Al Kinnear
Well, I would like to welcome our viewers and listeners to the Return on Intelligence podcast. And today I'm speaking with Craig Basford, the Executive vice president of product at Absorb Software. Thank you for joining me, Craig. Craig and I worked together for many, many, many years. I always enjoy conversations with you, Craig. And today we're going to talk about artificial intelligence and its role in business and learning and development.
Al Kinnear
And I'm quite excited to speak with you about this today.
Craig Basford
Likewise, Al. Thanks for having me.
Al Kinnear
So, Craig, I think why don't we back up a bit and talk about just over the last 3 or 4 or 5 or 6 years when we were going to market at absorb, we would run into, you know, folks talking about AI and artificial intelligence. And I think back then, a lot of times it was a lot about machine language more than AI.
Al Kinnear
Whereas now I think we're starting to see artificial intelligence being developed into solutions. So why don't we start back there and just talk about the difference to the audience and what you've seen over the years? As you know, when you're thinking about product, not just for absorb, but I know you think about it, you know, in broader terms, just talk about the difference between machine language and AI and how that's just continue to evolve.
Craig Basford
Sure. So, you can think of AI as a bit of an umbrella term. Machine learning is a subset of AI, but there's other elements that are fit under that same umbrella. So what we're seeing today, for example, with generative AI is a subset of both of those. It would fit under machine learning. And then also because machine learning is underwrites its AI as well.
Craig Basford
And for a long time there were companies that were advertising what they did is AI because it sounds better. You know, it's science fiction has taught us that AI is like is what robots do. You know, this is the big smart thing. Machine learning is like, that's a nerdy thing developers talk about. It doesn't doesn't sound as cool.
Craig Basford
And so I became that much more popular marketing term, and the nuance between the differences got lost. I would not consider myself an expert, but in broad strokes, the machine learning is any model you have where the machine is learning from a set of data over time. AI refers to a broad collection of technologies that are trying to approximate human intelligence using machines.
Craig Basford
So, yeah, I think what we've seen with this latest version of generative AI is interesting because now the AI part of it is almost the less interesting. It's the generative part that's now novel, and AI is almost like if you're not using generative, when you talk about AI now, that's old school, which is kind of, you know, speaks to the pace of technological change.
Craig Basford
That's something that five years ago would have put you, you know, on the top of that, the pace of innovation now is like, sideshow to generative AI, which is the big deal. And it's likely, you know, in 4 or 5 years that there'll be something new. You know, we don't know what that's going to be. But this overall field, which continues to try and approximate human intelligence, continues to move forwards.
Craig Basford
And at some point in the future, that sci fi version of AI may become reality. But today we're not there. And every technology is just some approximation of some portion, some percentage of how humans actually think or how we are intelligent.
Al Kinnear
Okay. And so in your role, you know, you're in charge of developing a product. And in SAS it's a LMS. But you know, what is the market? What is the market been telling you over the last 2 or 3 years? Like what do you hear at trade shows? And like do you guys listen to the market? Like how do you develop product and keep your ear to the ground?
Craig Basford
Yeah, I or any kind of innovative technology, it typically.
Al Kinnear
Is.
Craig Basford
For us anyway. It starts more as a push. you know, in the space we're in and learning and development, we often may be the first to bring in innovation to our customer base as opposed to waiting for them to ask us to be innovating in that area. I think part of that is because we have a very competitive space.
Craig Basford
And so there's a lot of different vendors trying to differentiate, and technology and innovation is a good way to do that. but also I think that the area of, you know, HR and training and learning is anywhere people tend to be pretty careful. You know, if you look at an area like sales or marketing, for example, they're often the first adopters because they're, you know, there's no there's no the risk is lower for them, right?
Craig Basford
You know, whereas if you're dealing with people and you have to, figure out how people will most use this technology, how you make it work with the way that they already are motivated, how they think that adds nuance to it. So I would just say broadly, we think about innovation very carefully. We try not to just be a first mover for the sake of going quickly to get something to market so that, you know, you can put a label on your website and say, hey, you know, we're doing the latest thing, you know, you should come look at us.
Craig Basford
Instead, we take what are already known problems, things that we know our customers are trying to solve already and match them to upcoming innovations. So generative AI is a good example of this. you look at something like course creation when you think about how customers would like to build courses, they would prefer that more content is made by subject matter experts than, you know, dedicated course authors.
Craig Basford
They've been a problem for the entire existence of LED. People have tried to solve it in various ways to outsource, to spend money, etc. but now generative AI comes around and it's like, hey, we can maybe lower that bar for the level of effort it takes to create a course low enough that anybody can make a course like literally anybody.
Craig Basford
You know, if you have enough subject matter expertise to craft a prompt or to upload a document, you've already written them, then you know, just review the content. Congratulations, you're now, of course, author. So that is an example where I think when we we take innovation in perspective of a customer first view, what are the problems they have?
Craig Basford
How can we bring technology to solve. And that's where we absorb choose to innovate in that area.
Al Kinnear
And would you say, you know when you're looking feature function, having roadmap discussions like is AI the common thread in a lot of these discussions, is it kind of the dominant theme? Right. You mentioned a marketing term, but the reality of AI right now is that I think it's it's coming to the point where it is important. Right?
