world is supported and mediated by AI except in learning, which is why he’s focused on AI in learning for the past four to five years. Not only has he built and invested in companies that focus on AI in learning, but he wrote Artificial Intelligence for Learning, which lays out the landscape around this really radical shift in technology.
He highlights several big shifts in technology, the biggest one being writing, which shaped and created the culture of our species, along with the alphabet and then printing. He explains that these are all learning technologies and multipliers of culture. The computer came along in the 1970s, and then the Internet itself, another event some would argue was as big as writing. Now the big paradigm shift is to AI and data, which start to mimic teaching and learning in a much more sophisticated way.
Over-Hyped Trends in Learntech
[05:15] – When you think about trends, do any fall in the camp of the over-hyped or dead-ends, things distracting us from other technology that might actually help with learning?
Donald thinks there are certainly some things that are more niche than general. He likes to look at the consumer technology that influences learning technology universally, and AI and data fit the bill—as do writing, computers, and social media.
But there are some niche things that people claim to be universal, such as Google Glass. He says virtual reality (VR) is not a dead-end, but it is a niche product currently and we don’t quite know where it’s going to end up. He also thinks augmented reality (AR) is over-hyped. There are trends whose efficacy in learning are somewhat exaggerated.
Another trend he thinks is exaggerated is gamification. Donald says it could be useful, but only in a Pavlovian badging sense. Badges are another trend he thinks is over-hyped.
Trends with Significant Potential for Learning
[07:08] – Which trends have significant potential for learning?
Donald likes to think about that question by considering which technologies reflect good learning theories and how we really learn. Some technologies fit learning; some don’t. The ones that really fit—and this is why he thinks AI and data are so important—are the ones that shape the interfaces, make learning easier, and avoid cognitive overload. For example, voice interface AI (such as with Google Alexa) has given us text-to-speech and speech-to-text that makes these interfaces almost frictionless.
The brain has evolved to where we don’t really learn how to speak or understand what people are saying—it comes effortlessly. But we did have to learn how to read and write. Our big fingers on tiny touch screens are a poor way of interfacing with anything. AI should make learning smoother and easier. We want to get to the learning, not worry about the interface.
The algorithmic side of AI, fueled by data, is allowing us to personalize learning, which is very important. This is the big shift in pedagogy that he thinks is afforded by AI. AI can be sensitive to what learners are finding difficult at a precise moment and help.
We vector through learning differently. We’re all in a different learning journey, even when you’re sitting in a classroom with 30 people. You’re all sort of going at different speeds, different things going on in your head, different types of difficulties. So this personalization of learning is terribly important.Donald Clark
AI for Personalization of Learning
[09:14] – Where are we in the application of AI to personalize learning? Do you have examples of where AI is being used successfully to support learning interactions?
Donald points out that we tend to think about AI as present and future tense, but it’s already been around for a couple of decades. For example, we all use Google, and Google is pure AI. It gives us what we’re interested in at the exact moment of need. He doesn’t just mean Google Search; he means the searchability of the content. Nobody doing PhD research would want to trade Google Scholar to return to wandering down shelves, looking for bits of paper in journals. AI makes research more efficient.
AI is already touching adaptive learning, which is increasingly being adopted.
There’s also the recommendation engine side of AI, which tries to determine where learners are on their learning journey in order to suggest appropriate content, as well as provide learning support through chatbots or the like. All this is enabled by AI. AI is also being used for anti-plagiarism, and there’s even the creation of online learning content using AI (such as what WildFire does).
I think we’re at this embryonic stage where the sophistication of AI is bringing sophistication to technology-based learning to get us out of a really, what was a necessary stage, which was almost multimedia production. Lots of videos, thinly punctuated by multiple-choice questions…. What it needs to do is become more like a teacher. Be more sensitive to you as a learner, be more adaptive, more responsive,to be able to tackle your needs at exactly that moment. And you can only do that by using data. This is terribly important. I need to know about you as an individual, where you are exactly, how you got there, the context you’re in, butI also need aggregated data from all the other learners that are taking this course so that I can bring that to bear,right at that moment of need.Donald Clark
The Near Future of Learntech
[12:40] – How would you characterize learntech in the near future? Are we going to have breakthroughs? Is it going to be disruptive innovation, more incremental innovation, something else? How would you describe what you think is going to come in the next three years or so?
