Richie Cotton: Welcome to DataFramed. This is Richie. Now, actually, it's not just DataFramed. Today's episode is a collaboration with the Analytics on Fire podcast. Now, the hype around generative AI is only the tip of the iceberg. There are so many ideas being touted as the next big thing that it's difficult to keep up.

More importantly, it's difficult to know which ideas will be the next chat GPT, and which will be the next NFT. Today I'm chatting with Mico Yuk, host of the aforementioned Analytics on Fire podcast, to dissect the trends and separate the wheat from the chaff. From data fabrics, to decision intelligence, to responsible AI, neither of us are shy about having an opinion.

And beyond podcasting, Mico has two decades of experience in data. She's the community manager at AcralData, And also has her own data trading consultancy. So I am incredibly excited to chat about the latest data and AI trends with her. 

Welcome to DataFramed. Good to be chatting with you.

Mico Yuk: Yeah, no, I, excellent to be here. And what a fun prep session.

Richie Cotton: Excellent. So I wanted to kick off just by asking you about jobs, because it seems like there's two camps here. There's like, some people think AI is going to make us all more productive, and then we're going to live happily ever after. And some people think that AI is just going to steal our jobs, because it's going to be better than humans at doing everything.

So, Where do you fall the spectrum?

Mico Yuk: So first and foremost, I just need to be very transparent that even though I'm a millennial, I think ChatGBT is revolutionary. Case in point, in my recent doctor visit, I just uploaded my test results. I asked questions and then I sat a month later, listen to my doctor, tell me everything Chattubiti said.

And then I also had asked Chattubiti what to ask my doctor. And then I told it, answer me. And I asked my doctor those questions. So full transparency, needless to say, I just put my life up in there. Every contract I have, my mortgage, I'm like, I got questions. So I might be a little bit different, a lot of millennials.

But with that said, even though I do believe in AI, really think that People who are smart are looking as at AI as a productivity hack, and that's how I use it. Right. So I'm not afraid about AI taking over my job or my roles or people's roles. I'm consistently telling our students and our teams, here's how to use AI to get back time.

the things that some of these things that you've been doing that have been taking days, hours, things like what KPIs are very similar for your metrics, or what's the best chart for this, or how should I draft this story, dude, AI. Like, you know, and it's not that you should get lazy, like one of the fundamental things that I'm afraid with AI is that critical thinking is going to decline tremendously because you have this machine that thinks for you.

But there's so much things that you could use AI as an accelerator for. to do much more than you could have within an eight hour workday. I see it as an accelerator for those people who are willing to literally have, sit down and think, what are the things that I'm doing that I'm not very good at? And what are the things I'm spending time on that I could use some help on?

I think if you make those lists in your job and you throw it into something like ChatTBT or one of these AI bots, you're going to realize that you can significantly reduce time, produce a better product, and then focus on the things that only you can do.

Richie Cotton: That's interesting. It seems like there's a different standard, like people where it's like, I never have time for everything. There's always more that I could be doing. And then AI is something that's completely necessary to help you be more productive. But if you're already like, Oh, you know, I do my job and that takes me 40 hours a week and then I'm done.

And I don't need to think about it anymore. Then maybe that's the sort of job where you're going to be more at risk.

Mico Yuk: Yeah, so you're going to be replaced. Like,

Richie Cotton: Very blunt. You will be replaced.

Mico Yuk: just be candid. So your days are numbered. If that is your mentality, you are already a dinosaur. You're aging on the company's tab and you're going to be replaced. So I don't have a candidate to put that like if, you know, and hopefully if you're, honestly, if you're toward retirement age and you're just, Hey, I just need to live up my last three years, it's fine.

But I think beyond two to three years, if you have that mentality, it's either time to find a new job, go to entrepreneurial route or do something else. Cause you are literally going to be replaced and not even by a machine. You're going to be replaced by someone who's forward thinking and embraces new tech.

Richie Cotton: And so for people who are working in data, how do you think that generative AI or AI more broadly is gonna be used? How it's gonna change people's roles?