Al Kinnear
And people are wanting that in their consideration set. Would that would that statement ring true?
Craig Basford
It's definitely something we think about. I wouldn't describe it as a dominant theme. You know, when we we think about technology, our business and the product we provided was as much more impacted by, you know, like AWS and cloud hosting as it was by innovations in AI, for example, technologies, kind of how we deliver solutions. And I view AI in that context.
Craig Basford
you know, we again, think about that roadmap. It's much more around like, you know, what are these segments of customers and what are the problems they have. The fact is now, though, that some problems that have been around for a really long time, that we've been like, man, we don't really know how to solve, so we're not going to bring that to the roadmap now we can solve.
Craig Basford
And so that's where I is coming to the forefront. It's these old problems that we can now look at in a different lens. And that's really become a focus. but again, I wouldn't say dominance. There's still a lot of areas in, in our space learning and development that wouldn't necessarily be made better, more by AI or maybe where it would be, it wouldn't be the first areas to go after, because I think people also get used to it.
Craig Basford
Sure. Like we, you know, where we work in this space, right? We we think about this all the time. But our end users, that's not always the case. Even when we surveyed our teams, like I had a conversation with the product and technology teams, what how many people were using generative AI tools to help them do their jobs?
Craig Basford
They all work in technology. Half of them are, you know, developers with degrees in the space. Less than half of them were using AI today already in some aspects of their job. And and so then you think about, you know, pushing that down to what our customers audiences look like, what their admins look like, their instructors or course authors.
Craig Basford
I think it's important that people get comfortable with the concepts. I don't want to go make our LMS like AI, everything. I don't want to go replace it all with AI. Yeah. so I would say in five years you'll definitely expect to see AI woven throughout the product. But I still at this point don't see that that's going to like replace.
Craig Basford
You know, that our whole roadmap will just be, you know, AI initiatives one through ten.
Al Kinnear
Yeah. Got it. I do think it's fascinating, though that that I will play a role, to assist you and others developing products, whether they're LMS or or CMS or whatever they are, in solving some of these long tail problems like return on investment. Yeah. That is a question that is asked universally to every SaaS vendor. Right. And do you see has I made any impact on that yet, you know, to your benefit.
Al Kinnear
Or have you seen other examples in the marketplace which might be true? I don't have a.
Craig Basford
Great example that's coming to mind immediately as like a direct like, hey, we could measure it. Now we have AI, now we can measure it. But there are, you know, maybe one step removed if we think about an area like what is the value of upskilling? For example, prior to AI, you would have assigned courses out. Maybe you would have built group learning paths, by job role or by seniority or something.
Craig Basford
It would have been not necessarily one size fits all, but still not personalized for the user. And so you might measure ROI by, okay, how many of my senior developers completed this or how many total courses did they take or something like that? Now, the version of that question today, you can build personalized learning paths with AI targeted to specific skill gaps to that learner.
Craig Basford
And again, specific career paths. And instead of answering how many courses did they complete, you could answer, how many skills did they get? And even more valuable, if you're talking about ROI. In that use case, you could be answering a question like how many job roles did we fill internally that we didn't have to fill externally? Sure that now that ROI can be measured at multiple levels in your organization, it saves money over recruiting and hiring costs.
Craig Basford
It helps improve employee retention, employee engagement, which most folks have. Other means of tying that back into their either top or bottom line. So I think that's a good example where these the question of ROI, where was trying to solve it in a generic way now can be answered more personalized and then rolled up into a number into a metric that can be measured.
Craig Basford
so that's what came to mind. First, remain, I have seen in other industries where I can help answer ROI much more directly. And you look at something like the customer success space, for example, or customer support, and you're trying to have, ticket avoidance, for example. So your measurement of success would be how many less tickets were created as a result of this content?
Craig Basford
While previously content was being generated by humans hard to build, hard to update, now generative AI can produce it off of your underlying artifacts that you built anyways, so you have more of this content. You can attribute ticket avoidance directly to those items. Immediate obvious ROI. Right now my support system avoids more support tickets. I'm saving money. I don't need as many support agents are more efficient, and so I think there's a lot of opportunities in many different industries for that.
Craig Basford
but for for alums right now? no, I can't think of anything immediately that would be like AI directly enabling ROI.
Al Kinnear
That's it had to be asked. I mean, yeah, we've been asked that question together a thousand times, if not 10,000 times about ROI. And yes, even in other SAS industries, I still see that being the, you know, a very popular question. And it's a it's a tough nut to crack. It is in terms of the inputs. So you're programing engineering, approach to building product.
Al Kinnear
Does I have any benefits there. I'm just curious. Yeah that on the programing side we.
Craig Basford
so this is one of the benefits of, the way the large. Okay. I'll keep this less technical. And the way that these AI models work now is they can be you can build specialized versions of them, specialized versions of them on specific domains. Okay. And so what we've seen is vendors like Microsoft bring out versions of their product that sit on top of their development tools, that enable developers to build products faster.
Craig Basford
so they have a product called Copilot, which we rolled out in our organization and is now available to our development team. And it does simple things. It might suggest make our recommendation as you're writing code for how you might want to finish it. It can also automatically generate things like test cases, which most developers will tell you is, a pain in the butt.
