Capacity deals with the people and the technology a learning business has in place, and capacity has a quantitative and a qualitative aspect. How well can the people and the technology do the work, and how much can the people and the technology do?
Artificial intelligence has the potential to change both the quality and the quantity of work that a learning business does, and so, in this episode of the Leading Learning Podcast, number 410, co-hosts Jeff Cobb and Celisa Steele focus on barriers to the adoption of AI and the potential accelerators.
To tune in, listen 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
Download a PDF transcript of this episode’s audio.
Read the Show Notes
Celisa Steele: [00:00:00] Artificial intelligence has the potential to increase both the quality and the quantity of work that a learning business does, and there are a variety of factors that might speed up or slow down a learning business’s adoption of AI.
Celisa Steele: [00:00:19] I’m Celisa Steele.
Jeff Cobb: [00:00:20] I’m Jeff Cobb, and this is the Leading Learning Podcast.
Jeff Cobb: [00:00:28] Capacity is one of the five domains in our Learning Business Maturity Model. Capacity deals with the people and the technology a learning business has in place—the capabilities and the competence staff, contractors, and consultants bring to the table, along with the abilities of relevant software, systems, and processes.
Celisa Steele: [00:00:48] There’s a quantitative and a qualitative aspect to capacity. How well can the people and the technology do the work? That’s the qualitative side. How much can the people and the technology do? That’s the quantitative side of capacity.
Jeff Cobb: [00:01:04] Artificial intelligence, in our view, has the potential to change both the quality and the quantity of work that a learning business does, and so we want to focus on barriers to the adoption of AI and the potential accelerators of AI adoption in this episode, number 410.
Growing Expectation that Individuals Use AI to Do Their Work
Celisa Steele: [00:01:24] There’s going to be a growing expectation that individuals, whether those are employees or contractors that are serving a learning business, make use of artificial intelligence in the work that they do.
Jeff Cobb: [00:01:37] Yes, I think that’s one of the things that does seem to be clear about how things are evolving around AI. There’s enough AI woven into enough places right now that seeing it mainly as a productivity tool bubbling up within learning businesses—using Microsoft Copilot or Zoom’s AI Companion—that’s naturally going to start happening, already is happening, regardless of whether organizations are really focused on AI and being strategic about it and how it fits into the future of their learning business.
Celisa Steele: [00:02:12] You mentioned productivity, and productivity emphasizes the quantity side. It’s this idea that you can have the same people on the team, but they can get more done by making use of some of these AI tools. Again, it’s more on the quantity side than necessarily on the quality side. The quality is a little bit of a deeper process and is probably likely to take longer for learning businesses to figure out what it looks like to deeply incorporate artificial intelligence into organizational-wide operations, to make use of it to create and deliver learning products. All of that seems a little bit more fraught in terms of how it plays out and how it ties into strategic goals.
Moving Toward More Strategic Uses of AI
Jeff Cobb: [00:02:58] Yes, it’s not going to happen as organically and by default the way that having Copilot in your Microsoft Office Suite is going to happen. It is going to require a lot more attention, and it’s going to require, in most organizations, breaking through some barriers that are going to be there to slow down adoption by learning businesses. And so that’s where we want to focus first in this episode, around those barriers that learning businesses may be facing. We will say, in our experience so far, people are talking about AI. They’re interested in it. Some of this—using Copilot, using Zoom AI Companion—is going on. But we’re not seeing a whole lot of aggressive, deep adoption of AI at this point, and I think it’s because of some of the barriers we’ll discuss here.
6 Barriers to AI Adoption in Learning Businesses
Celisa Steele: [00:03:44] We have six barriers in mind. As we run through these, we encourage you to listen with an ear towards which of these really do seem to be barriers in your organization, in your learning business? Of course, also think beyond the barriers that we are going to mention. What other barrier is your learning business facing in terms of adoption? That’s just a note in advance—a way that we encourage you to be listening.
1. Cost
Celisa Steele: [00:04:13] The first barrier that we’ll mention is a barrier that shows up in all sorts of areas whenever you talk about making change, and it is simply cost.
Jeff Cobb: [00:04:22] That’s right. To really integrate AI into your operations, your strategy is going to have some costs associated with the software and with everything that’s going to have to go with the software to use it well. You’ve got the free version of ChatGPT out there that folks can use. It isn’t the best or the most advanced version. You have to pay for access to get that. So, while you can use free AI in some cases, often there is that need to pay, especially when you start to think about that quality side versus the quantity.
