Content-type: text/html Downes.ca ~ Stephen's Web ~ AI in Enterprise Learning Systems

Stephen Downes

Knowledge, Learning, Community
AI in Enterprise Learning Systems


Automated transcript produced by Google Recorder / Pixel 8 Pro


[Speaker 2]
So I've been given the task of talking about Ai and Enterprise Learning Systems, and let's just get a sense, how many of you have knowledge of Enterprise Learning Systems? Yeah, that's kind of what I thought five. Yeah. How many of you have a deep background in AI? Yeah, so you see my challenge Um, But but I decided to do a bit more because I'm like that. And so I'm also going to talk about how to be a futurist. All right, right. So I'm going to give you some Pro tips on how to be a futurist. Uh, can everybody hear me? Okay. In the back. Okay, everyone can see the screen, okay? All right, so, this presentation is called Ai in Enterprise Learning Systems. Pro tip. Number one, put a QR code on your screen to the URL, where you've put your slides. So that people can get the slides without any fuss or masks. So that's where they are. Um, No, I discovered about five months before the talk that I'd put the PowerPoint slide link in incorrectly. Uh, it should be fixed but if it's not just give it time. It'll update the PDF one feels worth working. So, and after this presentation, At this URL or following QR code, you'll be able to get the The slides, the audio recording which is on here, the text transcript, which the audio recording is producing. Um the video recording which hopefully Zoom is producing. And if we're really lucky, the zoom recording will have captions So, you'll be able to experience this presentation over and over and over. Again. Um, just just this is the only AI produced piece in the entire presentation. Um, I asked it. I asked chat GPT. Chat cheap. You know, he began with 4-0. The newly released Omni which I really think is 3.5 in Disguise personally. And and I and all it would do is produce regression charts. Um, And that's not what I wanted. I, I wanted a diagram of the different capacities of AI, the different capabilities. So I went to chat gpt4, And this the best is the best. It could give me. It gave me such gems as 108. Only action. Dose. Grainy. Mo City. At least they're all like. Real letters.

So yeah, I'll be doing this the human generated way for the rest of this presentation. I want you to feel free. To stop me and ask me any questions as I go through. It's, it's the old thing about these talks, right? This talk isn't for me. Um, it's for you. So if it's not working for you, tell me. Uh, it's really important that you do that. Otherwise you'll be like not only bored but bored and disappointed and I don't want that, we'll just handle board. So yeah, uh, you know, uh, just fling up a hand. I'll finish my sentence because I'm usually in mid-flight and uh, then address your concern as thoroughly and completely as possible. Sounds good so far. All right. Here's the abstract. Which Saul got out of me after much coaxing. Including a desperate email saying, are you okay? Because I I tend to put off preparing presentations to the last possible moments and that's because like, you know, things happen in this field. Like Two weeks ago. Um chat gbt40 did not exist. No it does. That's that's the sort of thing. That's why I put it off but anyhow that's the the presentation. Um, for the sake of the recording, This presentation will provide a general perspective on the Probably use of AI. Yep, in Enterprise Learning Systems like learning experience platforms, Learning Management Systems and talent Management Systems. And what that might mean for learning, we will explore uses including learning performance, analytics, learning and competencies assessment content, and learning path recommendations and similar types of work. He will also consider how various standards such as x-api, facilitate intelligence, and Enterprise systems, and consider, wider issues, such as data, management and analytics. The Enterprise AI workflow and evolving needs in Enterprise learning and Development. My managers were very impressed. Oh, that's a mouthful. Um,

Whoops, other way. Here is actually what to expect. I got kind of keep to the abstract. I don't always Uh, some they bring me to speak anyways. Um, but uh, this is what I'm up to, for this particular presentation, we'll do a bit of an overview. Of of the technology. And then zoom in on what AI does. Then we'll look at these Enterprise Learning Systems. Then I'll do an interlude. Like I say I'm going to talk about how to be a futurist. So I'm going to talk a little bit about how we actually do the predicting. Um, Then we're going to actually make some predictions. That'll be the fun part. And then I'll give you a tip about what the pros do to get the big bucks. All those Consulting gigs. Um, you're just gonna kick yourself. And then looked briefly at some wider issues saw last, because he knows, All right.

So, We gotta put AI in a context if you don't put AI in a context. You oh I can't wander. Um, If you don't put AI in a context, you're going to get really confused. There we go. And just improving the video a bit. Um, Because, Ai depends on a lot of stuff that is not AI. Especially going forward is going to depend on a lot of stuff that's not AI and really I consider it. One of three key technological developments that have happened over the last. I don't know. Decade two decades. Now, as previously mentioned the history of AI goes back to like pre-history almost before I was born even um The same with these other Technologies, but they've come together recently. Um, especially with Chad GPT and stuff like that. But all of this has been years in the making and frankly, it was fairly easy to see coming. I don't know why everyone's so surprised. I get to say that. But okay, so here's the first. And the first is surprisingly, the metaverse. You guys eat some? What? Goggles on your well. Forget about the goggles on your eyes. Forget about, I mean, the metaverse is virtual reality, extended reality and all of that. But, and but VR has been around for 20 years. I mean, I played around in a system called active worlds. When I was