Mico Yuk: Oh my gosh. So I have to tell you a secret. Okay. one of the first certifications I had in college was SQL certification, Richie. And like, I was bad. I was, I mean, in a good way I could do SQL, I could fast forward to where I am right now without aging myself. I don't remember nothing but an if then clause, like, I don't, I mean, I'm an Excel guru at this point.

Right. So I've traded my SQL hat for Excel. I go into chat dbt and I'm like, I need this SQL statement. And I said, and also here's what I want to do. And it just pops out. I go in and I'm like, Python, like I need to get this done. You see how I'm smiling? It just pops out. I go in and I go, I need this script.

And here's what I needed to do. And I need this short and explain to me why you're doing this. Cause I don't understand what the hell you wrote and it still comes out and it works. And so I think that with regards to data people beyond code, I could tell people what I use it for and what I teach my students to use it for.

We use it for everything from, one of the common questions, for instance, that we get from companies is I specialize in, for instance, KPI and goals. A lot of times companies will say stuff like, Hey, what are other KPIs or metrics companies are using? What are other companies and metrics, you know, like I, like, I, I know it sounds lazy, but like anytime I get questions that require me going to Google, I just go to AI first.

Anytime I am struggling with something and I feel like I'm stuck and I need an additional perspective, I go to AI first, so I think that from a data perspective, again, really take a look at the things that you're spending time on. I think that fundamentally, our jobs are not just going to change in terms of productivity, but I've said this on a few different podcasts recently, and I said, if you're in a data field, you have to understand you're eventually going to become a professional AI prompt engineer.

And what do I mean by that? We are going to head to the stage where users are going to be coming to you at all stages, whether you're a data engineer, data analyst, whatever profession you're in, and say, Hey, my prompt isn't working. I'm not getting the right chart to develop. Like, you know what I mean?

Like, they're going to be coming to you saying, What's wrong with my prompt? Because every data tool we have now has an AI button in it. And you better believe that users are going, Oh, that's for me. I don't know what I'm doing, but I'm going to go in and try. And so I really think that to get ahead of that curve, data professionals themselves need to become prompt specialists.

That's where I think this is heading.

Richie Cotton: That's interesting that you should be doing everything AI first, rather than like, I'm not gonna write the code myself. I'm gonna ask AI to do it, and then maybe fix it. And then also you'd just gotta be like focusing on like, how do I get the best results outta ai rather than, how do I. Do the best work myself and then maybe get AI to augment it.

So it's a kind of flipped approach.

Mico Yuk: I recommend both. If you're a do it yourself person, put the thought process into it. I have stuff that I do when I put my heart into it. I still go to AI and ask it. I go, give me some case studies. Take a look at this. What am I missing? Like I ask it DAP analysis questions. What perspective am I missing?

Give me some studies, like, is there a gap in my analysis, you know, so even if you do a good body of work, I'm not, not recommending that, I'm saying that AI has access to so much data, it's a great soundbite. Now the caveat to that is that AI lies. A lot of people don't realize that. So if you're asking AI for quotes and stuff, it just fabricates stuff.

So make sure you ask it to give you valid links. A lot of people, like they're, they are lawyers who are heading to jail for perjury because they AI fabricated cases. So just be very careful when you ask for quotes and stuff and say, tell me who said it and give me a link because AI does lie. But I have a question for you, Richie.

Let me turn this over a little bit. I mean, Datacamp, you know, and I don't know if people who are listening know this, you guys have over 450 classes. That are data, AI, and analytics focused, you serve over 2, 500 organizations, 12 million students over nine years. I mean, you must be seeing a trend here, right?

Tell me what you guys are seeing from a learning perspective with regards to generative AI. Like, what are the kind of classes people are taking? Give some perspective there.

Richie Cotton: Yeah, absolutely. And there's two real sort of streams here for generative AI. So a lot of it is just people who want to know, how do I get started? How do I use chat GPT? What projects can I use this for? Why should I even care about generative AI? So there's this very broad interest, like it's actually AI's become cool again.

Uh, And so, most people's wanting to know the basics, getting that sort of Fundamental AI literacy, and the other streams, people are trying to build things with generative AI. So this is where you start adding AI feature, generative AI features to other bits of software, or maybe just trying to build something to help you do data analysis better, in this case, it's all about using API.