Donald predicts in the next three years we’ll have a hybrid phase, as we move from one paradigm to another. Netflix gives you an AI-mediated, tiled, personalized menu. Nobody has to learn the Netflix menu—it scrolls to the right across a topic and in-depth for new topics. It’s almost an effortless interface. But Netflix exists alongside broadcast television, which isn’t disappearing anytime soon (although it will diminish as the new streaming service comes into mode).
The best contrast might be between traditional VLEs (virtual learning environments) or LMSes (learning management systems), which will continue for some time because they manage and store stuff (though rather crudely) with the SCORM standard. But he thinks they are going to be replaced to a very large degree—and quite quickly—by LXPs (learning experience platforms). LXPs will enable AI- and data-driven learning journeys that are much more sophisticated, respond to the learner, and deliver personalized learning.
The More Distant Future of Learntech
[14:11] – In the more distant future, what do you see as the direction for learntech?
The shift is starting now. For example, there are now $3 million deals on LXPs, and that wouldn’t have happened just a few years ago, but it’s now happening with the larger global companies.
He admits that looking further out is always dangerous because the further you go out the probability drops off dramatically of you being correct. But he says the AI- and data-driven approach is irreversible. It’s is the technology of the age, and the idea that we’ll go back to a client-server model is plainly ridiculous.
There are some really interesting developments that he thinks don’t get enough attention when thinking five to ten years out. First, and this may be within the three-year horizon, is Starlink. The underlying technology for online delivery, especially AI and data, is bandwidth, is the Internet itself. The promise here (likely within a year or two) is that high bandwidth 5G will be available anywhere on planet Earth at a reasonable price because the prices will fall with volume. Donald thinks that will be revolutionary in our ability to deliver technology-based learning to anyone, anywhere, anytime.
Donald notes the great thing about having high bandwidth anywhere is it allows the use of personalization and AI and data in a way previously impossible because streaming is data-hungry. Also, the use of AI needs high bandwidth because you don’t want latency; the experience should be smooth.
Another area further out that Donald finds fascinating is the work by Neuralink and others. We have to be careful with these invasive and noninvasive techniques around the brain. He’s highly suspicious of the value of the noninvasive techniques, where you put a little helmet on and have a light flashing on your forehead, because what it’s measuring is EEG (electroencephalogram), which is a messy signal. He finds claims that we’ll be able to tell whether learners are paying attention or not farfetched.
However, the invasive techniques are starting to get interesting. We already have the Utah Array, and there are 150,000 people with that in their heads, helping them to move their arms and legs.
See below for an example from Neuralink, which shows a monkey playing a simple video game after getting implants of the new technology.
Donald says we know that putting fibers in the brain to stimulate portions of the brain works, but the interesting possibility is the ability to read and write to the brain. When you’ve got tiny fibers—a fraction of the width of a human hair—and you can put them in without bleeding, then you can start reading data, and you need AI to interpret that data accurately.
Writing data back to the brain begins to get into the realm of science fiction because we’re nowhere near this practically, but imagine sometime in the future if where we could pay $50 to learn Spanish.
But he points out there’s a bigger prize at stake here, which Neuralink is focused on: solving mental illness problems. For example, using technology to get rid of depression. So this isn’t just about toys, gimmicks, or gadgets. Rather, there are some real possibilities the technologies will afford us.
[20:53] – If you’re looking for a partner to help you realize the possibilities of learning technology, check out our sponsor for this series.
BenchPrep is a pioneer in the modern learning space, digitally transforming professional learning for corporations, credentialing bodies, associations, and training companies for over a decade. With an award-winning, learner-centric, cloud-based platform, BenchPrep enables learning organizations to deliver the best digital experience to drive learning outcomes and increase revenue.