You know, so it's no, what's new? And what makes the metaverse interesting is the idea of persistent objects. It's not simply VR. It's that you using VR, and somebody else using VR could be working with the very same objects, the very same things, virtual worlds, right? And, and so, you're in this environment, where, you know, you if you're playing a game, you're fighting the same monster, you're in the same room, whatever. Persistent objects. That's historically, been a challenge. Believe it or not for computing, the metaverse conceptualizes that it was still tons of work to do to make it useful. But we got with persistent objects, some examples. Distributed identities. Right, you'll see some nodding from the from the people who know here. Uh, the login is messy. Authentication is horrible. Uh, it's worse than a horrible and and and, you know, What do you do? Like Facebook, Twitter, you can't even talk through other people. In other systems, you're two completely different identities. We want one identity for one person but without a police state, right? And, and that's part of the trick. So it has to be a distributed system, it has to be a personal system Etc. Really important. Open educational resources. You might say, well what does that have to do with the metaverse? Well, Imagine every open educational resource was a distinct nullably distinct unique resource. And its identity was based on its content, it's called content-based addressing. Um, So now you can create entire graphs of open educational resources but the main thing here is you've solved the problem of authorship, you've solved, the problem of provenance citations attributions and all the rest. So that your resource, you don't need licenses and stuff like that, your resources open because it was made in open resource and nobody can change. That fact later. Badges and credentials. Yeah, and again. Uh, you want a system in education, where if somebody gets a credential, it's a real credential, they can't create their own credentials. They don't the credentials, don't go away just because you owe the university money, 20 years later. Speaking

[Speaker 3]
From

[Speaker 2]
Experience.

So, You get the idea, right? All of that works with AI or will work with AI. That's just one. The second one. Is blockchain and you should be going. No, not blockchain. Um, Yeah. There's Bitcoin, and ethereum, and financial markets, and all of that stuff and It's that's all well, very scammy Very flaky. I wouldn't trust it. I have never bought a single digital coin in my life and I won't because It's a scam, right? It's the financial markets doing what the financial markets do. But the technology behind blockchain is really cool and really interesting. For example, I mentioned content-based addressing just a few seconds ago, right? Remember that. So you have an address that's tied to the content of a resource. Now, you take that address, you make that part of the next resource. And now you have a new resource that contains the address. And now you give that an address that's based on the content in the previous address. I've just described blockchain to you that. That's what blockchain is. Right? So you're able to chain these uniquely addressed resources together to form up maybe a chain, but in the future, more likely a graph. We already use these today these these exist. How many of you have heard of GitHub? Github is 100 based on that system. It's called. A directional acyclic graph or dag and and that's how it keeps track of all the branches and the changes. And that's why you're doing all that stuff with commits and pushing is to make that bit work. Of lots of other stuff. Similarly, uses content-based addressing networks to create these kinds of things. This is leading now to distributed systems, distributed systems are really important. Uh, distributed system. Well, I'll give you an example. You know what went wrong with Facebook and Twitter, right? Right. Uh, you have one centralized system. Owned by one evil, billionaire. Right? And all of a sudden you're getting served up posts promoting Nazis and police state. It goes south really quickly. If it's not managed properly sometimes even if it is managed properly, uh, all kinds of other problems with it too. So people are developing alternatives to these closed, centralized social networks, called Federated social networks, or as it's usually called the fetiverse. So, it's interesting, it's gonna reverse the FED reverse And the federal verse question.

[Speaker 4]
Uh,

[Speaker 2]
Up.

[Speaker 4]
And DJ loves to be used envelopers because they include robotics thoughts of money or

[Speaker 2]
Uh, yeah.

[Speaker 3]
I I that

[Speaker 2]
Was my technical response.

Vetiverse captures what's happening in a way that Omniverse doesn't right? An omniverse could refi, could refer to a pile of rubble, right? But if Eden adverse implies importantly, That each individual social network is connected with other social networks. And that's how the Federers works. So there's a social network called Mastodon actually. Mastodon is the type of Technology. It's named after the not the rock band, but the great big elephant likes him. Um, And there were different instances of Mastodon and they connect to each other and you need all this kind of pure networky, sort of stuff to keep the whole system hanging together. I'm not going to get into the tech of it because it's not important to this talk but what is important is you can have these distributed connected networks of stuff all interoperator

That's important because when we think about AI, especially these days, look at what we're thinking of, we're thinking of the great big monster, that is open AI, the great big monster that is Google Genesis That's the current name for it. Uh the great big monster that is uh Whatever, right? But really down the line that's not what AI is going to be. It's not going to be a great big system. You work with. It's going to be a whole bunch of separate systems all into our operating together. That should really change a lot of what you're thinking about AI, and I haven't even gotten to talking about AI yet. So far so good. Okay, the third piece is AI.

It does a whole bunch of stuff for us. We saw a bit of a description in the previous, talk about how it works. Um, And and, and a bit of a description or a bit of an account of some of the ethics. If you go, if you follow, oh, not this place. Oh, Never mind. I did not say that. Let's start that sentence again. This and the previous, two slides. Can be found at this link. On the bottom of this slide which is why I gave you the length for the slides. And that's a whole other presentation building out. These three diagrams. So all the stuff in these diagrams actually refers to things and and all of that and try to explain and applied to a learning context. But that's a totally different subject. So, let's focus now. On an artificial intelligence itself.

There have been a billion different ways. In describing, what AI does This is my work. My way is the right way.

