Feb. 22, 2021

Unlocking the Value of Data, Analytics & AI

Unlocking the Value of Data, Analytics & AI

Bestselling author and data science thought leader, Mike Bugembe, talks getting value out of tech, building a PC with his son, GPT-3, education's silent disruption, and more

Apple Podcasts podcast player badge
Spotify podcast player badge
Google Podcasts podcast player badge
Amazon Music podcast player badge
Overcast podcast player badge
Castro podcast player badge
Stitcher podcast player badge
PocketCasts podcast player badge
TuneIn podcast player badge
Deezer podcast player badge
RadioPublic podcast player badge
YouTube podcast player badge
RSS Feed podcast player badge

Are you fascinated by data's potential to transform society? Then you won't want to miss this episode featuring Mike Bugembe, author of "Cracking the Data Code: Unlock the Hidden Value of Data for Your Organisation."

Mike and Ryan discuss how AI is shaping our world today and how it may change life for future generations.

Topics include:

  • Mike's career path
  • Mike's book and writing process
  • Building a PC (featuring Mike's son)
  • AI written code & GPT-3
  • Virtual learning & education's silent disruption

Meet Our Guest
Mike Bugembe is a bestselling author, international speaker, and thought leader in the world of data, analytics and AI. Mike has worked with organisations across a range of industries. His talks help leaders understand how they can generate real business value with AI and its family of technologies. Mike's research on the interplay between data and behavioural economics got him recognition as one of the UK's Top Digital Masters. He has won several awards and is a regular in the list of the most influential people in a data. Mike is the author of the bestselling book, “Cracking the Data Code”, and it has been used a blueprint on data for many businesses. He has a passion for using AI for good and is working on AI projects to address abuse, prejudice and child school exclusions. His team at JustGiving.com built a suite of algorithms that generated millions for charity and resulted in the eventual sale of JustGiving.com for over $100million in 2017.

Mike's book

Caleb and his rig
Caleb's excel sheet
GTP3 codes language

Show Links
Click here to join the Slack Workspace
Click here for the episode transcript

Follow us on Twitter: @thedwwpodcast

Email us: podcast@digitalworkspace.works

Visit us: www.digitalworkspace.works 

Subscribe to the podcast: click here
YouTube channel: click here


Ryan Purvis  0:00  
Hello and welcome to the digital workspace works podcast. I'm Ryan Purvis, your host supported by producer Heather Bicknell. In this series, you'll hear stories and opinions from experts in the field story from the frontlines, the problems they face and how they solve them. The years they're focused on from technology, people and processes to the approaches they took, they'll help you to get to the scripts with a digital workspace inner workings.

Welcome to the podcast. Do you want to give us a brief introduction to who you are?

Mike Bugembe  0:36  
Yes, yeah, I'm Mike, I'm passionate about the data analytics machine learning space. And actually, more specifically, how that interplays with human behavioral economics, I write a lot about the subject. I'm a Forbes columnist, and also a best selling author, professionally is my career began with Accenture, moved on to De Beers, the diamond miners where I helped establish their business intelligence function, a brief stint at Expedia. And then lastly, justgiving.com, where we use the suite of algorithms to help grow the world of giving, we were making 100 million 150 million a year in donations and then sold the company when we were making 400 million a year. And analytics data and machine learning had a really big part in growing the organization. Right now I'm the founder of a startup that's looking to disrupt that whole world of business intelligence. We've just completed our seed round and closed it literally, yesterday. And thank you and are hoping to launch in the next few months. So yeah, exciting times.

Ryan Purvis  1:48  
Cool. What's the name of the startup? I don't think I got that. Or can you not say it?

Mike Bugembe  1:51  

Ryan Purvis  1:53  
Oh, decidable.

Mike Bugembe  1:54  
Yeah. And essentially, if business intelligence and data is to help with human decision making, we want to make that world significantly more desirable than it is at the moment. So yeah,

Ryan Purvis  2:14  
I must admit, is one of the areas I've had huge, huge problems, but it's always a challenging space to take data and make it visible or viewable by someone. And ego isn't going to try works at almost lowest common denominator of the person that's going to be looking at it. Some people want a rich experience. And some people just want a red dot that says this is a problem.

Mike Bugembe  2:35  
Yeah, exactly. Some people just want the answer there. Yeah. Yeah. Yeah.

Ryan Purvis  2:41  
Great stuff. So we're talking about fiber and all the rest of it, I was gonna tell you, my, we just bought a house in West Sussex and everywhere around us has got fiber, except for our little area. No, and you can't find a reason why. It's almost like they deliberately ignored as adularia. So if someone didn't pay something,

Mike Bugembe  3:09  
is it like you're in a little island that doesn't want it or you just like on the edge.

Ryan Purvis  3:15  
So if I walk, you know, if I walk down the street, I can see all the cables being laid everywhere else except for our road. And I mean, our road. So there's a long road has been around. I mean, that's been a regional road of ocean. And then we've got a little branch offered, which is our road, but that whole long road doesn't have fiber. So it's just that long road and our road is attached to it. So I don't know why it was missed. I don't know if there was a real reason or just a planning problem, or what, that there's no, there's no end in sight to when we'll actually have it in the road. Yeah, let's

Mike Bugembe  3:48  
hope you don't have one of those neighbors that's caused the problem.

Ryan Purvis  3:52  
You probably find it is something silly like that. Because I mean, I have learned that new incidents as they come to your door, and they go no further. It's one of those things, they come to that point they like not going past it.