Celisa Steele: [00:04:55] Yes, and to just make a point about the availability of what’s free versus what you have to pay for, there’s a podcast episode that Ezra Klein did with Ethan Mollick. We’ll be sure to link to that in the show notes for this episode. But Ethan Mollick makes the point that when ChatGPT first came out, it was about as good as a sixth grader at writing. The free version, which is ChatGPT-3.5 at this point, is as good as a high schooler or a maybe college freshman. And then GPT-4, which is what you have to pay to get access to, is as good as a PhD in at least some forms of writing. Right there you can already see, if you’re really interested in quality, that cost investment almost becomes a necessity.
Jeff Cobb: [00:05:43] Yes, and that’s relatively small to get a ChatGPT subscription, even for your whole team. But, again, as you start to expand and get into other tools and the capacity to use them, the costs start to proliferate pretty quickly, and you’ve got to reckon with those. Those are the new costs. But I think what may even be a bigger issue for many learning businesses is those sunk costs that they already have in their existing technology and learntech investments.
Celisa Steele: [00:06:11] Right. A lot of learning businesses have a learning management system that they invested heavily in, not only in terms of the software costs but, again, the people, time to figure out how…
Jeff Cobb: [00:06:23] And a lot of pain.
Celisa Steele: [00:06:23] A lot of pain in how to integrate it, how to make it work with your processes, how to get everyone trained on it. You have the cost of legacy systems plus all the surrounding processes and education that you’ve already spent, essentially, money on for your team. It’s going to be a little bit hard, probably, to divert funds away from some of those legacy systems until there’s a crystal-clear, pretty fast path to the return on investment. We’re seeing that as a barrier.
2. Perceived Lack of Demand from Learners
Jeff Cobb: [00:06:56] So that’s cost—both the new cost and dealing with those sunk costs that you already have. Number two is perceived lack of demand from learners. This might take a little explaining, but you hear about these AI-driven learner scenarios that are all about personalization and dynamic and everything else. They’re predicated on this notion of what could be characterized as the “ideal learner”—someone who really appreciates and proactively seeks out learning experiences that are personalized, that do dynamically adapt to their needs.
Celisa Steele: [00:07:31] And not to be overly cynical, but to be realistic…
Jeff Cobb: [00:07:35] Realistically cynical.
Celisa Steele: [00:07:36] To be realistically cynical—that’s good. A lot of learners simply aren’t that ideal learner. They have a need. They’re trying to get their job done, and they need some piece of knowledge or a new skill to do it, so they’re really after that, and maybe they don’t care so much how personalized or dynamic the experience is. Or maybe they just have to check the box if they have some sort of CE requirement. Again, then in that case, they really just want to check that box and move on.
Jeff Cobb: [00:08:05] Yes, I think a lot of organizations, individuals who are creating learning programs like the idea of serving that ideal learner, but I think they know intuitively or feel like they know intuitively that maybe we’re not there yet. Maybe this isn’t the time to be aiming that high with what we’re doing. Now, that said—and this may already point towards the accelerator camp that we’re going to talk about in a little bit—the growth and enthusiasm for platforms like, say, Duolingo, which seems to have a pretty strong fan base…
Celisa Steele: [00:08:31] I’m one of them.
Jeff Cobb: [00:08:30] You’re one of them, and I’ve used it too. In fact, it keeps bugging me to come back. The enthusiasm around those types of platforms suggests that there might be more latent demand than many learning businesses appreciate at this point.
3. Lack of Time
Celisa Steele: [00:08:45] We’ve talked about cost as a potential barrier. We’ve talked about perceived lack of demand from learners as a potential barrier. A third barrier is lack of time. It’s very easy to say that learning businesses should make more time to explore artificial intelligence, to poke around and understand the possibilities, run some pilots, do some experiments. But I have met very few learning business professionals who feel like they have time to do that sort of piloting and experimentation and poking without a really clear return on investment, given all of the other things on their plate.
Jeff Cobb: [00:09:25] Yes, usually plenty of other things to do for the individual, for the team, and saying that you’re going to make this a priority right now can just be a stretch in a lot of instances. So lack of time. And then, number four, a lack of understanding.