[Speaker 3]
Um,

[Speaker 2]
Okay. Maybe just more right than other writers, but So, What I've done here. And I actually did the this is from a course that I did called um ethics analytics and the duty of care. Um it's it's the porcelain AI ethics. Uh, this particular Slide is in this particular presentation. So you can go to that if you want the main URL is ethics.mook.ca. And so a lot of the details on how AI works. The sorts of things AI can do Etc, are discussing quite a bit more detail. But again, that course came out before Um, openly I came up with chat GPT and arrest. That's why I say yeah it's easy to predict this stuff because most of that stuff except the exact names I didn't get the exact names is in this course. You don't like the echo? No, I don't like, yeah.

[Speaker 5]
I just realized it was there sorry.

[Speaker 2]
I've been working with this Echo all along. I just noticed his objects. That's what you call being a pro.

[Speaker 3]
All right.

[Speaker 2]
Still act close by the way um, my way of dealing with, maybe it's on my computer. Yeah. No, it's coming out of here. It doesn't matter. Uh,

Mine isn't based on the the types of things that it does so much, you know, just trying to imagine. Now what could it do? Call a cab plan a vacation My mine is actually based in. What we might call the logic of grammar. Um, you can think about it, right? Grammar. That useless subject you took in grade school or sometimes known as grammar school. Uh, great deal say with tenses, very intensives past, present future different types of Futures. It also deals with modalities could would should make Etc all the different tenses for verbs, right? So I used a bunch of those tenses. Not all of us. But but you can imagine how I could use others. Like, you know the history of AI, I didn't do that. So I'm not worried about the past tense. Um, But so, And and I kind of focused on analytics proper but and for the very broad conception of it and so I got Six major categories. The questions that we look at? Well what happened or you know what is happening now present What kind of thing is happening, classification? What will happen? So, past present future kind of Make it happen. The whole causal logic, make something new. And what should happen? The first four were already current in the literature when I created this course. So, I just borrowed those terms from other authors, descriptive AI, diagnostic AI, predictive AI. And prescriptive AI. The maximum was new when I wrote it, but it's all their age now. Generative AI And the last one nobody's talking about yet, but they will. Deonte Kai. What should happen? Yeah, it's when AI starts telling you what to do, instead of you telling it what to do.

[Speaker 3]
You

[Speaker 2]
Don't think that's going to happen. What do you do when you log in? Follow the instructions on your screen? Right, it's very simple. Example, of a computer telling you what to do, but you do it, you do it because you won't get access otherwise, In each of these. There's there's all kinds of different sub sets of AI. Uh, I'm not gonna go into them, I'm not gonna go into these in detail, I could In in the course that I did a few years ago, that was one of the modules, right? This was the first part of it. And I spent the next week, okay, we'll go through this and this, and this and this and this because I'm thorough that way, you'll see this But you know descriptive AI systems analysis, institutional compliance student profiles, right? This describing things all the different ways you can describe things. What kind of thing happened? Diagnosis Diagnostic spam detection. Proctoring that evil company proctorio. We don't like them. Uh, fakes detection, which we do like when it works. Sentiment analysis. All right, predictive what's gonna happen? Uh, resource planning, learning design, user testing. Identify students at risk of failing. This was all of analytics, three years ago They couldn't think of anything else. How do you predict that a student's going to fail? Well, maybe if they did the work, you know, Uh, making it happen. Um, learning recommendations, adaptive learning paths. Uh, placement, something, something hiring pricing. Generative AI. This is what everybody's talking about now chat. Pods AI generated content, I did a whole paper on AI generated, open educational resources, set up in a Content. Addressable Resource Network, I called it con content, addressable resources for Education or care. Has nothing to do with the care in ethics analytics and the duty of care. It's just a neat acronym. No, I went nowhere but you know five ten years from now. It'll be a thing. What should happen? Well, Community standards AI could determine what your community standards are probably more accurately than say the premier of the province. Uh, identifying the man amplifying the good defining. What's fair? Um, United States has a big gerrymandering problem in. Political Distributing, use an AI to design, you know, with with agreed upon parameters to design the system of congressional districts. It'll completely change their election results. Changing the content moderation, easing distress the area. That really does care. Um, You might think AI won't care, but an AI won't judge you. Well, unless we ask it to, but generally in AI won't judge you. That goes a long way. All right. So, Husky overview of AI, right?

[Speaker 3]
Pretty similar to

[Speaker 2]
What you just saw in the previous, uh, presentation, right? A lot of overlap, just organized slightly differently. The organization that's really the key here. Just as an aside education is a discipline that loves tax on earnings. Uh, I'm not a taxonomy lover. At all. Um, you know, taxonomies are useful. But unless a taxonomy is principled and based in something like, say English grammar, or has a practical purpose, there's no reason to do it. Right. Um, in previous. Now, I really liked the previous presentation and it was the one bit. Um, From a Swedish author built, I'm no good with names. Who did the, uh, The three types of Education, the qualification socialization and Anyone. Because I've forgotten that subjectification. Yeah, that's it. It's just attacks on me. But it's a useful taxonomy, if you keep in mind that most education is only doing the first two or the first one, mainly, maybe a little bit of the second. If you're in one of the elite ivy league, universities hardly ever the third But yeah, that's really interesting to me. But, you know, simply dividing the purposes of Education that way doesn't do anything, but actually applying that and and asking, okay well how can that be used to change, what we're doing, how can that be used to change? What our focus is. Then it becomes interesting and important. But yeah. Overall, too many taxonomies not enough cause and effect in education generally. Um, I'm old school, that way. My background was, as a philosopher of science. Can you tell?