The platform’s omni channel delivery incorporates personalized learning pathways, robust instructional design principles, gamification, and near real-time analytics that allow organizations across all industries to achieve their goals. More than 6 million learners have used BenchPrep’s platform to attain academic and professional success. BenchPrep publishes regular content sharing the latest in e-learning trends.
To download BenchPrep’s latest e-books, case studies, white papers, and more go to www.benchprep.com/resources.
Getting Learntech Right
[22:03] – If we get learning technology right as society, what’s the good that we might see coming from learntech? And what are some of the things that we need to do to make sure that we get it right?
There are people, especially in academia who say they’re involved in AI and ethics, but it’s not really ethics, according to Donald. Ethics is the study of moral principles, good and bad. But what they’re doing, he says, is more like activism. They’re looking for bias, gender issues, or racism around AI. They’re looking for the flaws, and that’s not really ethics because they’re discounting the other side, the good side. Ethics takes a more balanced view of what’s good and bad.
Donald thinks there are big moral issues here. Higher education in particular produces massive inequalities, and it’s become a generator of inequalities. The future technology-driven world offers a chance to balance things out and get education to everyone cheaply.
We cannot go on charging people tens of thousands of dollars a day for a little certificate at the end that’s really a sorting mechanism. Of course, it has value. Of course, people learn things at college, but about 80 percent of it is signaling. In other words, it’s giving employers a signal that you stuck at it a little bit, that you come from a certain class and background, and so on. I think that’s unacceptable now because it’s led to such inequalities and such a fractious world that I think it’s no longer sustainable. So I think we have to make learning cheaper, faster, better, more accessible globally.Donald Clark
Donald thinks globally is an important word if we’re going to solve climate change and get around the economic issues that we face today. The current system doesn’t work—it’s far too expensive, clumsy, and slow. We have to begin to rebalance system.
Donald doesn’t believe lifelong learning has anything to do with going back to college. Hardly any adult who went through college wants to go back, and hardly any of them do because that’s not what lifelong learning is about. Lifelong learning is about become an autonomous learner, being curious, and learning on your own.
Getting Learntech Wrong
[26:57] – What are the dangers we’re up against if we don’t approach learning technology in the right way? Any thoughts around what actions might lead us to get learntech wrong?
Donald thinks we’re already getting it wrong, particularly when it comes to cost, with $1.8 trillion in federal debt around student loans and people are struggling to pay that money back because they’re un- or underemployed. Academics fly everywhere at the drop of a hat to to to a conference, and, if we think that’s acceptable in terms of climate change, then we have to think again.
He says we’re now in sort of disaster avoidance mode, with economic issues and climate change. Technology promises a greener world. We went through an amazing experiment during COVID, where almost everybody on the planet had to do online learning (and he doesn’t buy the idea that kids were traumatized by it).
Donald suggests we need to look forward to a blend between the online and offline, with learning but also with work. We may spend two to four days a week working at home with the days in an office—or maybe wholly at home. Everything will be blended in the future. We’re now doing blended eating (with food getting delivered to our doors in COVID times) and blended entertainment (watching more movies than ever through Netflix).
Donald thinks the technology has shown we can help solve the climate change issue. And technological innovation is ultimately perhaps our only hope with climate change, and that’s a result of education and learning.
Donald is optimist. He thinks technology is a force for good, and that’s why he gets disappointed when he sees armies of people on social media looking for little bits of bias in algorithms when we know human beings are packed full of bias. Donald says we’re all racist and sexist, and it’s very difficult to get rid of that bias. But at least we can work towards improving these other systems to reduce the levels of bias. The danger is that we throw the baby out with the bathwater—and the bath. We throw the whole lot out because we think there’s some bias or unfairness in an algorithmic system that actually exists in the human system anyway.
See our related episodes that deal with bias: Diversity and Disruption with Shilpa Alimchandani and Getting Conscious About Bias with Howard Ross and Shilpa Alimchandani.
Donald thinks China perhaps will leap ahead because they don’t have the same view of these issues. He says Europe tends to be stuck in the middle here, with less innovation and loads of regulation. Americans have innovation on their side with some of the great tech innovations coming from the US (and he thinks this will continue to be the case).