4. Lack of Understanding
Celisa Steele: [00:09:40] Yes, there can be a skills gap. Even if you manage to carve out time or have people on your team who do have some time to devote to it, there may be a skills gap. Those people may not really understand artificial intelligence capabilities and the mechanics behind them, and then that can lead to a barrier to adoption because it can be, “Fine, play with this,” but, if you don’t really understand how the systems work and what they might do for you, that open-ended experimentation might not result in much coming of it that’s actually useful.
Jeff Cobb: [00:10:16] That’s true. It really is easy to get into ChatGPT-4—if you pay for it—and get it to do some things, but then really understanding how you get it to do the things that you most want it to do in a productive way, there is a learning curve on that. And, yes, it goes back to that time issue. The time to get the understanding often isn’t there.
Celisa Steele: [00:10:36] It also relates to an absence of a strategic or even a tactical framing for AI exploration. If the directive that is passed down in a learning business is “Experiment with AI,” that leaves so much open to that individual who’s playing with it, and you’re going to get widely different results depending on how good that individual’s understanding of AI and its possibilities are.
Jeff Cobb: [00:11:03] Yes, classic problem with innovation in general. There’s always that imperative, “Go innovate!” What does that mean? We haven’t actually developed a strategy and a plan around doing that. So that’s lack of understanding. Then, number five, content concerns.
5. Content Concerns
Celisa Steele: [00:11:18] This is an umbrella for us. There are a number of things that we’re categorizing as content concerns. Think about accuracy and hallucinations, for example. Obviously, that’s a very big deal in the context of learning businesses, where often the content absolutely has to be reliable and factual.
Jeff Cobb: [00:11:39] Yes, if you’re in healthcare (for example, medical), you can’t be putting stuff out there that is not completely reliable and accurate. Valid concerns around that. That does have to be addressed. There’re concerns around data privacy and security, particularly around sensitive learner data that might be coming in. A lot of data churning through AI, and exactly what’s happening with all that data isn’t always clear.
Celisa Steele: [00:12:05] Which leads to, then, ethical concerns around what data do you have the right to share with an AI tool that your learners have given to you? And tied up in those ethical concerns, there’s also concern around bias.
Jeff Cobb: [00:12:21] Yes, this is the kind of thing that’s always making headlines, I think, around AI. But the bottom line is even the developers don’t know exactly how it all works. There’s a bit of a black box in these large language models and other types of AI. The data used to train these large language models has the chance to perpetuate stereotypes, exclude or disadvantage certain groups of learners, depending on what goes into that training, and perpetuating equity issues and all sorts of other issues that we were already wrestling with before AI—we don’t need to have exacerbated by something that could take it to an even bigger scale than it is now.
Celisa Steele: [00:12:58] A lot of barriers wrapped up under that umbrella of content concerns. And then the sixth barrier that we’ll mention—this is the final one that we have brought to the table today to discuss—is culture.
6. Culture
Jeff Cobb: [00:13:12] Culture is always a big one there. Resistance to change. There’s a lot of comfort with traditional models that can slow down AI adoption. This happens with any new technology, any new approach coming into any sort of business. And, certainly, there are many areas of “We’ve always done it this way.” We’ve seen even resistance to advances that are driven by learning science within the learning field. You throw in something that feels a little less certain or harder to understand, like AI, and, yes, that resistance is bound to be there in many organizations.
Celisa Steele: [00:13:48] Yes, there’s a certain amount of inertia that can prevent the adoption. Again, like you were saying, we’ve always done things this way, and they’re working fine. Why do we need to change? I think silos, which can also come from culture, can further exacerbate adoption of AI because, even if one person or one team within a learning business or an overall organization figures out an effective use of artificial intelligence, will others in the organization even know about it? And, if they know about it, will they be encouraged to adopt it or apply it in their area? So I think silos are another aspect of culture that can really be a barrier to adoption of AI.
Jeff Cobb: [00:14:32] To loop back and note all of those again. The barriers we’re seeing are around cost, both new costs and sunk costs. Around a perceived lack of demand for learners—maybe it’s just not time yet to do this. A lack of time on the part of the learning business and its staff—you just don’t have the time to do this. A lack of understanding—this relates to the lack of time potentially but knowing enough to be able to really get the full potential out of AI. The fifth one, content concerns around IP issues, privacy, ethical considerations, bias. And then, finally, number six—the one that’s always there in any big shift in technology, strategy, or anything else—is culture.