[Speaker 3]
And

[Speaker 2]
So these issues matter to me, you know, issues of what counts, as evidence. Uh, You know what is the reason why we're doing the thing that we're doing all of that matters to me? And all of that lead to the creation of diagrams like this. So this is the part where I'm talking about being a futurist, okay? Continuing back with our regularly scheduled subject, All right. So, we got More than three actually. By the time we're done, monstrosities ahead of us. Enterprise Learning Systems. Now, I think Saul was just giving me a list. But I took that list seriously and made that the content of the presentation just in case that is actually what he wanted. Um, so I'm gonna go over those now. Keep in mind, these are big huge pieces of software there. They've got multiple components, tons of moving Parts. At their core. They're pretty simple. But because they're dealing with thousands and thousands of students for a given institution, they get complex in a hurry. I'm going to try to keep the complexity to a minimum and focus on what they do. Now there's a risk there, right? The the risk is maybe I'm not addressing what's really important about Enterprise Learning Systems but that's the direction I've chosen to go, sometimes you just gotta choose. So learning management system. You've probably all used one, haven't you? Yeah. What do you use D2L here? No, no, poodle noodle.

[Speaker 3]
Oh, okay.

[Speaker 2]
Oh man. Moodle. Oh, there's a blast. No, never mind.

[Speaker 6]
Pardon

[Speaker 4]
Him

[Speaker 2]
Just upgraded, well, even more open source thing. So, okay, with all all of these parts will be familiar too, right?

Well, accepted men controls. Hopefully So you're going to have online course, materials. You're going to have alert tracking of learning. You're going to have a testing system with new and improved metrics problem with Off-Blowing course material, you know. So maybe a link to the library or whatever. Communication web. Mobile Moodle does have discussion capabilities. That's nice. A lot of lms's don't. Especially the lms's that they made for moocs and they called them mooc systems like Coursera and Udacity, they're just lms's. Cloud-Based backhand so they can handle hundreds of thousands of people and no discussion because that's hard. I don't want to criticize them, but, They're they're a bit more up to date now but that's what they were when they came out. So that's basically a learning management system, that's what it does. Right. And then the whole idea is that it gives you as a student access to these things access to the materials access to the testing, and then it'll track you and hopefully it'll be connected into other systems like student information system or whatever true. That's, So that's, you know, that's all I want to say about the LMS. Can you believe it?

The Learning Experience platform, is an LMS with attitude. Working did quite a bit of work over the years with desire to learn, good company, good people. I like them. I like the Moodle people too. But um, they changed their branding from a learning management system to a learning experience platform. And they they go red in the face when electrically. But when you refer to them as an LMS, no, we're a learning experience, platform, the same technology.

But again, this got parts. Where we've got our learner again. So here's the and it's really hard to read. Lxp l m x d or similar One, Stop Shop competency based learning Journeys. For example. Degreed, sum total Uh, and NRC we use when we use skillsoft. Uh, We have other names for it. And so here are some of the things, right? External content, adaptive learning. Uh, this is old style adaptive learning where it was basically. It was basically very primitive machine learning until very recently. The AI in these systems was really primitive. A coaching platform. Here's a link to the LMS. For example, success factors. Oh, I think we use success. No, I don't. Uh gaming, which you're not really supposed to acknowledge uh whiteboarding Etc. But also down here. We've got, it's called a data Lake. And if you're wondering what the technical term, technical meaning of data Lake, is, you take all of your data. Just put it in a big repository. That's a data Lake. So but your data Lake's useful, because you do some analysis and stuff like that, you organize it, sort it filter it, flake it form it, and then that'll create for you a data Mart. And a date of March is one shot stopping for all your data needs. But that really kind of goes outside the realm of a learning experience platform. But there is a data Lake, and the data Lake is used by all of these systems. We're talking a little bit more about that. And then down here, the content creation, bit and a purple other button. So, Yes, of course.

[Speaker 3]
For the slides are going to separate

[Speaker 2]
The walls. For our concentration, was the different ways. That's a good question. Let's have a look.

So, Content creation, for example, articulate podcast videos Etc. And content creation, I.E, red nucleus illuminate, ERS, professional learning and whole systems, Etc. I have no idea why those are sound. Honestly, I, I thought when you asked, I thought, okay, well maybe one box refers to formal learning resources, Called scorm shareable. Um, course object. Shareable. Yeah shareable courseware object reference model I think. Yeah got it good. It's been a while. All right. So actual packaged learning resources and then I thought the other one might be just stuff. You you created on the Fly like a lecture recording like we're doing here or something. But you

[Speaker 6]
Know on the right there those are all content development companies for the new keywords Vision, they're in the arm of space. So he'll head to the Pharma

[Speaker 2]
Um learning

[Speaker 6]
Organization. So they must have been the ones who developed that. So you probably have licensed content versus just, you know, like we're saying.

[Speaker 2]
Versus content, he created in-house. Perfect. Very good.

[Speaker 3]
Thank you.