Unknown Speaker  4:05  

Ryan Purvis  4:06  
Great. So so I think just to give everyone context, so we spoke because I saw posts that you put on LinkedIn with your son put putting together an Excel spreadsheet and build his computer. And I really got that and I wanted to explore that. But that was one of the reasons I wanted to chat to you that obviously I read your book as well, which was another thing that's that's worth chatting about. Do you wanna give some some some some continuance on the story with your son and his computer?

Mike Bugembe  4:30  
Yes, absolutely. And he shouldn't. He shouldn't be. He shouldn't be with us shortly. Actually, he's, I think he was. He was mid game. So apparently, that's more important. But yeah, I think so. Let me tell you a bit about the book before before he joined. That's okay, if we do the round. So the book actually took me about three years to write and that's because I'm not Author English perhaps I'm a little dyslexic. And English isn't my speciality if you like, I'm more on the mathematic side of things. And it began as an angry letter, and then sort of evolved into a book. And the frustration grew from organizations that were investing quite heavily in this world of data, machine learning and AI, but failing to see any return on investment. And then on the back of that, making it a technical reason as to why they couldn't get it to work. And anybody who's been in that space, particularly now, with all the advancements of the tools and the technology, we know that it's more to do with the people. And and the organizational elements that are required to get that, right. So that's where the frustration came from. And it eventually grew into a formula, if you like, for what businesses and organizations should look at a blueprint, let's call it of things they should do to make sure they at least tick the right boxes to make sure that data and analytics actually delivers value for them. So So yeah, that's the journey, if you like on the book,

Ryan Purvis  6:13  
what was it? Yeah, it resonated so well with me, because I mean, I basically highlighted most of the book, I think it is, it is probably less and highlighted sections of the highlighted sections. And it's probably because of the exact same similar frustrations. Yeah, you know, getting involved in a project and just seeing that it's not gonna work because of the people and the process being being just, you know, the technology is usually not the problem. Yeah. Because, especially if you know what I'm talking about 10 years ago, what we have now compared to what we had 10 years ago, yeah, there's no excuse for the technology to not be absolutely suitable.

Mike Bugembe  6:48  
Absolutely. And we're talking about challenges, like, Do you even know what you want to do with them? Right. It's all good well, and good having and buying it and purchasing it and getting a team in place, but for to one end, right. And then to put that responsibility on a technical team whose domain expertise is to deliver technology, not to understand your business, to the extent that they can identify all of those opportunities, and you need the balance, right? whole thing of it's up to you to work it out. And tech, love doing tech, we don't enjoy reading, understanding the business to the degree that you do as a domain owner, to be able to then find holes and opportunities, and realizing that there is the balance, because I even found when it was left with the tech teams, the tech teams would be like, yeah, I'm happy to go in and find a use case. But they'd often choose a use case that satisfies the tech needs, if you like, you know, it's saying this is a really interesting, very technical project, that's what we're gonna do. Whereas this low hanging fruit that may be less interesting on the tech side delivers huge value, you know, and so that balance is so key to get right.

Ryan Purvis  8:01  
Now you're spot on, I mean, I've got a problem I see often. And that that criteria, the sort of priority versus impact, sort of Cartesian plane, if you like, the tech guys always pick the ones that they can look the best at and get the best, you know, bells and whistles associated to it has no real business value at all. So I totally get that. And the behavioral economics piece. I mean, is that is that something that's been out of that?

Mike Bugembe  8:28  
Well, that also i can say is perhaps a historical, you know, generational one, almost because my granddad was a psychiatrist in sorry, we got a little joke that thinks it's reflection is another. And I've got my son here now. So I'll, I'll explain very briefly about the

Ryan Purvis  8:52  
how to beat you. I'm Ryan.

Mike Bugembe  8:57  
That's right. Yeah. So my granddad was actually Uganda's first psychiatrist. And, yeah, and my dad, being an economist with the UN, and the computer being left at home, it was a natural marriage of all of those disciplines coming coming together. And when obviously, you jump into the world of data, machine learning or at that time, it was artificial intelligence, or what Hollywood showed us. There's a human element to that, whether it's mimicking a human or interfering with the way we do life, and looking at it from both a philosophical and psychological perspective, came naturally because I was surrounded by that in my family. So yeah, yeah. Okay. Yeah,

Ryan Purvis  9:48  
that's good. I mean, it's something that you know, we use a lot. We build products, especially now if you read books like hooked on this other one, really game start or something. I don't remember the name of the title. But, but everything we do is is gamified. Right? Because we because humans are looking for a game that seems kick them into? Well,

Mike Bugembe  10:10  
there's a lot in that I think if you just begin to understand the way the brain functions, what is it we need? What is it, we crave hooked? There's a really great, you know, summary of a lot of BJ Fogg says, science around, you know, addictive technology or, you know, a technology that you can use to increase engagement significantly. And there there are, you know, I think a lot of good use cases, particularly from a public policy perspective, why that technology is really useful. And then now, if you watch the social dilemma on Netflix, you can understand the more sense of of jumping into that technology. Yeah, yeah, I

Ryan Purvis  10:50  
think that was a great movie. I think it really expressed a lot of basic concepts really well.

Mike Bugembe  10:55  
Yeah. Yeah, yeah. And then how, for example, we do need that ethical layer on technology. A lot of the time we talk about ethics, purely in artificial intelligence, but in technology as a whole, particularly as we are able to now build very, you know, increasingly persuasive technology. And we do need an ethical governing body of some sort, just to make sure we're getting all of that right. Yeah.