Partner with Tagoras
Jeff Cobb: [00:15:23] At Tagoras, we partner with professional and trade associations, continuing education units, training firms, and other learning businesses to help them to understand market realities and potential, to connect better with existing customers and find new ones, and to make smart investment decisions around product development and portfolio management. Drawing on our expertise in lifelong learning, market assessment, and strategy formulation, we can help you achieve greater reach, revenue, and impact. Learn more at tagoras.com/more.
5 Accelerators of AI Adoption in Learning Businesses
Celisa Steele: [00:15:59] We’ve looked at barriers to AI adoption. We’d like to talk next about accelerators. What might speed the adoption of artificial intelligence in learning businesses? We have five that we’ll bring to the table. Again, we encourage you to listen to these five and think, “Okay, which ones of these could actually accelerate the adoption of AI in my organization?” And also think about what else might accelerate the adoption of AI in your organization.
1. Vendor Leadership
Jeff Cobb: [00:16:32] The first one we’re going to list is vendor leadership. This may particularly apply in the case of those legacy systems where you’ve got sunk costs and what the vendors are going to do there. But, overall, vendor leadership is going to be important here. If vendors, technology platform providers start integrating AI into those existing systems or investing heavily to smooth the path into new systems, that can make a big difference.
Celisa Steele: [00:16:59] Yes, and, so far, what we’ve seen in terms of efforts to incorporate AI into existing systems, it seems relatively rudimentary, more on that productivity side of things and the quantity rather than the quality side. Incorporating GPT to help generate content ideas or do a first draft of assessment questions, for example. And that’s useful, but it doesn’t really alter much about how things are being done.
Jeff Cobb: [00:17:30] Right. Hopefully, it’s hard to remember…. In some ways, it’s late days, but, in other ways, it’s early days because the whole ChatGPT shift that really marked a turning point didn’t happen all that long ago, so you could argue we’re still in somewhat early days, and vendors are starting to catch up and accelerate their efforts. We probably will see a lot more vendor leadership coming in in the next one to five years. But that would be a big factor to speed adoption. The next one would have to do with the broader market.
2. The Broader Market
Celisa Steele: [00:18:01] Again, this is an umbrella point for us. But you can imagine, for example, if a learning business is being outpaced by key competitors, that’s going to really urge them to get their act together and invest more in artificial intelligence. If more explicit demand starts coming from learners around artificial intelligence being used in the products and services that you’re offering, again, that sort of market demand is going to really push a learning business to move faster to integrate AI.
Jeff Cobb: [00:18:36] Generally, a fear of diminished reputation or relevance that, if you’re not keeping up with whoever the Joneses are in your competitive set, it just doesn’t reflect well on you. Those sorts of market pressures can come to bear. We also think about the partnerships and collaborations that can occur out there as being part of this broader market perspective.
Celisa Steele: [00:19:00] Yes, with the vendor leadership piece, where a learning business can take advantage of what a vendor of a learning platform might be making available. Another way to potentially leapfrog or accelerate your own adoption of AI is by partnering or collaborating with someone who already is further along that AI path than you are. You seek out new relationships—might be in the academic field, for example—and you can share knowledge, resources, and use cases, and maybe that speeds the adoption and application of AI.
Jeff Cobb: [00:19:37] Vendor leadership, number one. Number two, changes in that broader market and environment that are going to impact how you’re thinking about AI. And then number three would be another major shift in AI itself.
3. Another Major Shift in AI
Celisa Steele: [00:19:53] Right. You mentioned that, in some ways, these are early days, some ways maybe late days. But, with the introduction of ChatGPT, that really got the world’s attention because it allowed many people to see, firsthand, what the potential for this generative AI was. If there were to be some other sort of shift like that or some sort of event, that could speed adoption even further.
Jeff Cobb: [00:20:21] Right. One that’s out there potentially right now is the broader release of Sora, which is basically the video version of ChatGPT, using generative AI for video creation. It’s in beta or limited distribution right now. I have not been able to get my hands on it so far, but I’ve heard fairly incredible things about it. If you suddenly have this tool that allows you to crank out relevant, high-quality video very easily, that could be an inflection point. We don’t know. We’ll see exactly how much impact that has. But that’s just one potential example. ChatGPT was in many ways that proverbial tip of the iceberg because so much had been going on underneath that for so many years, and that’s what became visible and caught the eye of the general public and suddenly created this big shift. Something else is going to bubble up, no doubt, within the next one, two, or three years.