Craig Basford
for them. And so it helps save time in that regard. we're still in the process of measuring our own ROI of that implementation. But if you measure it subjectively, which we've done, just asking developers, for those who are highly adopted and those who are like really into it, it is having a significant positive impact on their efficiency.
Craig Basford
I would add an interesting note, though. Again, even in this very technological space, people who are live and breathe, you know, code all day, every day, and they think about all this stuff all the time. But not everyone is excited at the prospect or like immediately an arms wide open embracing it. So I do think that's fascinating.
Craig Basford
I wouldn't have guessed that. I would have assumed, you know, this is the space you live in. You know, your world is always being disrupted. That's what it means to be a developer. You know, most of your technology skills can be obsolete in, you know, five years or less in some of these cases. so it was interesting to me to see that AI is definitely being embraced by many, but not necessarily all.
Craig Basford
I do think I think folks will come around as it becomes more natural, and certainly as the new generation of developers get more used to this as they come up, it'll probably become, a requirement. Right? Right. I didn't see on those tools, but yeah, I'm really excited about the prospect. you know, compared to any other initiatives that you might spend money on or time on to try and deliver efficiency, it's pretty.
Craig Basford
It's pretty special when you find a technology or a tool that can enable this kind of efficiency at what is a relatively low cost. Yeah. for what it can deliver.
Al Kinnear
Well, you're talking about enablement here, right? You there's sales enablement. You're talking about enabling your. Yeah, Alex etc.. Yeah. Honestly that's exciting. Like we you know we talk about AI all the time, you know, in different conversations. But when you start thinking about getting efficient in your business and costs and, just efficiency of development and lead time like, yeah, it's it's really it's it's cool.
Al Kinnear
I don't know how else to say it. I agree. That's a great example. so one thing we did, Craig, throughout this season's podcast is we interviewed, many of our customers and we actually asked them about AI and we've got some really interesting clips. So I'd like to introduce a few of them to you. Sure, we can watch them and then we can just get some thoughts and spur some more conversation about, you know, product and, and just the impact and, and, and even the business world.
Al Kinnear
so I'd like to start Hamish Knox of Sandler, has a very interesting, perspective on, on the role of AI and human replacement. So let's watch that one. Now.
Craig Basford
You know, in terms of AI, there's, cliche going out there about how.
Al Kinnear
We're not going to get.
Craig Basford
Replaced by AI, we're going to get replaced by a human who's using AI. And my thought is, well, if every person is using AI, then nobody's getting replaced. Yeah. So this is an interesting concept and I love I love that Hamish brought this up. okay. I'll first say that I agree. I think that AI is going to and is already becoming in certain roles you like.
Craig Basford
I don't think you get hired as an SDR now and not have, for sure, you know, a tool, at least some knowledge. Yeah. but yes, being comfortable using AI, if you are in a knowledge worker role, I'll clarify that not all jobs are going to get impacted by AI, but if you're in a knowledge worker role, you definitely you got to get comfortable with this because I do not.
Craig Basford
I agree with Hamish. I don't think you're gonna get replaced by AI, you know, wholesale, right? But if you're up against someone else who could be five times more productive than you because they've invested the time to learn how I can specifically be tailored to their job. well, I mean, you're going to lose, right? Like that's you is not at 10% difference here.
Craig Basford
Some of that, some of the roles or this is having the biggest impact. It is huge, huge productivity increase is like total game changer. So I agree with him in that regard. I do, however, have to say when you look at technological advancements that have occurred or innovations, a lot of the big ones have caused job losses.
Craig Basford
Sometimes a job type, you know, goes away. Companies, sometimes entire industries. you know, I always think of phonebooks when I think of this, everyone probably has their own version of that product or that space in their mind that was like existential importance. And then somehow so quickly just disappeared. and so I don't want, I don't wanna make light of the fact, I think in a role like sales, where, you know, Hamish is coming from there is I don't I don't see that sales is one of those roles that I think is just going to get replaced, you know, by AI.
Craig Basford
But there will be in our industries at large and in our countries and all of these places, I do believe that there will be some that just cease to exist. as people work more and more to AI to solve some problems that it's really good at. And humans where maybe if.
Al Kinnear
We were really good at.
Craig Basford
Hired to compete with the kind of scale and cost you can get from AI.
Al Kinnear
Well, my feeling and you're going to love this because it's an opportunity for companies like absorb, is that we really need to train people on how to embrace and utilize a AI in their day to day, because you're talking about efficiencies. You've already mentioned one previously with your programmers, right? And if you're not embracing that, it's going to impact your efficiency when looking, you know, across the bell curve.
Al Kinnear
Yeah. Right. So I think there's a massive role for training actually in companies that want to utilize and benefit from AI. I think there's a huge role in training folks correctly on how to maybe not respect AI, but utilize it within their day to day appropriately have guardrails like there's there's a whole bunch of things you got to think about.
Craig Basford
Yeah, we we actually built a course for our employees, as part of our policy as well to help educate them about it. We've also seen a lot of vendors in the space in the last six months rapidly roll out content on it. I do think there's a lot of different kinds. You're seeing some larger, very big organizations decide that AI is risky to roll out at scale.