Ryan Purvis  11:20  
Yeah, yeah, sure. I can see that. The guests getting bored.

Mike Bugembe  11:28  
Yeah, so let me tell you about Caleb, and the story behind that. So why why did you want to build a PC? To

Caleb Bugembe  11:49  
a laptop, I thought I thought I thought I am. So I wanted a PC. And so I'm so dark came up with this great idea to build one instead of having to buy the whole thing. And then a couple days after that, all the boxes that we started buying the pieces and the money okay, we started building.

Mike Bugembe  12:14  
So who told you about? Who came up with which pieces needed and so forth? You did. How did you do that?

Caleb Bugembe  12:22  
That was a video. There was a video and there was lots of different pieces in the description of the video. So yeah.

Ryan Purvis  12:35  
Just give us a give us a bit of background yourself can be what your How old? that was at nine, I think. Nine. And when you started building a PC, you will still learn was that what you've just turned on? I can't remember. Okay, and did you build a machine yourself? Or did your dad do the?

Caleb Bugembe  12:55  
I guess

Ryan Purvis  12:58  
he held the torch for you while you screwed everything in. So what did you end up buying? What What did what was the things you were looking for?

Caleb Bugembe  13:24  
Inside the

graphics? Yeah. And then and then they are? Yeah, and then CP cola and CP CPU on the same day as the shaft?

Mike Bugembe  13:52  
Yeah. So he's very literal, like like me. So we asked him what order and he's never remembering it in sequence.

Ryan Purvis  14:00  
When I was about to say, I mean, this this is probably the one skill that every young kid should know how to do is to build a computer from scratch in order, because I'll tell you, the reason why I say choosing this chassis last is such a good thing because I've done it the other way around once or twice and you've ended up with us. There's no moment when you realize that the motherboard doesn't actually fit. And it changes the whole design

Mike Bugembe  14:27  
because I've never built a PC before. And it wasn't until Caleb sent me his spreadsheet of all the components that are required that I thought okay, we better get on with this. I planted the idea not thinking it will take off did a little research and then basically got us to the stage of building it. So yeah, you're right. We did do some fiddly bits one day when we were trying to get it in. Yeah,

Caleb Bugembe  14:54  
I symptom wise.

Mike Bugembe  14:55  
Yeah, yeah, the wiring was was pretty tricky, particularly within that chassis. We got it in the end.

Ryan Purvis  15:01  
Did you go with it without having seen it? Or you ever did you go with the gamers sort of box with colored lights and all that kind of stuff? Or did you go with like a corporate corporate box? Again, nice.

Caleb Bugembe  15:21  
Find friends.

Mike Bugembe  15:22  
Yeah. You got a really good rig. I think I yeah, I was suckered in by the spreadsheet. So

Ryan Purvis  15:35  
almost a blank check cuz he was looking at the spreadsheet.

Mike Bugembe  15:41  
I was more impressed by the spreadsheet than anything else.

Ryan Purvis  15:45  
You had screens? What did you do about screens? Did you go with one big one? Or did you go with two or?

Caleb Bugembe  15:50  
So I have one. There's one big one. It was my brother's before. But now now I've choked it my PC into into them. incident incident? incident?

Unknown Speaker  16:06  

Mike Bugembe  16:10  
So we had we had a monitor previously. And that was for his older brother who used it for a driving simulator. He's into karting. So so he has a simulator that he uses at home to practice the circuits. And we we borrowed that indefinitely.

Ryan Purvis  16:29  
Yes. Lots of families for to share. Exactly. Yeah. And then from a keyboard and mouse point of view, did you go with with any specific brand or type?

Caleb Bugembe  16:46  
We had we have this? Again, my brother brother had. And I used that for a while, but then a different different one. A new one?

Unknown Speaker  17:02  

Ryan Purvis  17:05  
So I think you I think it was a photo of your your setup rig. And you playing it so we can we can all experience your your design.

Mike Bugembe  17:15  
You're gonna go get some pictures for us, and then we'll send them to Ryan. Okay. All right. Brilliant.

Ryan Purvis  17:22  
Great. But I think it's, I hear a fish, our youngest guests. So you've got that as well to to go with your Excel spreadsheet.

You can go you can go.

All the best.

Mike Bugembe  17:44  
Yeah. So I have three, three boys done. It's interesting, you. They're all they're all reasonably technical, I would say he's the most technical out of all of them. But he's also the most. If ever, I was to predict what career he could go into, if if we were back, you know, 20 years ago, it would be that he's going into, he will be a developer. He just had that he already has a lot of the mannerisms, where he's happy to sit in front of a computer with pizza and hairbows. And just, you know, spend the whole night.

Ryan Purvis  18:19  
Stop talking to me, I'm busy, kind of

Mike Bugembe  18:23  
when I say that interesting, because when we're talking about tech, there's a big part of me that wonders whether we will need that many developers and whether AI will be writing a lot of the code for us.

Ryan Purvis  18:39  
Yeah, and you know, it's it's very much like the Uber problem, where you have a whole bunch of drivers that are driving cars that could be replaced by self driving vehicles. It's a possibility. But I wonder how if it will ever happen, because you still need that creativity. So yes, you might have I mean, I'd love it if the basic stuff was always taken care of, like, you know, building the login screens and building the you know, whatever they are, they're naturally easy components and let you just focus on that really piece of complicated functionality. I'd love that show. I could if you see it. Have you seen much in your work of coming up?