[Speaker 2]
Thank him. Didn't occur to me. But yeah. Oh, how about that? I love boys problems. Get solved. Sometimes here's another picture of a digital experience platform. Right? And the main reason for this picture is to, is to underline the idea that these are kind of fluid Concepts, right? Uh, you know, not every LMS is exactly the same, not every LX. Lxp is exactly the same. So, Uh, they're gonna have slightly different components, but more or less. You know, we have interfaces With uh, other pieces of software. We have. Experience apis, what? I'll talk which I'll talk about and user or ux user experience Design Systems. So that's what you use to actually create the shell or the template of of your learning experience. Again, I'm there's more details there. Supporting digital platforms. So social management conversation management payment Gateway.

Digital intelligence, contents and digital sets, Etc. And then the integration and data layer to Enterprise integration platforms. And then down here at the bottom, as usual, your data platforms, they call it data platforms and then data and data domain, but really, you know, it's data, Lakes data marks, whatever. So there's a whole science of that. That's all I want to say about learning experience platforms. You kind of get the idea. Can the details aren't that important? But the overall thing is important. What's the difference between a learning management system and a learning experience platform is the one is old school, content management and teaching. The other one is more about creating a learning experience with different kinds of resources, different kinds of services, so that you're not stuck in the one single system. And again, that's a fast and loose distinction between the two. Third. And on the Venture, capitalists, love this. Talent management Systems. Yeah. Every large Enterprise has a talent management system. Your University probably has one. Do they have one here? Yeah. Oh yeah. Yeah. Um so basically it's the thing they call it Talent management system. It's the thing that handles the whole human resources function And in a lot of organizations including my own uh training learning development is considered a human resource function, not core of the institution, the way it should be. Um So, Some of the things while planning, but recruitment employee onboarding, performance Management. This is actually on the job, not taking tests, right, compensation management. The, uh, big thing about the government of Canada and Phoenix, Compensation management. Did it work? No does it work? I'm getting paid but it took three years to register but I moved from New Brunswick to Ontario back. 2015 and learning a professional development. So he got the idea, right? This was the simplest diagram of any of them that I could find. I tried, this is the diagram. I used to try to get Chat GPT to organize because I wanted something like this for artificial intelligence, right? Diagnostic. But no, it gave me that Abomination. You saw All right. So, Let's put it all together. So this is my Rough and Ready diagram. So, the learning management system. Basically, about learning management. Correct course, contents, testing tracking people keeping track of your grades stuff like that. Um, over here. Off the chart. We might have a student information system. Um, although in a corporate Enterprise, that'll be handled over in the TMS. I'm also off the chart. We might have what used to be called an lcms learning content management system. Those are all the rates for a few years but you know, some sort of content authoring. This is where we'll have our, I guess, our two types of, you know, the content producing companies and the in-house content creation systems articulate stuff like that. Okay, The Learning Experience platform really is kind of at the center of everything. Um, when now, I'm not talking about personal learning so much in this talk, but when we talked about building a personal learning environment, It was this, except based on a person, not an institution. This is based on the institution. And it connects, To the learning management system using scorm. Um, it'll connect to a login or authentication system. Um, do you use edu roam here? Yeah, yeah, yeah. So AJ room he might it might connect you into IG Rome shibboleth is another popular one. Never really gotten to hang a Chevrolet to be honest. It's it's yeah. So, Uh there's a thing down here. Called a learning learning or learner record store lrs. An X API. Uh, called the experience API, which is kind of, we have a learning experience platform. Experience API used to be called tin can and then some company trademarked the name Tin can. So the entire discipline said, well, now it's ax API. That that is absolutely what happened. I'm not making it up. So, Uh, and other apis, okay? So here's our talent management system. That'll be linked to the learning experience platform. They'll be able to tie into all the employment, records in that and Oh yeah, this is the credential store. I looked for an abstract representation of the credential store and I didn't actually find one which I thought was kind of interesting. Maybe it's an opportunity for work there. And again he's connecting. So this is your badging system. Your degree system, any anything your skills profile, your your portfolio, anything that relates to your actual learning credentials. This overall is a rough, very high level, approximation of Enterprise Learning Systems

This is years in development. Uh, millions of dollars for the different pieces of this. And like, this is the top layer and then behind this. Well, I've got a page at the end of the presentation on what did I call it? Advanced issues or further issues or whatever? Or I'll talk about some of this stuff. But why is behind all of this? But this is The Superficial level. The user facing level, we might say. Any questions on this part? Yes,

[Speaker 7]
I didn't register. Um, so you said when you were at the Atlas key you said this is based on the way I registered and based on the it's not based on the individuals based on the Enterprise. Yeah. Can you explain more about that please? Yeah.

[Speaker 2]
Just trying to come up with the metaphor. Sometimes it works. Sometimes it doesn't.

Okay.

Think of third-party applications like say, A discussion list. Um, discourse. This course is a wonderful discussion list software. It sits alone on its own server. Right. It has discussion channels and threads, and things like that.

That's the sword of service that an lxp might connect to Right, because you want to give Your students or your participants access to a discussion list. Now, if it's based on the Enterprise, Then your access to this discussion list is almost always like 99 of the time exclusive to people who are in this Enterprise. So it's the, you know, it's the Ryerson discussion list. Is there such a thing? Oh, we're not at Ryerson. How are you?

[Speaker 3]
We're Concordian Ryerson doesn't exist anymore, right?