4. Regulation and Guidelines
Celisa Steele: [00:21:15] The fourth accelerator, or potential accelerator, that we see around adoption of AI is regulation and guidelines. This would help in terms of clarifying a lot. Like we were saying earlier around some of the barriers, there’s just a lot of lack of understanding about what AI can actually do and how it works. As regulations are rolled out, as clear guidelines are developed, that’s going to help build trust in artificial intelligence, and it’s going to make it clearer how to deal with some of the barriers that we talked about, things like intellectual property and what can or can’t you own, around bias, around ethical issues. If you have regulatory bodies developing those guidelines and overseeing things, then that takes the burden off of the individual learning business to have to work through legal opinions with a lawyer on their own, and then they’re allowed to think about, “Okay, within these rules, within these guidelines, how can we make use of it?”
Jeff Cobb: [00:22:21] Yes, this is almost certain to happen on a global stage too, which gives me a little more confidence that it will happen because, if it had to happen in the United States right now, given what the political environment is, I’m not so sure. But just like GDPR had such a big influence on how we think about Web privacy and security, regulations and guidelines are going to be coming from all over the world, and, hopefully, they will help to address some of those barriers that are giving organizations caution right now.
5. A Business Case for AI
Celisa Steele: [00:22:51] And then the fifth and final potential accelerator that we wanted to talk about is simply a good business case for AI.
Jeff Cobb: [00:23:01] Always good to have a good business case for things. This, of course, will go back to things like cost and understanding learner demand and all of those barriers that we spoke to earlier. A good business case is going to address and figure out how you work your way through those. It’s going to address our classic triumvirate of reach, revenue, and impact.
Celisa Steele: [00:23:25] The business case can be broken down into how is AI going to help you with your reach? How is AI going to help you with your revenue? How is AI going to help you with your impact? On the reach side, AI potentially helps you to do more with the same resources, which could translate very naturally into the potential to reach and serve more learners, even without greatly changing the size of your team, for example.
Jeff Cobb: [00:23:51] On that revenue side, if you’re able to do the analysis and see that potential for long-term savings despite upfront costs or new revenue generation because it allows you to develop products more quickly, get them to market more quickly—but just getting to that point where you can quantify, you can put the financial framework around it, that certainly could help with speeding the investment.
Celisa Steele: [00:24:16] Around impact, AI offers the potential to unlock some of those holy grails of learning that we’ve been talking about for a long time—decades at this point—things like personalization, things like just-in-time or point-of-need learning. If you can do that, that’s going to allow you to better serve those learners. It’s going to allow you to have more impact as a learning business. If you can begin to see that, that can be part of that good business case for investing and adopting AI.
Jeff Cobb: [00:24:48] Yes, and the same technology that’s going to make those things possible is also going to make it possible to get the data from them and know how effective they are, which goes back to regulation and all of the content concerns and everything else, but the potential for being able to serve the individual learner well, the organization well, the broader communities in which the individuals and the organizations work, and tracking the data around whether it’s actually having an impact or not. We’ve never been at a point before where that was as possible as it’s starting to be now.
Celisa Steele: [00:25:23] A good, strong business case for AI that helps you quantify and qualify the potential for impact on reach, revenue, impact can help accelerate adoption. The fourth point was around regulation and guidelines; as those are rolled out, that can help with speeding the adoption of AI. Third, we had mentioned another shift in artificial intelligence, some sort of development akin or similar to the release of ChatGPT. The second thing we had talked about in terms of an accelerator was broader market changes; as competitors adopt AI or as learners begin to clamor for AI or other partners and players becoming more attuned to AI, that’s going to push learning businesses to also pay attention. And then we started our list of potential accelerators with vendor leadership; if you have tech companies, for example, that invest in adding AI to their existing systems or developing new systems that then you can relatively seamlessly transition into, that will help with adoption.
AI Can Impact the Quality and Quantity of Work a Learning Business Does
Celisa Steele: [00:26:42] Artificial intelligence has the potential to increase both the quality and the quantity of work that a learning business does, and there are a variety of factors that might speed up or slow down a learning business’s adoption of AI.
To make sure you don’t miss 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 gives us some data on the impact of the podcast.
We’d also be grateful if you would take a minute to rate us on Apple Podcasts or wherever you listen. We personally appreciate reviews and ratings, and they help us show up when people search for content on leading a learning business.
Finally, consider following us and sharing the good word about Leading Learning. You can find us on X (formerly Twitter), Facebook, and LinkedIn.
Using Data to Drive Decision-Making
Leave a Reply