Mike Bugembe  19:18  
Yeah, well, I've seen I don't know if you've seen any in any of the examples of GPT, three that have been released. some fantastic examples. There are people basically saying, code builds me a social network. And straightaway it gets all the code all the front end, and you've got social network, you know, wow.

Ryan Purvis  19:41  
That's amazing.

Mike Bugembe  19:42  
I'll see if I can find that link and send it to someone and it's quite easy to build because because of GPT. Three was trained on I don't know how many billion elements it was essentially trained on the whole internet, right? It's got all imagine all of sakal flows inside an AI brain, you know. So it's, it's really advanced.

Ryan Purvis  20:07  
As you said, I was thinking, I can imagine it logging a problem on Stack Overflow and then solving the same problem. So the only thing I seen was GDP three was where they were they gave it a whole lot of articles to read. And then a wrote an article and people couldn't discern if there was a human article or an AI written article that's seen.

Mike Bugembe  20:30  
Yeah, no, they I think those people who managed to get early access before it was, I think Microsoft had the charging license for it now. But there were there were quite a few applications that were built that were genuinely fascinating. I mean, really, you were limited by your bounds on creativity, as you say, in terms of what you could get it to do, just imagine that you've been trained on pretty much the whole internet, you know, so. Yeah, in terms of a language model, it was on another

Ryan Purvis  21:00  
planet. That's a scary and amazing thought that that something exists already today that you could, in essence, tailor what you want out of out of it? Well,

Mike Bugembe  21:12  
yeah, it completely changes the game, right? Because if you remember that with with machine learning, the basic concept is we get a lot of data. And there's a challenge to getting the data because if it doesn't exist in a database, you have to start storing it in a database, then you have split that into your training set, and then your test set train the machine. And then you have all these overfitting problems, because your data only covers one small part of the universe. And then eventually it does it. Whereas now with with GPT, three, you can do short shot or zero shot learning, where you don't have a training set, you know, you just put the task together, but because it's been trained on the whole internet, there's so much that it can do.

Ryan Purvis  22:02  
So is that I mean, I haven't really looked into this in depth. So is gdb three hosted somewhere, and you're just you're just using the libraries.

Mike Bugembe  22:11  
Yeah. Something something to that, that that sort of setup, I think it's basically an API based access. From from what I understand, yeah.

Ryan Purvis  22:24  
That's pretty cool.

Unknown Speaker  22:25  
That's looking at,

Ryan Purvis  22:29  
yeah, cuz I mean, what you're saying now, in the sense of approaching and machine learning project is so true. And it's that getting unstructured stuff into structured formats. And, and the overhead not only have to skew the data, the right format, but trying to understand the context of what you've got. And what you're seeing more often than not you've got is such a challenge.

Mike Bugembe  22:52  
That's true. Yeah. It's that it's, it's, it's and I've just pulled out, I had the numbers on what it was trained on. So if Turing nlg, which was released in 2019, by Microsoft, was trained on 17 billion. Let's call it elements of data. Right? Yeah. gbd, three was 175 billion.

Ryan Purvis  23:17  
Wow. So that's amazing. That's an

Mike Bugembe  23:22  
order of magnitude bigger, you know,

Ryan Purvis  23:28  
out of interest in your new startup, or you're gonna use something like GDP three, or you got a different approach completely? Well, we've

Mike Bugembe  23:34  
got a slightly different approach, but it doesn't hurt. If we could get, you know, access to it. Why not? Because that's a lot of work done for us already. But you know, when I think one of the big advantages with GPT. Three is because it's been trained on so much so many variables in so many different, perhaps contexts, and languages and so forth. Like you said, it could write an article, right. And the, the, that's a pre trained language model that is more advanced than any other pre trained language model that exists. So machines that can begin to talk to you communicate with you, and so forth, you know, understand sentiment, understand context, understand all of those things. It's a, it's, I think it's a different world. You know, it's huge, and this stuff is happening much faster than we're, we're ready for. Yeah.

Ryan Purvis  24:31  
Yeah. When you mentioned that the sort of the technology and how that impacts thing, I think that's one of our challenges as almost as too many options. But also the technology. Change is so quick, that you just think you've got your head around something and then there's a new piece of technology that does it slightly differently, or you know, you've just been, it makes what you knew before redundant. Yeah. To see. I mean, when you look at a project now or Deal with customer? Do you have a preset of things you'd use to solve the problem? Or do you take it on the problem first, and then look at what tools you'd use?

Mike Bugembe  25:09  
And, yeah, so I would say it's a, it's looking at the problem first, because then because there's so many advancements that happens so quickly, like you said, you can jump into something, and then within a year, it's completely redundant, you know, and, and it doesn't work. But looking at the problem set is absolutely critical. And spending more time on that, because there was so many intricate details around, you know, particularly if you remember back in the day when we used to capture requirements for demo projects, non functional requirements as an example, right? And so doing the task right now is not enough, you need to get a better understanding of some function, for example, in the world of API is explainability, right? We live in a world where you can't just do something you have to explain a little bit more. Why, right? Why were you rejected for the mortgage? Why was this happening? You know, and why is it picking these kinds over those bonds? and that sort of thing. And if you have a requirement for high explainability, there's certain algorithms and approaches you just can't use, you know? And so getting the problem set really clear is absolutely key to the success of the project.

Ryan Purvis  26:24  
And how do you how do you lead someone through? You know, getting getting clear on what that problem is? Because often people will think the problem is actually is ABC. Meanwhile, when you actually asked like, you do the five why's technique, for example, you realize that the problem is actually something else completely, but it was just a symptom that you're actually thinking was the problem? Yeah.