[Speaker 2]
I told you I'm not good with names.

[Speaker 3]
Very honest,

[Speaker 2]
This Concordia have a discussion, they've gone out of fashion recently because people actually use them the community. Very specialized enough for everybody. They're very specialized and not for everyone. Yeah, that's pretty typical. Yeah. Government of Canada, same thing, right? Uh government of Canada has a thing called uh GC connect. Which is a government of Canada. Discussion area for employees. And then there's a whole bunch of sub forums and things like that. It's very extensive. Um and you'd get into it using your your credentials and you go into this. So that's the Enterprise focused one. But what about all those people? Who aren't University, students don't work for a big company don't work for the government, right? Where do they go? Well, they could go to Reddit. All right. But, Really. I, you know, I mean, it's like they have to go to these Standalone discussion areas. But they want to learn too, they want access to all of, you know, learning content and such. So we set up a special environment that we call a ple That's just for one person. And that person will still log into their own PLU their own credentials. But now when they log into the discussion list, they log in as themselves, Now, it's not just a discussion list with one person that would be useless. So other people using their own peles log into it as well, their own credentials. So it's distributed right, one discussion list. Multiple accesses.

[Speaker 8]
What do

[Speaker 2]
We got?

[Speaker 8]
Okay.

[Speaker 2]
I don't think they're in here. No, it's five hours a week. Yeah. I'm in Ontario Eastern Ontario. Well actually not even not far from Montreal. You can hamburger alerts like from Thunder Bay. I'm serious. And and and the kid's been missing, you know, I mean it's a serious subject and of course we support Amber Alerts but maybe those that are Located near us. Um, because you know, it's it's an hour old and they've been missing for like an hour and we're not even a one hour flight from time nearby Okay. Where was I? Oh yeah. Um, Ple. One system for an individual person, multiple people accessing the discussion list. The problem with ples is, I've just described it. They don't really exist. So annoying, it's something that I've been working on for years and years and years. Really hard. I mean, you think Enterprise technology is hard. Try attributed. Learning technology. That's hard. And and there's no market for it which is what really makes it hard. Right? Uh yeah. Places like Concordia will pay millions of dollars for a learning experience platform. Nobody will pay for a personal learning environment. Who's gonna pay for it? So, there's the problem. But the concept is there question.

[Speaker 3]
Um, so something like GC connects, they were mentioning your current employees use their government ID to login. Yep. Thinking about a PLE will be like any member of the public could. Also participate these discussions that they'd be using a separate personal login, instead of conductive body.

[Speaker 2]
Uh, If the, if the GC connect allowed it to happen, yeah. And there are some, there's GC collab. Does that actually, is set up that way, there's still a little personal learning environment, but Members of the public can log into it using their own credentials. And actually talk with government employees. And then again, there are different subgroups and discussions and things like that. If you're curious, yeah, look up GC, collab. There's a whole bunch of stuff like that, you know? I mean people don't talk about it but it exists. Not to tote the government or anything because I, I wouldn't really, but the last three years, especially They've gone from Stone Age. Digital technology to almost mainstream. It's been incredible. They've done a lot and nobody sees it, but they've done a lot, a lot of it was caused by the pandemic, right? Which is why so many people are really upset that. Now they're being forced to return to the office because for the technology doesn't work. It works better at home. Okay. Yeah. Yeah, like going to the office, all my internet's going to be slow, if it's up. All right. Yeah, you got my home trouble here. But if they fire me, I get severance. Six and one half dozen of the other. Okay. Now, this is the interlude. I hope I'm sort of at halfway. How am I doing on time,

[Speaker 3]
Um, about 17 minutes?

[Speaker 2]
Crap. I really hope this was interesting to this point because because this was supposed to be the halfway mark,

[Speaker 6]
Right. Oh

[Speaker 2]
Okay. But but it wasn't, you know, I mean like I say, I sometimes promise to follow the agenda, Okay. This is the interlude. Here's how we predict. Um, and just as an aside, Which I can't really afford with 17 minutes. There's so many you talk to futurists and they create these four quadrant scenarios right to me. That's fake futurism, right? What they're doing is they're creating Um, one range of possibilities, another range of possibilities. Right, giving a square for each quadrant and giving it a cute name. And the quadrant in the upper, right, that's what they really want you to do. It's more of a marketing device than anything. Yeah, I I don't do that. I think predictions should be clear, precise and accurate. I don't do the scenario thing. Well, how do you do that? I mean, it's the future hasn't happened yet. Well, I used to say long long ago. And I still say today, we read the future, the same way, we read the past, right? We look at the science, we look at the data. We've got We've got tons of data, especially now, right? We can predict the future really easily, you know, like, uh, like they're gonna kick me out of this room sometimes. Sometime within the next hour, I'm going to be asked to leave. Guaranteed to happen. Um, All kinds of things. The sun will come up tomorrow, the trains will continue to be late. Actually via rail is pretty good for on time these days for me, but again, Brand new trailer, it took a brand new train anyhow. So What do we got? You got the learning management learning experience talent management, three systems that I've just described in some level of detail to you, Those exist. We've got artificial intelligence metaverse, blockchain, those are types of Technology that exist today. That I've just got described in a little bit of detail for you. Those exist. So, the futurism is What's in these boxes? All right. Easy.