Technology to Address Learning-Related Problems and Opportunities
[31:33] – Are there learning-related problems or opportunities that we need technology to address? Are there things we can’t solve without learning technology or technology more broadly?
Almost everyone with learning difficulties has been helped by AI and data, and Donald thinks we have a chance of tackling the difficulties every individual has in learning. This has already happened in the world of accessibility. For people with hearing and visual impairment, it’s been AI that’s led the way with text-to-speech and speech-to-text systems.
Mathematics and language learning are areas of catastrophic failure for most people. We can use technology to help solve failure and get people through obstacles.
I think once we pay attention to the learning theory and match that with the technology we can massively accelerate learning for the good of all, for the good of everyone.Donald Clark
Advice for Learning Businesses
[34:05] – What advice do you have for a learning business looking to effectively use learning technology and trying to decide what to focus on, what to invest in, and where to put their resources?
Donald recommends learning businesses look at AI and data with more vigor. Remember that learning is a process, not an event. Nobody learns anything by just sitting in a lecture or a one-off course, online or offline. Almost all learning is asynchronous and after the event. We know how people learn, and, more importantly, we know a lot about how people forget and fail. If we start addressing those issues using the technology, we will accelerate learning and have a much higher degrees of success.
Donald only invests in AI- and data-based learning companies now because that’s the future. We live in the age of algorithms, AI, and data. And if your learning business is not thinking about using those, then another organization will come along that is.
AI- and data-driven companies are growing, with the biggest ones being either American or Chinese. Google, Facebook, Microsoft, and Apple are all really AI companies. If you ask any of the CEOs of those companies what the underlying technology is that they use and develop, they would say AI. Donald says these companies are now starting to dabble in education and learning.
He also adds that we’re looking at hybrid AI-with-human models coming along. That’s happening already in adaptive learning, in learning experience platforms (LXPs).
Consumer Technology with the Most Potential for Learning
[37:25] – What consumer technology do you see having the most potential for learning?
Donald clarifies that when he refers to AI, it’s not one thing—it is many, many things. Text-to-speech/speech-to-text is one species of AI that relies on natural language processing (NLP).
Then there are the big models, such as GPT-3, that generate online learning content automatically. He’s built a system that does that. You just send him a video, a PowerPoint, text, etc., and he cuts and pastes it into the system that then generates online learning content, not multiple-choice questions, but questions that can be interrogative. So it will ask a learner a question, she types a short paragraph, and AI interpret that paragraph semantically and meaningfully. These little bits of AI are moving ahead on a broad front and allowing us to do things we could never do with technology before.
Donald thinks AI will have big cultural and economic impact because it’s a massive multiplier. You can scale it, and it extends cognition in a way that other technology doesn’t. For example, video is a very poor medium for learning (unless the goal is emotional impact or to cause an attitudinal shift). AI is replacing some teaching tasks, and that’s why it’s more important. It can scale, and it allows us to enable pedagogies like spaced practice, interleaving, all the things we know will work, but never had the chance to apply.
AI comes from consumer technology. If you buy something on Amazon or go on Netflix, the reason they have these personalized screens is they’re trying to target what you want. That’s what we need in learning. AI melds hardware and software together in a very powerful fashion. Donald thinks there’s a huge amount of good to come from this in the learning and education sphere.
Donald points out that we don’t regard things like writing, printing, or books as technology, but they were in their day. What we tend to do is historically declassify or discount the technology from before we were born. We don’t see a dishwasher as a piece of technology now, but it was. He predicts we’ll see the current e-learning, LMSes, MOOCs, and so on as old hat very soon because of this newer age of algorithms, data, and AI.
AI As Comprehension Without Competence
[41:05] – I’ve heard you describe AI as comprehension without competence. Would you talk a little bit about what the implications of that might mean when applying AI in the service of learning?
When describing AI to people in the education and learning profession Donald stresses the importance of truly understanding what it is. He describes AI as an idiot savant. It’s incredibly smart at very precise things but incredibly stupid at general intelligence.