Craig Basford
And so a lot of their training is about the risks of AI, the dangers of it, the concerns around it. And for some organizations, I get it right. I mean, if you deal with, you know, life and death type of data situations, then risking a hallucination in your AI, like, that's that's a really big deal. but I think we're still I haven't yet, although I've seen some of them, but I haven't yet see it become mainstream about courses that are like, as a blank, you know, insert job role here.
Craig Basford
You know, how to use AI effectively. and so I think those will become more and more common. If you look at your LinkedIn feeds, you'll definitely see it's always sales. Who's moving first. You know, their course, their sales, they're always selling. So you see them like they're pitching, you know, hey, take my course I've got training courses on this.
Craig Basford
But when you look at some of the big vendors I it's still pretty early days. And so the kinds of courses that are out there to help teach people. But I would agree that this concept of upskilling has always been present and has become ever more important as innovation has become more rapid. This is one of those areas is a perfect case study for how you can take advantage of this opportunity with learning and activate your your workforce.
Craig Basford
Yeah.
Al Kinnear
Great thoughts, great thoughts. Yeah. So, Craig, Hamish continued, in our conversation about AI, to talk about the impact it's having on B2B sales and the buying process for sellers who are taking.
Craig Basford
Content that they.
Al Kinnear
Find through an.
Craig Basford
AI, whatever that might be, and.
Al Kinnear
Blasting it out.
Craig Basford
Well, God sent a messenger, and that's awesome. You and everybody else. so.
Al Kinnear
There's it's completely undifferentiated. And in fact, it's probably quite annoying for the buyers.
Craig Basford
Yeah. I don't know how much time everyone else spends on LinkedIn. I feel like I'm starting to spend an unhealthy amount there. But in the first three months after, you know, I came out, it was like half of my feed was folks who were in the outbound marketing or outbound sales space. They were talking about how ChatGPT, you know, they they came up with the latest hack, the latest prompt that would generate that most personalized, unique message.
Craig Basford
Right. And I saw it go from the LinkedIn feeds to my inbox. I, like many of us, received many outbound emails about different products and services people are trying to sell me, and I couldn't pinpoint the shift, but there's a.
Al Kinnear
Certain.
Craig Basford
flavor almost, if you will, to these AI generated emails and I, I don't know if I can put it into words, but I think once you see enough of them, you start to notice it. And it was fascinating to me because the whole purpose, the whole ROI behind using it, is that it would feel more personalized to me, you know, a prospect.
Craig Basford
And what actually one of happening is they converged more quickly than they ever would have prior to these models and current. They used to come in all kinds of different flavors. I get the, you know, my name in the line or you know, my ROI in the first sentence or all the other tactics. I tell you now, with AI, there's a certain sameness to all of them.
Craig Basford
Yeah. which is fascinating. And then back to Misha's point. You've lost the point of using it in the first place, which is being differentiate it. And I should clarify that there's always the risk of like, confirmation bias here, because maybe the good ones that are still coming through are also using AI, but they're using it well enough that I don't they're quote.
Craig Basford
Yeah, exactly. Exactly. They're using it in a way that is differentiate it. But anyway, I do think it's it's very amusing as we think about these large language models, many of them are trained on the same set of inputs and they're starting to run into problems. I say this as a layman. I don't know if these are real problems, but you hear about them running into problems or they're training now on AI generated data.
Craig Basford
Yeah. And I don't know what that looks like in two years, but my guess is that it looks there'll be an awful lot of, you know, convergence in these answers and all that. They're going to have the same tone and the same flavor. And, you know, God forbid, potentially the same misinformation could have been generated previously. So, yeah, that's, I totally agree with Hamish.
Craig Basford
You absolutely cannot rely on AI to differentiate yourselves now that everybody is using it.
Al Kinnear
Exactly. And I think so in the B2B model, like, you know, I run a sales operation and we as ChatGPT as kindling, kind of like the kindling for a fire, like it's building a playbook and you've got writer's block. There's a way to stimulate some thought. Now you're not going to take the output. And that's your playbook. That's not the intent.
Al Kinnear
But as I mentioned, it's kind of the kindling to the bonfire that gets you on your way. And I think there is use there. And, and you know, I'm fascinated by some software in the sales floor, really. Like there's a company called Gong, which I know you're familiar with because I'm sure, absorb is using it in their sales, floor.
Al Kinnear
But, you know, they're using AI to review interactions and in a really smart way, look at the buyer's journeys and be able to actually personalize those journeys by customer style. Right. Like so absorb sales folks would have, different conversations and say web TMA, right. But this AI is actually able to help folks who are trying to look in at both pipelines and decide where there's risk.
Al Kinnear
And I think it's a fascinating utilization of AI. It's not doing anyone's job for it. But a lot of times, pipeline management is identifying risk. And I think there is some, some sameness in in some of the risks that are out there. Once you've been through someone's pipeline for a while and listen to calls and you know, there's tag words and and that the other thing I like to think about in B2B and in sales and Hamish, of course, the sales guru himself, I think about flipping the script, like people are looking for a seller lists, buyer's journey.
Al Kinnear
But you know what's going to happen when buyers start using AI to benefit their purchase?
Craig Basford
Yeah. And their purchase.
Al Kinnear
Cycle. I think, you know, we are going to see, you know, things occur in that area in the next few years where we're going to have buyers who are well-educated on what they're buying, simply because they're using a tool that is taking advantage of the AI that's out there about the companies in their consideration sets. So, again, you know, salespeople may not want to hear that.