Mike Bugembe  26:48  
So I think five why's is a tried and tested method to try and get to that. And the other approach is business goals. And the last one that we tend to do is I used to now this, this has all changed with COVID. But Previously, we used to use Lego. Because you get a bunch of executives in the room to play with Lego. Firstly, they're disarmed by the fact that hang on, this is supposed to be a serious session. And you brought me in with a bunch of toys, right? And a lot of them get upset. They're like, what is what on earth is this going to do? By the end of the session, they're all like, look at my amazing model, you know, and everything completely changes. But level, we use a process called Lego serious play, which I really am put together by I believe it was Lego themselves with a couple of professors from an MBA school somewhere where they you know, that the challenge they had was that they were saying they're a creative organization, how come their strategy meetings are so boring and ineffective. And this is at a time when Lego was being taken over by Nintendo and games and so forth. So Lego is really struggling to capture the attention of certain children. And, and they came up with Lego serious Play Doh, the science behind that is fascinating, because they also call it thinking with your hands. And the moment you start, you know, doing that you stimulate a whole bunch of extra quality, or those elements in the brain that stimulate creativity to a degree that you've never thought of before. Right. And so Lego serious play, I think is fantastic of that. But the one trick to get that to work even more effectively, is to put people in the room to increase what we call the diversity of thought. So put people in the room wouldn't normally be in the room to have that discussion. And on a couple of occasions, I remember facts were gathering requirements and trying to find, let's say the problem or a new way of looking at something. And it was the VA who was in there taking notes that came up with the answer. You know, and just, and this is a great book, I wish I could remember when they talk about innovation. Innovation tends to happen on the edges, not in the middle where all the contemplated knowledge is that happens on the edge. And it's about getting people from the edge into the room to have that conversation and participate because that's where innovation comes from.

Ryan Purvis  29:11  
It's funny it's this has come up twice now this week as a mechanism where you're trying to get one using Lego. The two you're trying to get people to get out of the the roles in our organization and share things using sort of indirect means of looking at the website now it was developed 20 years ago by Lego by a guy by the name of Kirk Christiansen who was the owner of Lego at the time, and Christian and now he's joined trivium which is actually the if you want to get certified, which I've ended up looking to because it's twice it must mean something. They've got a beautiful picture here of guys building all sorts of contraptions and you create a landscape of all these contraptions and that becomes Is your world.

Mike Bugembe  30:01  
Yeah, there's a, there's a structured approach to the thinking in that, and it's quite abstract, but it really jumps into your business, you know, or into your way of thinking and really pulls out creativity is great for team building, it's great for strategic thinking absolutely fantastic for strategic thinking, I use it on the AI side of things, because you do kind of need to get out of the box, the trouble we have with AI, is that a lot of people tend to think of the use cases that everybody else is doing, it's my turn, then I also have to build the terrain model. If my competitors do recommendations, then I have to do recommendations, and just missing all the extra value that AI could potentially bring you.

Ryan Purvis  30:42  
But you don't think that those sort of models churn and recommendation and, and whatever else, are commodities, now, you should just go and rent them from Microsoft, or Amazon or Google, use their models. And

Mike Bugembe  30:56  
yeah, it all depends on the domain. So I'll give you an example where a canned recommendation model which they exist, by the way, couldn't work for us. And this was at just giving. So most recommendation models are built on either product sales, primarily product sales, or even content delivery, which plays on the look alike approach. Whereas just doing the domain, where giving is so inherently personal, and giving changes over time. So, you know, what we did is we studied giving for a long time, and we saw that, you know, for example, the moment I became a parent, children related charities were more important to me than they were before, you know, because I'm from, you know, anything given to Africa, I have a very different perspective to charities in Africa than perhaps somebody else might. And, and you then go through things like I have a friend who, who's he's a, he's an ex rugby player. And when he got married, he has three daughters, two daughters, and now he's so focused on women's education. And this is thing I've ever seen him, you know, care about before, but now, that's something he's so passionate about. And so I'll give him changes so frequently that you can't say, because you've been like, because you bought this product, this one is one we can recommend for you. Because you give to this charity, we can recommend this one doesn't work that way. You know, it's, I'll give you an example how personal it is that you can care about cancer, but actually, you're really passionate about eradicating the disease, or you're really passionate about palliative care, you know, and those are two different things, two completely different types of charities. And you can't just sort of replace one for the other because I'm getting palliative care because of how they took care of my granddad. You know, and then that sort of thing. So it's, so those kind of approaches, it depends on the domain, as well. Yeah, I

Ryan Purvis  33:03  
mean, you had that, that. That old example of, I think it was Tesco. There was a target there was predicting people were pregnant based on what they were buying, for dads and all that kind of stuff. And that actually, I think the old urban legend that that's how the dad found out his daughter was pregnant. But I find it interesting. If you look at something like Amazon, for example, where you'd think by now that have that thing sussed, the recommendation engine, but the amount of times that recommend stuff that I've already bought, I've already already read. And you're like, surely you guys know that I read this book, because I read on Kindle and keynotes on your apps, or surely you bought this because I bought this in your off your orders.