But AI doesn’t have to be a form of human intelligence to be effective. It’s terribly important in terms of the application of AI that we don’t over anthropomorphize.
The future of AI is not going to be little robot teachers. It’s not going to be robot teachers, full stop. It’s such a great fallacy this. It’s going to be more like Google. It’s going to be online. It’s going to be from Starlink. It’s going to be on whatever devices we carry around with us, coming into our glasses, AR, whatever. It’s going to be online stuff. Incredibly smart because of the hidden force of AI behind the scenes. It will be invisible, and that’s the point.Donald Clark
AI shouldn’t replace the human brain because the human brain is a terrible, messy organ. It’s inattentive and gets easily distracted, emotional, and depressed. It forgets almost everything, gets dementia, Alzheimer’s, etc. We can’t network from other brains or upload or download. Computers can do a lot of that stuff, so we can take a load off ourselves by using this technology.
AI has a long way to go before being human, but it doesn’t have to be human to be effective.
A Shift to LXPs
[43:49] – LXPS are getting a lot of attention these days, the Netflix-like learning being very attractive. What are your thoughts on the differences that LXPs represent versus other learning technologies, like LMSes? What’s driving the shift?
Donald thinks what’s driving the shift is the idea that you spend a lot of money just storing and managing stuff on an LMS, which is really mimicking the old course structure. Learning is a process, not an event. With LXPs, we flip that model and look at how people actually learn. We learn in a fragmented and informal fashion. He references the 1991 book Electronic Performance Support Systems by Gloria Gery, which talks about unintentional learning and how most of learning just happens to us.
He says LXPs are more dynamic. They use AI for recommending and predicting the future and automating processes. They push stuff to learners, but also allow learners to pull stuff.
There are other species of push that are more dynamic and process-like. For example, Donald talks about spaced practice, which works superbly well. But how many of us use it? Almost nobody. But now we have the technology where the algorithms can personalize spaced practice, just as Duolingo does with a language. Duolingo is worth $1.5 billion and has 100 million learners in their sy,stem and that’s because it uses this type of tech. It’s adaptive. It’s an LXP-type system that will send notifications to you. If you go away for the weekend and stop learning Spanish, it knows you’re going to forget some so it takes, you back a little bit. Donald asserts this type of smart technology will shape the future of learning—it already is.
LXPs are data-driven, often using a learning record store (LRS) to fuel the LXP.
When we look into the future, we need to have an ethical concern and focus then on what really matters—and that’s people and their needs in terms of accessibility and cost. Donald points out that everything has become too expensive, making it far too elitist. We need to make sure that people can afford learning, that it’s available, most likely for free.
All the learning he does is free—for example, listening to podcasts. Some 60 percent of all Americans listen regularly to podcasts, and Donald says nobody could have predicted that, which is what’s so exciting.
This interview is taking place on different parts of the planet, and then we publish it, and anybody can hear it for free. When Starlink comes along, anybody, literally anywhere on planet Earth, can listen to that podcast for free. Donald notes that we’ve gotten pessimistic about technology; we need to have more optimism.
[49:35] – Wrap-up
Donald Clark is CEO of WildFire, an AI content creation company, and author of Artificial Intelligence for Learning, which is recommended reading if you want to better understand how AI can support learning. We’re pleased to be able to offer Leading Learning Podcast listeners a 20-percent discount with free delivery in the US. (Listen to the audio to get the discount code.)
To make sure you don’t miss the new episodes, we encourage you to subscribe via RSS, Apple Podcasts, Spotify, Stitcher Radio, iHeartRadio, PodBean, or any podcatcher service you may use (e.g., Overcast). Subscribing also helps us get some data on the impact of the podcast. And remember how important data is!
We’d also appreciate if you give us a rating on Apple Podcasts by going to https://www.leadinglearning.com/apple. We personally appreciate your rating and review, but more importantly reviews and ratings play a big role in helping the podcast show up when people search for content on leading a learning business.
We encourage you to learn more about the sponsor for this series by visiting benchprep.com/resources.
[51:52] – Sign-off
Other Episodes in This Series:
Episodes on Related Topics: