Marc Zao-Sanders is the CEO of Filtered (Artificial Intelligence), a leading artificial intelligence learning platform. After graduating from Oxford in Maths & Philosophy, Marc started his career in strategy at Mars & Co and founded Pure Potential in 2005 and the online charity accessprofessions.com in 2010.
Broadcasting live from Business RadioX Studios in Atlanta, Georgia, it’s time for Learning Insight, featuring learning professionals improving performance to drive business results.
: Welcome to another exciting and informative edition of Learning Insights. We are delighted to announce that today’s interview with Mark Zao-Sanders, Founder and CEO of Filtered, is sponsored by Training Pros and broadcast live from the Common Vines and Taste Wine Bar in Boston, where we also want to thank Jennifer Scott, the owner and general manager, for allowing us to have this wonderful event at her beautiful restaurant and function facility right in the heart of Boston. Hello, Marc. How are you?
: Yeah, I’m very well. Hi, guys. Good afternoon. It is a very nice restaurant here.
: Well, Marc, can you share a little bit about Filtered. How are you serving folks?
: Yes, cool. So, Filtered is called Filtered because we are about putting in front of learners just the stuff that they need to see, that they need to learn. So, we use some AI to strip out any material that’s unlikely to be useful for them, just make recommendations, which are relevant to them and their role. So, that’s that in a nutshell. It’s the opposite of the sheep dip approach to learning whereby everyone goes through the same thing.
: So, now, you’re leveraging some artificial intelligence in your learning platform. Is that right?
: Leveraging, yeah. I mean, we’re creating and we’re booting it first. And we’re booting it because we see a problem that we think that can be solved or partly solved by AI. AI could resolve all of it. The problem that we see is a potentially soluble or part soluble by AI is getting those recommendations to learners.
: And so, if we understand the learner well enough, we use a chatbot to get information about the learner and find out about them. If we understand the learning well enough, and as we look at the material in the learning, and we adapt the algorithms to entire data, then we can match the two up with some prioritization within.
: It’s really much like the Spotify algorithm, or the Netflix, or Facebook, or LinkedIn, or Twitter, or any of these guys. They are trying to match up with individuals the content from how that … you know, from all the different contents listing there are, the most relevant material, and we take that with corporate learning.
: So, now, when you’re communicating that to your clients, are they embracing this, or are you, right now, at the stage where the most forward-thinking people are the ones that are dipping their toe into this world?
: Yeah. Well, the clients are embracing it because they bought it. And with some of the other guys that we’re targeting, yeah, we are looking at the most progressive individuals, but also the organizations that are in a position to be able to embrace, you know, or experiment with newer technologies, find newer solutions. And that isn’t everyone just yet. We are at the start of this journey.
: I mean, in learning, it’s not like maybe you can sort it all that day throughout. A lot of companies are still using technologies like XCore that are a couple of decades old. So, it will be a journey for a lot of us to get into AI whether you’re doing it with Filtered or not. And, yeah, for now, you have to rely with looking to the guys that are in a position to look ahead a little bit more.
: Now, is your background from learning and training, or is it more from AI?
: Neither really. My background is … What’s my background? My background, I started my career anyway in strategy consulting. And in strategy, as you may know, there’s not a spreadsheet thing and analysis. The insight that I had when I started my career was just the training involved to get me through then that could be useful for the firm was relatively limited. So, if you could find a way to get the most relevant learning and present that to a learner, you’d really significantly uplift productivity for the company and a sense of for fulfillment for the individual.
: So, we came in. I mean, I came in and with the other co-founders from that angle. The problem here, we weren’t thinking in terms of terminology like learning, or development, or school, or LMSs, or anything like that. I mean, we had to develop our knowledge in those areas. But it was … Yeah, we came in with a typical set of entrepreneur roots of, “Yeah, there’s a problem here. We think that we can do something about it. Let’s make that happen.” And then, of course, we got to know all these wonderful things like school, LMSs, and the virtual learning environment more heavily.
: So, now, can you share some best practices?
: I’m sorry. Say that again. Share some?
: Can you share some best practices if a company is going to kind of dabble in AI?
: Yes, of course. Well, one thing … Okay, I’ll share a couple of things. One is to be sure that you’re ready for an AI journey. And that means partly that the culture, and, you know, having enough of a groundswell of influences at the company that the top in AI solution is going to be attracted.
: Another part of that is having the infrastructure to do it. And I’m not talking about just money because, actually, to make an initial throw-in into AI, it’s not necessarily expensive. It’s potentially expensive. But what you do need to is have the data in order to be able to feed the algorithm. If you don’t have the data, then algorithm is paralyzed, and they can’t do anything. It’s going to be particularly special for you. So, make sure that your company is well inclined for this move both from a people perspective, and also from a from a data perspective. Those two things I have done.