I told you it's not rocket surgery. So, let's get a little narrower. All right, so This is where the taxonomy kind of comes useful in organizing your thoughts just like, Grammar. All right, the script of diagnostic for Learning Management system, learning experience, platform Talent management system, the predictions are in the boxes. So far so good. Okay. I'm actually going to make this in 17 minutes, but we're not going to get the breakouts I'm afraid. I hope that's okay. Let's go even narrower because we have the details we need. All right. There's the all the types of AI across the top. It doesn't really matter how you break it down. Just break it down rationally. You know. Use basic principles of categorization. Um, you know You know, each category should refer to all and only something. All and only something they shouldn't cross-categorize, right? The the the standard way of producing a taxonomy is to show How they're the same and how they're different from the other things. I think there are many ways of producing a taxonomy. I like to use grammar because grammar is connected to logic. Logic is connected to cause and effect. There's probably going to be something useful there. And down the side we have the elements of the learning management system. I've picked out three just for that's position. Coarse materials, testing of learning tracking of learning. Simple right. So now, How could I use artificial intelligence? Descriptive. Artificial intelligence. For course, materials. One of these hearts. Pardon.

[Speaker 6]
Not very, I had a few ideas,

[Speaker 2]
Let's hear one.

[Speaker 6]
Well, if you wanted to create like a smaller, uh, whatever you would call it a description of all of the calendar of learning materials available in two sentences, three sentences. Whatever. Yeah.

[Speaker 2]
Kind of a summarizing function, but yeah. I'm thinking. One of the big challenges and this was a huge challenge in the early 2000s was learning object metadata People have forgotten it now, but So learning object metadata is data about the, the learning object. So Uh what its total is what format it's in if it's a video, how long it is, if it's an image, where this dimensions are, what the typical age range is, there's a whole settlement learning object metadata, l-o-m-i, Triple E 14. Something something 28, whatever. People is 57 87, whatever separate items. But people used to describe learning resources. And their accuracy rate was pretty good, but you could count under being errors Javi. I do it. Create your own automated learning object metadata. So that you don't have to classify your learning resources anymore. Your AI is looked into that's a PDF, that's an image. This is about Cowboys. This is about firemen whatever.

Do that for each box. It's easy. I didn't say it was. Quick.

So here, here's my predictions. Sorry about the script but I just kind of wanted to make it here. For course, materials. For descriptive, count word frequencies. How many reads was there? How many upvotes, you know, just any anything that describes the material Uh, you might think well that's not very. Interesting. Well, think about and they were they were just showing it on. You know, one chat jpt, oh came out. Cheapy teal. Yeah. Borough world. One of the things it does is that you ask, what am I looking at? And it describes the scene in front of you, all all of a sudden descriptive is calling it. Right. So, you know, uh, okay. For the next one. It's hard to read Diagnostics or classify the resources, identify topics extract, keywords, automated, and keyword extraction. Oh, that would save me. Titan six words and I'm just kidding that. I hate doing keywords. It's editors always come back and you say please add keywords your article, I'm sorry. Great quality Etc. Um, Of course materials for I think that's predictive access readability. Give it a readability score, you've actually seen systems that do that probably All right. Prescriptive do content recommendations. Learning path planning. Right for, uh, generative create novel content. Ai, that writes. Course materials for you, it is coming. Yeah, you might be saying, well yeah. But what about the hallucinations and all of that? Well, that's why I started this talk. The matter verse and blockchain the metaverse in the blockchain. Keep you honest, right? You can't hallucinate. If you have to refer to actually existing objects. Right. There's already some of that. They they it's in, in AI. You have your your model and then you give it a quote unquote context in which you're going to use it. That's that's how they're using that term at the moment and the context can actually be like An entire paper, an entire library of papers written by somebody. And then the instruction in the prompt which is the third part actually, part of the context. But the third part, you say, do not vary from the facts stated in the context. And then your AI is not going to make stuff up. It has stuff it has to use.

But as we apply, Uh, you know, persist uh, persistent objects and and uh, Contact-Based addressing things like that. Our AI gets more and more accurate. So, accurate, in fact, it will correct. Correct us, rather than us correcting it. Count on it. It's coming. Think about your relation with a calculator. How often do you question? What the calculator does? Times you should. But usually if you add it in your head and you look at the calculator and there are two different numbers. You figure the calculator was probably right question.

[Speaker 3]
I work with programmers in an agile environment.

[Speaker 2]
Yep.

[Speaker 3]
Changing on an hourly basis. Yep. Will AI also, facilitate course updates and course course, accuracy. So that say, of course it was drafted, a week ago and there's an under changes, all of a sudden, boom. No, identify those changes for me or I need to implement them

[Speaker 2]
Or

The short answer is. I mean, well, next week, you'll still have to implement it right. 10 years from now, it'll all be automatic. Because your AI will be tied into Um, Um, the global linked data Network. Um, we're able to used to be called web of data. Now there's there's an acronym, it's not Global link data Network but that's basically what I'm talking about. But yeah, absolutely. And how will, you know, that what it's getting from this data network is accurate, every piece of data will be tagged with a unique identifier that can be traced using a blockchain back to the person who put it in the system. Thank you. It's come, it's it's most of the all the, all the pieces already exist, easy to make the prediction. Yeah,

[Speaker 6]
I'm just curious what this last thing is just right there because I kind of figure out where instructional designers live in this world, where you're talking about you know, that internet and knowledge. So is it do you think it's instructional designers, who will become the Gatekeepers of what gets to be considered tagged and trusted and references? That'd be something that we should be thinking of, as a, the brand of an instructional designer is somebody who can actually I've been doing for decades already packaging, learning objects now, and to another

[Speaker 2]
Why Short term. Yes long-term help. Yeah, long-term that function becomes completely automated

[Speaker 6]
Easily.