Mike Bugembe  33:47  
That is really interesting. And I would say I'm also surprised that they're getting wrong. You see the things. So I teach a course on trying to get people to understand the world of AI data and machine learning. And I try and get non technical people to try and understand that. And the reason why is because a lot of the time, again, like we saw when we were talking earlier, a tech person who build a tech product is excited by that tech. And then they're ready to move on to the new thing, right? No one wants to stay developing HTML when now you can develop in Python and so forth, right? You want to move on. And so that desire to continue to look after that goes and then you often get this thing of we've got to replace the whole thing with Nico right, you know, which is probably more exciting. But then on the business side, there's a lot of work to do around the skills of identifying really useful use cases for artificial intelligence, delivering the projects where we talked about things like explainability, those non functional requirements where you as the domain owner really know them. And the last thing is managing them as as they go through the lifecycle because things change. In fact, half the B to C Type algorithms that were around before COVID are going to have to change now, because by behavior has changed. So you'd be silly to just stick with the same algorithm that was trained on that previous data set, it's got to use an add more advanced than an updated training set to begin to understand the new world or a new environment that is operating. So yeah,

Ryan Purvis  35:24  
yeah, I mean, I helped a friend of mine, they both called singular decisions. I think I called and I was going back to the churn model and that they were they were looking at how you how do you look after the consumer of content? Because as you look at TV today, that's going more and more down the IPTV route. So Netflix, Netflix is not IPTV, I learned that the other day. But it's a it's an actual like black Sky Sports been transmitted over, you know, over the Internet, but it's, you know, looking at needing a specific minimum speed have sort of 25 meg lines and that sort of thing. Whereas Netflix are run on a to make life better so that you as a consumer can have a boat, okay, have personalized channels. And then, but there's a need to almost manage all your subscriptions. And you can kind of see this with Amazon a little bit where you can sign up for Amazon Prime. And then you could they can recommend to you to watch like belowdecks, my favorite TV show at the moment. But then in order to watch, in order to watch that I have to sign up for a soap subscription called I think it's how you write, which is like five bucks. But now I've got a subscription and a subscription. But one of them want to watch it. As long as I've got the season that I want to watch available to me then once that's done, I'll cancel it. But I've still got my main one. And it's really the single decisions algorithms all about trying to keep you on that, hey, you subscription as an example. So you don't cancel, and you're entered. And then you've and you're doing it. So it's there. They've got a an eight step framework, which is quite interesting. is as you move through the stages, they can they can manipulate that's the wrong word in the negative sense. But to keep you on to keep you consuming, but there could be a leaving moment or there could be a joining moment, or there could be an upgrade moment or downgrade moments or maybe you you're about to about to leave and they say okay, well, instead of paying five bucks a month, you pay three for six months. Okay, well, I'll save two bucks. That's that's a cup of coffee almost. So I'll stick around. I'm gonna give you three months free, and then it goes back up to April by then you've forgotten that you've actually stayed on?

Mike Bugembe  37:32  
Yeah, yeah. No, that's interesting. Yeah, very, very interesting. I think I'll lift them up senior decisions to decide what they call. Yep. So what I like about that is it's again, it's decision focused, right. So it's, you know, it's it's thing is to try and, you know, minimize the decision of unsubscribing. Completely, you know, that that is that it's a sort of optimization function.

Ryan Purvis  37:59  
Look, I'm happy to introduce you to profession and connect you guys up to have a chat.

Mike Bugembe  38:04  
Yeah, I would love to. I would love to Absolutely. Yeah, yeah. Yeah. Really good. Yeah. Excellent. Yeah. So

Ryan Purvis  38:13  
I mean, the other thing that I wanted to do so you talk through your sort of history of working on the consulting side, yeah, you know, and then moving into the company side analysis, you got a start up? How you seen the market change over time, especially now with everyone working from home, and data becoming more and more important to decide on renting office space for storage and office space? Or are you seeing a bit of question being asked of data is that that's maybe a better way of calling it and

Mike Bugembe  38:44  
it's a it's a great question to ask, am I seeing a better question asked of data. I'm seeing more questions asked of data for sure. Right. And, you know, if, if the consumer is now doing significantly more online, and is likely to according to McKinsey, and those sort of reports continue doing a lot of stuff online. That is going to create more data. But it also means that the corresponding businesses have to do more online as well. Right. So if the online market begins to get a little saturated, you need to find a way to differentiate yourself online and become a bit smarter. And look, ultimately, every organization is still trying to get customers get maximize their, their their time with that customer. If it's financially a financial payoff or not increase retention, and reduce the chat and increase the chances of them spreading word of mouth. Every organization wants to do that. How would you decide to do that without investing more in data and asking the right questions about your activities within each of those fit. I think what we're seeing is actually was saying because I saw a great article by I don't know whose organization it was, that had predicted around 300,000 jobs from the data analytic space for 2021. And they revise that number towards the end of that last year to 2 million. Right. Wow. When we're talking about a decrease in jobs, I found it quite interesting that that was saying there was an increase in jobs in that particular sphere, you know?

Ryan Purvis  40:19  
Well, I mean, that's made with the industrial revolutions, are they they always talk about losing jobs? Well, actually, it's a shifting of jobs, to to higher skilled. And yes, you feel for the people that are, let's say, lower skilled for one of a better phrase. Because they don't know anything else. I mean, you know, you don't even grow up in Uganda, did you? Or do you?