Actually, I mean, another one would be they’re really critical of those organizations that are trying to sell an AI solution. They will often be able to point a marketing literature. But first, we’ll be sure that the solution that they bring is that they solve … that they are … the problem that they’re solving with AI really is a problem that needs AI. Maybe a manual job will do. Maybe an automation job will do. Maybe a semi-automated job will do instead.
: Also, find out whether the AI is owned by the company, and make sure that it’s real AI. So, what I mean by that is if it’s AI and if it’s machine-learning, then the system will get better at a certain task, performing whatever the task is over time. And if it’s not actually going on, then you’re perfectly within your rights to ask that vendor, “Well, how much better did they get over time? And can you quantify that?” The best vendors will be able to do that.
: Now, when you say-
: That should be there.
: When you say, “Can you quantify that?”, what are some metrics to quantify?
: For example, the AI thing I’m using AI for. So, one of the … I mean, for example, one of the things that we use AI for is the text learning asset. So, this means looking at a given learning asset, looking at some of the data that we do have for it, like type and description, and then assigning text to that asset, so that we can make recommendations to individual.
: Now, that’s the task. With data, AI should get better at that task over time. And you conclude that if you have some sort of measure of what good performance is. So, that might be maybe a scorecard versus a human user. That might be how many times you’re getting it correct according to some definition. As long as you have that, and then you have some sort of baseline, and then, with data, you know, should be improving it. And maybe it’s something small, but the improvement should be there over time. So, that is the whole point of machine learning. You’re creating a task over time with data, working on and under the experience.
: And, actually, that’s what human beings do. They get better at a stuff over time because, you know, we’ve evolved to be adaptive. So, you are natural in your networks. Obviously, our brain adapts and improves how we respond to the world, to the environment. And you want a human machine learning to do that for you.
: Now, what is a reasonable amount of time to implement an AI program platform?
: Okay. Well, it depends again on what you’re asking AI to do. Actually, it could be a project that might take a number of years if it’s super complex and involves a lot of people, different countries, different departments, and what have you.
: But a lot of … For a lot of companies, they just want to make a start, and maybe even learn themselves that they’re closer to AI. There are stations that come out of the box from, you know, from vendors. There will be things it will do with you, or with some of the tasks, some of the … bring some of the benefits that I described earlier.
: And don’t forget that we’re all using AI all the time anyway. I mean, every time you go on to YouTube, or use Spotify, or use Twitter, you’re feeding a series of algorithm data to make better recommendations, provide a better learning or better experience for you in the future.
: So, I’m sorry. To come back to question on how long, it might be within a few weeks if it’s an out-of-the-box solution. I mean, you know, what we provide, where there’s minimal customization, that can be done in a few weeks.
: Now, let’s-
: It should be that quick.
: Let’s talk about Filtered specifically. What is kind of that pain point that your customer is having where they go, “You know what, we should call those folks over at Filtered”?
: Okay. Well, it’s … Of course, it’s a range of things. They sure are talking differently. But, essentially, it’s that there is so much content that our clients’ staff have access to. And that comes from libraries that they bought for their staff. It comes from materials that they’ve created themselves in their proprietary material. It also comes from, you know, the internet, the world wide web. So, there’s just so much material. There’s an absolutely monstrous content overwhelm, and that’s getting worse if it were by the day.
: And at the same time, there will be skills gaps. It’s not like they have a workforce that goes around. They’re optimally scaled up. So, there are challenges that lie ahead. So, the training and the learning is there. The problem that companies report, and the ones that they feel the vast need, and the problems they see, getting back to that, is how to get better utilization. I mean, there are, ultimately, better productivity and sense of fulfillment from their staff using the training that they have invested in, in most cases, over a number of years. There’s content overwhelm will be how I would summarize that.
: Now, since you’ve been doing this, have you had some stories you can share people that have tangibly benefited from your platform?
: Yeah. There’s probably. Yeah, we’ve been around almost 10 years. And with the new product, it’s been … We launched it in the start of last year. We got a consultancy that has 100% utilization of learning materials in the professional services firm. They’ve found literally thousands of hours. And for guys like that, that’s a huge cost saving, hundreds of thousands of dollars or pounds. But, also, particularly from work with consumers directly since our inception, we’ve seen a lot of nice stories about learners getting learning that they wouldn’t have otherwise giving themselves time and hearing about that directly from them.
: And the stats, one that I particularly like is that you’re learning the right stuff. And you’re … So, intelligent learning recommendation that’s coming to the individual, and that’s being seen by that individual. We see improvements in proficiency of 5% per hour. Now, 5% per hour of a skill that you use daily has a huge cumulative effect at the course of a career, which will drop tens or hundreds of thousands of dollars.