Al Kinnear
That sounds like risky, you know? But to me it's like, look where this technology is going to take us. So. Right. Yeah. Go ahead.
Craig Basford
And interestingly, when, shortly after ChatGPT came out, we, we couldn't attribute all possibilities that might come from. But we got our first lead lead from ChatGPT, within a matter of like weeks, I think of it sort of entering the mainstream. and actually the first reaction was kind of an oh crap moment, because part of the challenge is ChatGPT at the time was very dated, right.
Craig Basford
Who was trained on data. That was a year and a half out of date. And so when you ask it a question like, you know, what are the top five alarms, for example, you know, you as like if it's reporting outdated information, now you have this, this product that people are coming to and it's like it's not accurate.
Craig Basford
Right. now I would say, of course we ever did show up if you asked that question, who is a top five alums? But there were ways you could ask the question where we wouldn't show up, and there were lots of questions you could ask it that were wrong. based on like what even at the time was accurate.
Craig Basford
So it's, it's, very new area and I don't know where it'll shake out. but yeah, we're definitely seeing buyers use it. And this only the ones we know of where it's super obvious. Yes. There's like a link out to the observed website from it. I mean, who knows how many folks are using it, you know, write RFPs or help them draft requirements, help them review, information or a product capability.
Craig Basford
So like, it's it's a powerful tool for a lot of those areas.
Al Kinnear
And I think, I think the immediate result you.
Craig Basford
Would see.
Al Kinnear
From a buyer's journey perspective is shorter times to close, and not because of what your sales or my sales staff are doing, but because of what the buyers are doing. They're coming in more informed. Right. it it's going to be very interesting to see where this heads in the nail for yours for sure I agree. So Craig we we interviewed Devin hasty event as are Bush and fellow Canadian.
Craig Basford
Huge hockey fan.
Al Kinnear
he was the first guest on the podcast and it was just a great interview. I'd encourage any of our listeners to, go back and watch that, but Devin talked about the disruption of technology and education that could happen from AI. So let's watch this clip and, see what he has to say.
Speaker 3
If you put ten, ten year olds in a room with technology and ten adults in a room without technology and give them the same problem that the ten year olds will come up with a better solution to the problem than the adults, which is really interesting, right?
Craig Basford
I do I find it a fascinating thought experiment. there's a lot of studies on how groups work, which I think has an interesting dynamic to this, but if you just take the technology piece, you would have these kids, they'd have essentially the entire database of knowledge that, accumulated human knowledge at their fingertips. And we all have that, you know, with a cell phone in our pockets.
Craig Basford
the idea of being, like, walked off from it is kind of horrifying when you think about it. But these ten kids, they have all the knowledge. But they wouldn't have experience. they wouldn't have. They wouldn't necessarily know where to look. Although I would also say that my eye isn't either. Right, right. Yeah. They don't have that.
Craig Basford
And so I think that would be kind of their constraints. And then on the flip side, you have the ten adults who would have that experience. They'd have, you know, their own collective knowledge. They bring into the room. But that would be it, right? Anything or wasn't in their memories, it wasn't already in their minds they wouldn't have access to.
Craig Basford
And so if I had to, you know, I had to put a bet on this, I think I wouldn't bet on the kids specifically because of one concept in the thought experiment, which is the idea that you don't necessarily know what the problem is going to be. I would give the adults the edge if it was a problem that any one of them, or perhaps some of them had experience.
Craig Basford
Right. but if it's totally net new, like, you know, you're asking a group of software developers something about like a medical problem, I would give the edge to the kids because they would have access to tools to be able to teach themselves to go answer it. And I is solving a really generative I in particular, solving a really interesting problem, which is you don't even necessarily have to know where to look for your answer anymore.
Craig Basford
It can now index out for you in a very human way. You just have to know what questions to ask, which is not still the easiest thing, but I, I would take that bet. I would bet on the ten kids.
Al Kinnear
So I think, I think I, I agree with you but the the the reason I agree with you, I think is more to the point that the adults in their room having no access to data, right, would be plagued by recency bias. and so I think that, you know, when you've got access to data and, and, and you know, the information that the, the other group would have, I just it's it's to me they've they've got the advantage.
Al Kinnear
So I mean so you're.
Craig Basford
Describing this in a case where the adults have experience in the problem, but it's. Yeah. Right. Recent you know they've been on in-depth knowledge. They have you know.
Al Kinnear
So yeah I might not. It's just such a, you know, it could be a subjective problem or something that, you know, if assuming the adults are, let's say average age 35, we've all experienced a lot of things in our 35 years. but I find when I'm speaking to folks at work and outside of work, sometimes without data, like recency bias is a is it's an issue.
Al Kinnear
Yeah. And so it's just my own thought.
Craig Basford
But yeah. You know just expanding on this a little bit. We see this in learning in organizations all the time. And I think about employee onboarding is in particular is a good example because it's this really condensed period of learning. If I were to kind of counterargument my own, you know, I'm saying the kids would come in. I see, particularly with new employees.
Craig Basford
And I think I will assume that my experience at absorb is relative to other companies. You can give them all the knowledge. You can give them these, you know, big pages of, here's where you can go find this information and here's easy ways to find it. You know, search capabilities, AI capabilities. But at the end of the day, when they want to learn something new that they're doing for the first time, nine times out of ten they want to go talk to someone else who's done it before.
Craig Basford
That is the primary question they ask. And like even having a simple buddy system or cohort learning approaches, which are much more popular these days, and learning, giving that access to a peer, it's like you could do that and not do all the other stuff and probably still be successful. Yeah, and I don't think you can do the opposite.
Craig Basford
I don't like you see this with remote work and the impacts of just having, you know, documentation and systems based onboarding versus, you know, a learning program that brings in instructor led training, a learning program that brings in peers and community and collaboration. that latter program, I think is much more successful. So in that context, I think I don't know if I'm changing my own mind here and arguing with yourself.
Craig Basford
It depends on the question or it turns on the problem. I agree, but I don't think experience can be, underestimated in that regard. And knowledge is not everything. anyway.
Al Kinnear
Interesting concept, right? Yes. Excellent. So, Craig, Devin Hastie had a really interesting, comment about the the ability for AI to support, improved data, insight and outcomes in learning and development. So let's watch this clip.
Speaker 3
From an LED perspective, we don't necessarily want people doing things they like to do or things that they may like to do. We want people to lean into things that they need to be better at their roles or their jobs. So I think the opportunity for AI into the future is to help us personalized learning journeys so that we're meeting the learners where they're at and that learners have the ability to lean into something that's going to help them a be better at their job, or b grow in their careers because they're leaning into their own personalized learning journey.
Craig Basford
Yeah, I couldn't agree more. one of the things that you learn, if you look at how adults learn versus how kids learn, is there's a lot of differences. and when you think about how you want to build an LMS, you really need to tap into the adult learning piece, because at least in my world, that absorb, we are not training kids.
Craig Basford
or at least, you know, I don't know if you consider college students kids still, but, you know, they're very relative group. and so one of the big differences is, is adults need to know why, they need to know why they are learning a thing. And they also typically want to know they want to know it close to when it will be applied.
Craig Basford
Right. So like teaching them an abstract concepts like, like a trivia topic if you will, is not going to align well with the way most adults learn. and the why. You have to be very specific about the why. You can't necessarily rely on external motivation like a grade or giving someone you know, even even points or a gamified model or dollars or anything like that little star.
Craig Basford
Yeah. What's what's most effective for adults generally is tapping into their own intrinsic motivation. So, where he's talking about this, personalized learning paths, it really creates the opportunity to tap in to that motivation already by building learning paths that are specific to you as a learner and aligned with your existing Y, which for most employees, it's pretty safe bet that they care about their career path.
Craig Basford
And if learning can help them get from A to B to C, help them along that career path, that's an extremely powerful motivation to tap into. And we're really starting to see platforms in the last couple of years. bring that to the table, and absorb itself actually been working on this now for a while. We're really excited to talk about it, in the next couple months here, but I couldn't agree more.
Craig Basford
I absolutely think that being able to personalize learning, which is only now being enabled by I, I should add, this is not a new concept. I'm sure ten years ago people wanted to personalize still, but you can only do that for your senior leaders because it was so expensive. So, now AI is kind of democratizing this, bringing it to the masses, and we're seeing it in platforms that are able to develop these personalized learning paths for learners.
Al Kinnear
So what is it about the the input of AI that allows the development of this personalization at such a rapid pace?
Craig Basford
it's about, the number of data points. So if you think about what you would need to generate a personalized learning path for you, I would need to know a couple things. I would need to know what skills you already have today or what competencies you already have today. And we then need to know what job you have today.
Craig Basford
I would also need to know what skills and competencies required for the job you have today. Then you have to know the job that you might be interested in, or that you would pick, or that your career would naturally evolve into the skills and competencies required for that. Then do a gap analysis on those two things by measuring where you are today, and then you have to map all of that to a library of content, which has to be broad enough that it covers not just your skills and competency gaps, but also, you know, a thousand other employees at the same time.
Craig Basford
And that mapping is itself quite challenging because learners typically also have preferences. So you might prefer short form versus long form or video versus, you know, simulations or otherwise. So all of those things, it's like technically possible to do as humans. And you did see it again in programs ten years ago, but it was just a huge scaling problem.
Craig Basford
And until I came to the table, it would have involved you might have had a thousand people in an HR team just trying to figure this out. And that would have been for a select group of your employees. Now I can take on each of these steps.
Al Kinnear
It's not,
Craig Basford
It's not a silver bullet, you know, it doesn't like immediately just make all this work. You need good data in any AI model that has bad data in. Well, this will be bad data out. So you still have to, you know, really work on the data and build up good pipelines and understand all those things and make sure you have a loop to learn from as well.
Craig Basford
I recommend you a course and you say this is way too easy for me. I need to know that so I don't recommend it to the next al. but yeah, AI is is totally revolutionized this area and is really enabling personalization in a way that was just never before possible.
Al Kinnear
That's great man, great insight. Thank you for that. Yeah. So Craig, in our episode with Josh Romanowski of, Ford EV, he speaks about the importance of quality and accuracy in AI and how we are checking for bias. So let's watch his clip and then, we'll talk about it.