Mike Bugembe  40:42  
I actually, as a, as a child of a diplomat, we were in many different countries and the continent. So

Ryan Purvis  40:50  
yeah, well, what sort of examples and again, when I'm back in South Africa at the moment, and I'm realizing more and more that when I speak to certain people, that they've got no other job than to be a gardener, or a domestic because they've got no other vision for anything else. Right? And how do you how do you tell that will help that person so that their kids have the opportunities that they didn't have, you know, to be an accountant or a lawyer or whatever they want to be? Yeah, and not be not be stuck in the same rut and to an extent, and it's, you know, it's a, it's a depressing situation, sometimes, because there's only so much you could do and you can't help everybody. And most of us are most affected by it by COVID. Because, you know, the first thing to keep first thing to do is keep people away from you. So you're going to keep people away that are traveling on taxis every day, and you know, living in squatter camps and, and that sort of stuff.

Mike Bugembe  41:49  
Ryan jump scares me even more, because, you know, if we think about the world of disruption, and how it is everything is getting disrupted, right, you can name any industry is going to go through some form of disruption. Education, is going through what I believe to be a very silent disruption.

Ryan Purvis  42:06  
Oh, yeah, definitely.

Mike Bugembe  42:08  
where, like, now they're gonna they're doing school from home. Right. And let's not be surprised if something more comes off on the back of that, right? Where we see some some change, it would be a shame for it not to, because it hasn't changed for a millennia, right? You know, a lot of what we're doing, that's pretty much exactly the same. Now, as the job market is shifting. The education market can continue to do the same thing, if it's if its role is to equip our children for the future, right? Just do the same thing. And so you end up finding, I mean, we've seen a revolution, what do they call the MOOCs? I can't remember.

Ryan Purvis  42:49  
Yeah, I see online courses or something.

Mike Bugembe  42:52  
That's right. You know, so we've seen a lot of that where people are now, you know, the online market has shot through the roof during this period. And you're like, well, they're selling things to people who don't have to have gone to uni, and, and all of those sort of things. And you're like, I do see a sneaky shift taking place here that, you know, we're not keeping an eye on it. And most of this, most of the time disruption happens, because we're just not focused on it. It's some other people who are paying attention to it. But I do think the education space, you know, edtech is ripe for massive disruption. The government already tried to get AI to mock exams. You know, there's, there's, there's a lot that's happening that, you know,

Ryan Purvis  43:35  
yeah, and it is something that if you look, what do enough, a year, but since the last lockdown, I see Yeah. So almost in March, or since the first lockdown for the for the UK, at least. The amount of kids that had to be educated remotely, and also been educated remotely, but they're losing out on the social aspects of the social skills that they need. But they're still doing this old fashioned way of worksheets. And yes, some some some schools are using apps and some aren't. But not every school in every sector has the

Unknown Speaker  44:08  
tools. Yeah.

Ryan Purvis  44:11  
There's got to be a way I mean, you got to go to Google or to Microsoft and say, please sponsor million laptops, with Google Classroom or something installed, so we can do this a different way.

Mike Bugembe  44:22  
Yeah, yeah. But also when we're talking about, you know, so in the tech world, we have a world of hyper personalization. So why not personalized education? And it's so interesting. Caleb, here is the academic of my three kids. He's the most academic and doesn't enjoy learning via lockdown. Right? And my eldest, Josh, who's the least academic, has thrived during during a lockdown. Like he's, you know, his grades have gone through the roof, you know, and it's interesting because there's just something different about the environment that that is meant. This is his learning style. And then there's a bit about me that's a little reluctant to let him, why would I put him in a place where he was really struggling to somewhere where he's so thriving, like genuinely thriving, you know, he manages his own time, everything, and he just made sure we're keeping tabs on it, of course. And he's flying, you know.

Ryan Purvis  45:22  
But is that is that not the point is, you said hyper personalization. But that's, that's the problem with education system has always developed to the lowest common denominator. So how do we train you to work in a factory, there's going to be a shift that's going to be from, you know, eight to three or 95, or whatever it is. So you're used to working like that? How do you change that? How do you think you've answered the question in a way? But how do you? How do you treat people? Because, I mean, these things were put in place based on some very old theory, which was right at the time, it was what Henry Ford used to build his, you know, conglomerate? what it is now? And but how do you go? Fast forward? 20 years with people haven't been educated in high school the same, you know, they're damaging courses they want to do? They haven't got a degree, because I've done 60 courses that are two weeks at a time. You know, you gotta gotta hire them differently, then. Yeah. critical, critical thinking, problem solving that sort of stuff.

Mike Bugembe  46:24  
Yeah. You know what, let's mix it up, right. And so it's not just the jobs that are changing technology that's changing in the way that people get educated. But also, what's changing is we have a big social revolution taking place as well, right away there, there is an increased desire for equity and equality across the spectrum across the globe. Which means that like an organization that I worked with, after uni, Accenture will have to change its hiring practices, if it's to meet the, you know, the the the equality, equity, equality numbers, or the diversity and inclusion numbers that it needs to. So previously that that organization would hire from, let's just say, five universities. Right? Okay. And to get into those five universities, you probably needed to go into private school. And to get private school, you probably needed a certain amount of wealth within your family. Right? Yep. And now that's a cascading effect, which means that everyone who worked in Accenture my time was the same type of person. Right? So now you have a change in the way everyone's being educated. And as Asians are focused on growth, so they need the skill set. Right? You know, you're gonna say no, to a top class developer, because he didn't go to this particular unit is then I have to change hiring practices, which means that old way of looking at value, you know, with education may change. I don't know, I just feel like there's a real shift coming, that, you know, perhaps we need to prepare for?