: So, the company will feel, then, that the individual benefit from as well. So, we see that … We see big changes, improvement in productivity, and measurable effect of that belonging to big numbers. I mean, it’s just, in a sense, a nice and lovely company be involved in things seeing these benefits for both individuals and companies.
: Now when you’re working with some companies that have multi-generations in the companies, is this something that the millennial workforce is adopting quickly, and the older people are kind of slower to adopt, or are you finding that everybody is embracing this?
: Yeah, interesting question. I don’t really see a huge difference between the millennials, to use that term, and everyone else. I know what we can’t deny is that … Well, for one thing, millennials aren’t just in one block. No offense. They’re not even a very well-defined group of sub-population anyway. But whoever they are, that age range, admits that it created itself, but there’s all sorts of variation within that block,
: And then, the other side of millennials as well, we who are way up 38, I’m not a millennial, but my own technological understanding has improved significantly through the last 20 years. I think it’s more that the workforce genuinely has enjoyed … It’s not the right term, but, anyway, benefited from and responded to changes in technology over the last few years. And that’s the incremental thing.
: I don’t think that the divide between millennials and the rest. It’s helpful to us in forming our statute. It’s more thinking about all the workforce is changing. These technological changes mean that there are going to be more of them, remote workers. Okay. So, how do you tackle that? Attention span is maybe shorter. Consumer base, software and services that people enjoy have an implication for how learning needs to be provided in the corporate environment. These are all changes that affect everyone. I think that’s a more important perspective than just the age cutoff.
: Now, as technology advances and the speed of computing power increases, are you seeing … are there things outside of artificial intelligence? Like what are some of the things that are on your road map to really leverage all of that?
: Well, I mean, first, it’s really, like I said at the time, it’s not just about artificial intelligence. We see AI as a means to an end. The end is to make useful, relevant recommendations.
: And to give an example of one other method that’s really important with that is curation. So, if you can make the best recommendations with the most populous algorithm, but if that … If they’re only going to ever be drawn from a pool of content that isn’t high quality or relevant for that workforce, the best the algorithm can do is not going to be very good.
: So, that sort of curation at the start, which can be algorithmically enabled, or enhanced, or augmented, we’re still at a point with AI where humans need to be in the loop. So, with curation, that final decision about whether or not, say, it’s going to be relevant to this population, probably the decision is best made by human being still.
: So, for us, yes, AI is going to be part of the future, but it’s not going to be a pure AI future any time in the next 50 years, I would say and probably not in the next 10. So, yeah, there are other facets to our solution, which are non-AI. I think you asked about advice, like on this initially, that are just as important. And in certain situations, that’d be more important than AI.
: If you asked me about the technologies that I think are going to be important and influential, well, I think possibly more even than AI is, first, a well-chosen automation. So, there’ll be more and more tasks that require or that can be automated. They don’t necessarily need artificial intelligence. They didn’t necessarily need to get better and better at those tasks over time but picking which those are and getting the benefit to be felt and enjoyed by a human being is going to continue to be a really important part of business. And I think that’s it. That’s certain over the next decade.
: And the other thing is immersive technologies, VR, and AI, AR. I share the predictions for those markets. They’re just as bullish as they are for AI.
: Now, for you, as a company, how difficult is it for you to have trained people to work there?
: Change it’s hard. It is hard because if you’re selling something sophisticated, and that needs to be relayed internally and then externally. And in some cases, you’re working with partners that need to tell the story themselves. I mean, the tough stuff that were led externally for them to try the relay, it is hard.
: So, that means to the hiring process, which is still hugely an entirely human process, is all the more important. And so, it’s timing. It’s just training. And then, it’s that ongoing support, coaching, and learning. I mean, it comes back to learning for everyone that works at the company and me included.
: I spent most of my knowledge of AI, which is still naissance, but I really enjoyed the last two and a half years with my job as CEO of Filtered. You bring all these on that journey with me. And I think we’ve got that culture. So, if you have that as a company, then, yeah, with any company, if you were … there’s some …it’s easier than for self-directed learning, and most of the time in careers. And a lot can come from that. It’s hard to put it, but AI is going to good.
: Well, Mark Zao-Sanders, Founder and CEO of Filtrate. It has been an absolute delight having you on the show this afternoon. I won’t keep you from the wine bar, but you have to earn your tip a little bit. You’re going to go to a talk here and just a little bit, aren’t you?
: Uh-huh. Yeah, another talk. Yeah. It’s been a real delight for me as well. Thanks so much, guys.
: Absolutely. Our pleasure. Okay. Until next time, this is Stone Payton for Lee Kantor. Our good friends at Training Pros, our guest today, Marc Zao-Sanders, and everyone here at the Business RadioX family saying we’ll see you next time on Learning Insights.