Speaker 3
I think risks are what you're finding is AI is going to make people more lazy than they already are. And I think that's a dangerous thing from a what type of information and quality of information are we providing. So I think it's telling professionals don't put in kind of the parameters that say, how are we checking this thing for accuracy?
Speaker 3
How are we checking this, for bias?
Craig Basford
Okay. first off Josh, you know, he's he's definitely on the money when it comes to the risks that are present with AI. You know, a lot of these new technologies, people move to embrace them. And then the risks become apparent after they're embraced. And we kind of learn to live with them because technologies have become so, we're so dependent on them now that, like, there's no choice.
Craig Basford
AI is dangerous in, in a number of regards. The one I think Josh is referring to specifically is like, what if it generates a wrong answer? Yeah. What if it generates that answer in a way that propagates an existing bias? and that could lead to, discrimination. Like, these are these are real risks and I definitely worry about them.
Craig Basford
I think in the context of learning and development. So I would I would add some caveats to this. First, we're not an industry, generally speaking. that has to deal with like high stakes, situations where we do have some of those. If you think about training that maybe is has regulatory require requirements, or otherwise really sensitive areas, I think that would be an example where you would not want to apply AI, but when I talk to our customers and I hear the problems they have, one of the biggies would be around content creation.
Craig Basford
And again, how do you build up this, you know, knowledge base of this, this library of learning and that can help train your employees to train your customers, train your supply, train, train your vendors, all those areas. And the trade off right now is you can either have information and it's accurate and it's human vetted or you have nothing.
Craig Basford
Right. Like those are your two choices.
Al Kinnear
Then generative AI.
Craig Basford
Brings a third choice to the table. And it's not perfect. But if that third choice is, you can now generate more information and you can't guarantee its accuracy. And it may come with some risks. But if it is able to answer these questions, even 80% of the time, I think that for certain areas, particularly things like product training, internal processes, you know, new product launches, things like that, I would take that risk.
Craig Basford
I would take the risk that it might be 20% wrong in exchange for knowing that it exists. Right. and that you can get it out there, you can refine content over time. You can build feedback loops into your products, into your courses to improve the accuracy over time. but you can't iterate on nothing. So I do think generative AI definitely has a space.
Craig Basford
it is not without risks. I, I completely agree, but for a lot of areas which we work in and I'm, I'm just going to totally stay away from everyone else's space because I don't have to worry about that problem other than as a citizen of the world. But for my area now, I think the risk is, worth it.
Craig Basford
for, for most of the areas and where we're talking about in this example, again, content, you know, content creation.
Al Kinnear
Yeah, sure. I mean, to me it's like, well, how do you fact check A.I.. Well yeah. Right. And so again, I'm going to go back to my comment earlier. It's about training. Like what are the guardrails you're going to employ. If you are going to embrace AI in your organization. Right. Just make sure you've got a clear set.
Al Kinnear
They're concise and they're easy to understand. And you know they have to be somewhat pragmatic as well. So. Right. you're identification of that risk and reward, I think, would be important in creating those guardrails because as you know, as I just said, you can't fact check AI. It's, you know, I will.
Craig Basford
I will say on that point. So we're absorb is going to be launching generative AI capabilities for a great tool. in the very like next month and a half or so here. And one of the ways we're trying to solve for this is bringing in we're taking advantage of the existing reviewer concept we already have. So if you think about you have a human build a course, most courses that get produced today don't get produced by one person.
Craig Basford
AI is not like one person who has all the possible knowledge you want, and often if they do, they're no good at making courses or they're not interested, or they're too busy or otherwise. So most course authoring tools, including our create offering, have reviewer functionality, collaboration capabilities, and it's used to review and fact check the human generated information.
Craig Basford
Today, I believe we can leverage that same concept still for AI generated content. so the model that we've built is the AI gets involved right up until it kind of comes in to create, and after that, now it's someone indistinguishable to content produced by human. And you can bring in those reviewer steps and get your subject matter experts to fact check this for you.
Craig Basford
it takes it from a model where, like, there's no involvement whatsoever by these folks to their some level of involvement. Sure. But I think if that is some you're concerned about, you're worried about accuracy. You know, you do want to go from, you know, 70, 80% to 90% or something like that. You can still bring in some of the exact same approaches we use today to validate and ensure content is accurate on AI generated content as well.
Al Kinnear
That's very well thought out. Yeah. That that's that's just great. Yeah. So finally, Craig, Brittany Warner of Sony Electronics, when we spoke to her about AI, she, she spoke about IP copyright plagiarism and really authenticity of content. Yeah. so let's watch her clip. And I'm really interested in your points and ideas, about this subject.
Speaker 3
Some colleagues have shared things like, you know, once you put something into like program like a chat GPT, they actually own that content because you're utilizing it in their system. So just thinking about, you know, intellectual property and things like that, and who owns that content, you know, anything that we create here, what we're proud of is it's our, you know, intellectual property here at Sony.
Speaker 3
But if you start working and interacting with, you know, AI programs, you know, who owns that. so I think about, you know, those as, you know, potential risks.
Craig Basford
So I think Brittany talks about two topics here, and I'd like to address them separately. The first one was this idea of when you're using one of these products like a chat, GPT or otherwise, that the information you send to it is now owned by them. That actually depends on how you use the models or what version of them you use.