[Speaker 2]
Count on it. Yeah, so you do this systematically right now. I've got now when I had more than five minutes left now, it's three minutes. I was gonna break you all into groups and you're each going to do one of these. So I'll leave that as an exercise for you. But Here are the oh before we before we even get to that, after you do the boxes. Here's one right now. Ask the important questions. This is what makes you a futurist, all right.


What problem does this solve? It might not solve any problem, that's not necessarily a problem. Lots of things don't solve problems are still really useful. Um, What new thing can we do. That's usually the really useful thing. Right? We weren't able to fly before. Now we can fly. Yeah, we didn't really solve a problem with airplanes, it wasn't a real problem, but now we can go to the Bahamas anytime we want. Wasn't an issue before what needs to exist to make this possible? Some of you should be thinking about Marshall mcloone here, what does this replace? What does this amplify Etc? Similar sort of set of questions. What would make this impossible? Right, many people say many things about AI are impossible but you get right down to the box. And ask what would make this impossible and you learn, there's far fewer things make any aspect of our AI impossible. When people think it's when people talk in Broad generalizations that it all seems like magic get into the little boxes, it's not magic at all. And then up here who or what would be harmed? Important question to ask. Obviously that's not necessarily going to stop people. But if you're the school of management, right? If you're cold-hearted business person that's opportunity.

All right, so Here's the the your turn thing. So, Here are the topics again. It's an exercise. And then we're going to come back and here are my predictions. You can read through these on your own because you have access to those slides.

And, But you can you look at each one of these boxes? Each of the different types of AI. All these predictions become really obvious and hard to resist. And the point I was trying to make here is you could have come up with them. You didn't need me at all. All you had to do was organize stuff properly. Now, the pro tip. Because there is a pro tip. This is what the pros do for the big bucks. And and it's really interesting and and I hope your indulge me just to weave it because I got a great story that won't close with this. Okay, so these are all the boxes that I created and all those previous slides, right? So, You connect them together because there's a workflow and you can fairly easily identify the workflow and Enterprise learning management Systems. And now you look at each one of these little boxes. And tell yourself what would happen. Now, what I've done is I've framed this And I did this, like, a few days ago. So using AI to support staff diversity at institution P. That's my Franny, right? So I've just picked that out of the other. The funny thing is, when I came to the conference today, I went to the wrong conference.

I went to the accessibility

[Speaker 6]
Across

[Speaker 2]
Canada conference and I thought Gee. That's pretty interesting as Saul, didn't tell me that. There was an accessibility Focus to this but I'm really glad I chose the example that I did and it looked like I planned for it and I look like a real genius. But but now you have the story, right? You're assessing and pre-screening, incoming applications. So that allows you to be selective. Now you're you're recommending training materials policies goals for that person specific for that person. You're also redefining your acceptance logic. For registration for sign in or employment and recommending for Target groups. Because of the demographics, you know, how to improve success rates. So, you're creating custom content, images, videos, Etc. They're able to assess the readability of that. So you can create test assessment rubrics, Etc, specific to the person. Uh, that's being assessed. We go staff diversity, right? So, how to improve that success rate? We're also going to identify areas of strength, weaknesses, topic preferences, for that person. We're also going to do learning path recommendations. And to help them out. We're going to identify compatible discussion lists and put them in touch with correspondence and mentors who will help them out. Um, we can predict what their role and their con in these conversations will be in areas where they'll be successful. So that allows us to design a custom user interface designed for them. And we can talk about what their responses would be. For specific experience. Designs or identify optimal presentation Styles. Um, digital modalities Etc, especially for people who need have accessibility needs That leads us to a recommended metadata perform profile data requirements. Etc. So the we can predict what their data needs will be what their data will be produced. Takes us down to being able to predict the sorts of requests that they'll make, what kind of demand that they'll put on a system. We can also, Therefore recommend what external sources of data and sources of Truth. That's the formal name. That database analysts, use for that, uh, you know, the web book data, the the verified data, And that will also, Suggest what their individual performance indicator? Should be again, supporting diversity. And that'll help us generate performance assessments, formative training, and recommend a compensation profile specific to that individual. That is the sort of futurism that you can do if you nail down all the details. And you can be very specific, you can flush out each one of these. You've got yourself a grade, A compensate, a grade, A consultant's report.

That's what instructional designers will be doing in the future. Not designing instruction. Who would do that computers? Can do about it. We don't need that anymore. What you do need is this story, right? Be able to tell this story. If you can tell this story, you'll learn your salary. If if you can design a web page not so much. Wider issues management and analytics. We talked a little bit about that the Enterprise workflow and evolving needs identity identity identity. That's it. Thank you very much.
 


Stephen Downes Stephen Downes, Casselman, Canada
stephen@downes.ca

Copyright 2024
Last Updated: Dec 09, 2024 2:05 p.m.

Canadian Flag Creative Commons License.

Force:yes