Ryan Purvis  47:57  
Yeah, I think you're right. I mean, there was that there was a news agency, it could have been NBC could stand corrected, where they used AI to go through all their previous hires, so all their CVS and their careers and stuff. So they could hire based on what they thought were the best people that match to their culture, they end up hiring the same, you know, white male, same mask, same article, same varsity whatever, and completely discounted most females and parents of color because they didn't fit this biased profile. I think bias comes into to a large extent. And I think you're right, that the rules have changed. Yeah. But there's a bit it's rippling through faster. And maybe it says the hidden wave, it's building up to tsunami that's gonna hit us all the face at some point.

Mike Bugembe  48:43  
Yeah, well, let's, let's talk about the time, and there's the rate of change rates of change, right. So AI is an exponential technology. So that means not a straight line, which means it is changing is going to be faster as we move along, not slower. Right. So I think it will hit us in the face, I think we will just be shocked by it. You know, in 2014, a whole bunch of experts gathered together to ask the question, when will go, when will the machine beat a human being at the game of Go? Right? And all the experts said 10 years, at the very least. And it happened in two, you know, that it's just that's what we're dealing with. Right now. Google did not do protein folding, you know, a problem that didn't expect to be solved for another 50 years, you know? So, yeah. Fascinating.

Ryan Purvis  49:34  
Did you think I mean, that's, that's all leading up to the singularity. Do you think that's something we'll see in our lifetimes?

Mike Bugembe  49:41  
You know, what, I do think there was one thing that will slow it down, right? Is, is things like GDPR because the capture of data is what's needed for us to get that we need literally all the data, right? What we have as humans is we you at the moment was still capturing And processing more data than a computer can. You know, I think I can't remember anecdotally, I have something like 17. Between it's either seven or 17 million, can't remember data points per second, right. And it's such a wide variety of data points that haven't all been captured yet in the database. And therefore, our decision making capabilities are far more superior than that of a machine. In the wider context, I've got to try and remember that number two, it was a great book that I read that talks about how much they were actually. And I think that's the one thing that I'm so

Unknown Speaker  50:40  
sorry about

Ryan Purvis  50:42  
that. Sorry, my, my wife's calling me, right.

Unknown Speaker  50:47  
Yeah, yeah,

Ryan Purvis  50:49  
I'm sorry, I didn't hear loss a little bit because she was calling me in I was busy.

Mike Bugembe  50:54  
So I think the thing that will slow us down is things like GDP, other people's awareness on the privacy of the data that needs to be captured. And if the data isn't captured, we will it will take a while for us to get to that that singularity, singularity point, if you like, and that's capturing known data sources. There's also an unknown data sources that, you know, where you and I still have a gut feeling, or a God thing. We haven't codify that yet. So, you know, until that gets in the database, then we're still away from that.

Ryan Purvis  51:28  
Yeah, I mean, it's still it still makes me think of data in the next generation, Big Star Trek fan, clearly. We aren't exploiting exploiting emotions, and taking hunches and all that kind of stuff. And he's, you know, in the 23rd century, but and is unique. There's only one of him to, if you really look at it, it's probably a good time to tie up. So what's the best way for people to get hold of you?

Mike Bugembe  51:57  
Yeah, so I'm on Twitter. I'm on LinkedIn, Twitter, I'm an admirer. jambay. me on LinkedIn, and also on Instagram. So yeah, any of those approaches is perfectly perfectly fine. Get in touch. I love speaking to people about the whole world of technology. So I do tend to respond or jump onto my website, Mike gambi.com. So there's a there's a contact us section as well.

Ryan Purvis  52:27  
Lovely. Well, thanks for thanks for coming on.

Unknown Speaker  52:30  
Thank you. It's fantastic discussion.

Ryan Purvis  52:33  
Thank you for listening. Today's episode, hit the big nose our producer and editor. Thank you, Heather. For your hardware for this episode. Please subscribe to the series and rate us on iTunes or the Google Play Store. Follow us on Twitter at the DW w podcast. The show notes and transcripts will be available on the website WWW dot digital workspace that works. Please also visit our website www dot digital workspace that works and subscribe to our newsletter. And lastly, if you found this episode useful, please share with your friends and colleagues.

Transcribed by https://otter.ai

Mike BugembeProfile Photo

Mike Bugembe

Author, Speaker, Consultant - Helping business leaders get real business value from data, analytics and AI

Award-winning big data evangelist, author of Cracking The Data (Amazon Best Seller in three categories - Grab your copy on Amazon here: https://amzn.to/2Gd1Gnk), blogger & executive advisor working with organisations to get value from the deluge of data that is being produced in our modern era. Advises and mentors CDOs, CTOs, CIOs and data science teams looking to transform and unlock the full power of their data & AI efforts, and always on the lookout for new interesting projects.
My greatest motivation comes from helping business leaders understand how to utilise their data science team most effectively, how to make the most of their AI initiatives and to appreciate the value of their data. Using my processes, I have been able to transform organisations, establishing data & analytics as the real lifeblood of modern companies.
By identifying opportunities for data science, building an effective team and launching a social giving product, I was able to unlock the power of data to help raise millions of pounds in donations for good causes during my time with JustGiving. To learn a bit about my time as Chief Analytics at JustGiving from May 2010 to January 2018, check out the Forbes article here ➡️ http://bit.ly/2XVHN9X
I am honoured to have been named a top digital one of the UK’s top digital masters (check out the Guardian article here ➡️ http://bit.ly/2TUmg2a) and to have won the award as the leading executive in the Big Data Analytics space. You can also check out some of my work in the book Predictive Analysis or in Big Data and Predictive Analysis, which I co-authored.