Craig Basford
So the open to the public free version of ChatGPT, or even the paid version of ChatGPT. That is absolutely true. The information you send back to it part of your terms and services that you put into it, it's theirs now, and they're learning from that information, and they use it to make their learning large language model better. but they're not the only that's not the only way to interface with what they built.
Craig Basford
So, at absorb, actually, our generative AI capabilities sit on top of a different version of the same large language model. But in ours, we are the information we send does not go back to the vendor. So it stays. we essentially send a prompt over, it responds to that prompt, and it retains that prompt for a very short period of time, usually about 30 minutes or so, I think, ours is hosted with Microsoft.
Craig Basford
There's different versions out there, and they mostly retain it for quality assurance purposes. It does not go back to train the large language model. So that's really key. And you'll see this depending on how organizations have implemented this commercially. this not it's not a guarantee that they're like taking the information. Definitely something you should check on.
Craig Basford
And,
Al Kinnear
And be selective in your choice.
Craig Basford
Yes, absolutely. Because, I mean, I think it'll be very difficult once these tools are in place to then after the fact, try and put limits on how people will use them. I think it's much safer to make sure your system is architected with that safety built in, knowing that if someone does upload something they shouldn't have, wow, it's not going to go anywhere anyway, right?
Craig Basford
So that's the first point. And the second point about what? About the actual like, legality of how these, large language models were trained. And that is an unanswered question. you've seen things now. So vendors like Microsoft, for example, and I think I'll open AI has done this as well, are offering what basically amounts to indemnification. They say if you use our tools and you get sued for any reason because of it, we will cover it, which, is a pretty bold thing to say for something that has really not been decided in, you know, courts.
Craig Basford
there is no answer yet as to how it will work. One argument is, look, these models are no different than if you had a human read 500 books, or if you had someone go out and read 500 web pages and then wrote an answer, you know, in that context you can. Interesting. Take coffee. Right. This kind of understanding, the flip side is these companies are now generating massive commercial returns, right.
Craig Basford
The valuation companies on and and yeah. And then the large language models like you know, is that an original work. and again, not a lawyer. I don't have, a legal opinion to share on it. All I can say is at this point is it's not decided. If you look, though, at previous technological innovations, once the cat's out of the bag, it usually doesn't come back.
Craig Basford
Now, if you think about the, Napster case, for example, you know, it's not like they made it illegal. And then CD sales, you know, were booming again. Right? instead it led to things like iTunes existing and which is in many ways a very similar concept to what Napster brought in terms of ease of use and how you can get access to this information.
Craig Basford
There was a cost associated with it, and they had to ensure they were paying the rights holders. But the concept was there. So my if I had to make a bet, I would say that these concepts are here to stay and if something will change through the courts, it'll be how, how those rights holders are compensated through it.
Craig Basford
so something other than zero. but I don't know if it's going to get crazy either. I mean, you look at Spotify as one of these examples of companies who is democratizing this and they're, you know, doing crazy amounts, but it's a running joke among artists. The amount you get paid. I saw some the other day, you know, for like a billion Spotify views, you might getting getting paid tens of thousands of dollars.
Craig Basford
So maybe they'll get paid out of it, maybe they'll win the court cases. But they might also turn out that, you know, what is the value of one sentence in one book by one author? Maybe that's you get 0001 sense. Yeah, I know.
Al Kinnear
It's interesting because this this is the hot topic of.
Craig Basford
AI. Yeah. And it's being talked about.
Al Kinnear
By book publishers. By music publishers. in learning, in development, I mean, people are concerned about IP and copyright, and I don't think there's a clear answer. Yeah. And I think it's going to be pretty interesting to watch what courts do decide. and again, we're neither of us are legal experts. But you know, when you're feeding 200,000 published books into an engine to teach it.
Al Kinnear
Yeah. You got to wonder what the ramifications for that will be long term. And, you know, these folks that make their living through copyright, you know, definitely we've seen an impact in, in the music world, as you alluded to. Yeah. but it's going to be interesting to see AI's impact across all rights like artists and, and music artists.
Al Kinnear
But like, you know, people, people who just, you know, make their living based on their, their IP. It right. It could be tough for them.
Craig Basford
And books is tricky because they're somewhat, you know, analog, if you will. We've seen now some major publishers of web content, figure out how to try and monetize their data places like StackOverflow, New York Times, Reddit. They all have either implemented ways to try and protect themselves from having their data be scraped, or are monetizing APIs to try and have access to the data.
Craig Basford
So I do think that those kinds of companies, those people will figure out sooner rather than later how to get compensated for what's on their platform, but that leaves an awful lot of people out in the cold. And as you said, you you think of like an author or, you know, individual musician, your ability to get monetized independently is incredibly small.
Craig Basford
Yeah. and so I do think about, especially in AI.
Al Kinnear
If your music influenced an AI engine to help write new music, it will be cloaked and and literally invisible. Yeah.
Craig Basford
So yeah, that's a.
Al Kinnear
Topic you brought up, but, yeah, very, very interesting one. Craig, I have to thank you for joining me today. This has been one of the most, enjoyable episodes this year and, couldn't have done it without you, man. So thanks so much for.
Craig Basford
Thank you for having me on. It's been a pleasure.