So one thing that's been a kind of, yeah, so a side effect of this is like suddenly knowing how to use an API, yeah. like pull an API from Python is a very popular essential skill again.

Mico Yuk: I love it. And so, in a data professional realm or any professional realm, who are you seeing as more inclined to take these courses?

Richie Cotton: Yeah, so it's interesting because it's all sort of data professionals are wanting to know a bit about genitive AI. So you see the common roles, data analyst, data scientist, data engineer. What we also see is interest from software engineers because it, Quite often, it's the software engineering teams that are having to build these AI features.

They don't have a background in AI, so now software engineers need AI skills as well.

Mico Yuk: now do you think that data people need to be wary of the fact that software engineers are kind of intercepting and coming over because traditionally AI is like a data thing, right? But I mean, the software right now, the software gravy train for funding is drying up a bit. So do you think that that's why they're coming over and saying, Hey, we need to start to embrace this because it because I noticed tons of money.

BC's are pouring tons of money into AI.

Richie Cotton: Yeah, absolutely. So I think a lot of software a lot of data professionals do need to be aware of this. One interesting trend is that data teams are now moving into engineering departments. So, the data team is going to lie under the chief technology officer, for example. So this is happening tend to be your organizations that are doing this.

But it's an interesting trend and data science and software engineer, maybe those, maybe those two roles are kind of converging a little bit.

Mico Yuk: Converging a little bit. Yeah.

Richie Cotton: So the one data job that is more safe from this is the data engineer. So Mm you still need to have the right data pipelines in order to feed the generative AI.

Mico Yuk: So I have a controversial opinion on that. I think they're safe until we get over what I call a Dropbox error. So you remember Dropbox where it was like, I am on premise. Like, no matter what you tell me, I'm not putting it in the cloud. I'm like, dude, the cloud is a box in somebody's, in a building somewhere.

Just to be clear, like it's still to some extent on premise, it's just that you've outsourced it. Okay. So anyways you remember that era where it was like, I'm on premise, then I'm hybrid, and then, oh shoot, Dropbox drop the bomb and now it's okay to put everything in the cloud and everybody's going to the cloud.

I almost feel like, wait. AI, it's kind of going through that honeymoon phase or that phase of like, well, you know, in data or in data engineering, I should say of like, I know my stuff and data pipelines are important, but I think the minute they get the security right and companies are like, just dump the damn pipeline in there, like, like, you know, or there's a secure way to attach it.

I don't know if that's going to last.

Richie Cotton: Oh, so you think the data pipelines are going to be more automated then?

Mico Yuk: Absolutely. I think once the security element, like what happened with the cloud and on premise era, I think once the security element is satisfied or they're happy with it, I think that that can be automated as well.

Richie Cotton: So what happens to the data engineering roles then, do you think?

Mico Yuk: No, I don't think it's going away. I just think they'll become more efficient and be able to do more. They'll be able to work smarter, right? Because, again, you still need them to look at, this is the thing data people. When the outputs come out, we still have to make sense of it. there's a term called data informed.

I know we're going to get into this. We still have to make sense of it. We still have to go back to prompts. If you use AI and I use AI every day, like I literally, one of the things I do in the morning, I just talk to chat, GBT. I talk through things like, I don't know if you know, there's a talking function and I talk and my family is like, what are you doing?

I'm like, I'm talking through this idea with chat, GBT. And she talks back to me. I put a woman voice is very confident. And I talked to her and she thinks it through at me and Miko, this is how you'd approach this. And I'm like, well, tell me why. this is how I'm thinking. So one of the things I think that. is anybody uses AI knows that to get to a final state requires a series of refined prompts. And I think that even though you're going to get output, mastering that refinement is going to honestly be one of the most valuable skill sets coming out. And when I say prompts, whether it's coming from your mouth, your hands, soon it'll just be like death to Elon would just be sitting on and thinking, to be honest, and it'll kind of bring it together.

I still think human beings have to have an input or take what comes out and make sure it's correct.

Richie Cotton: That's interesting. I like the end game where it's just like you're thinking to an AI that's inside your head. That's very science fiction and also mildly terrifying, but I'm sure, a few generations down people will be like, Oh, yeah, it's totally normal. I'm just thinking with my AI.

Mico Yuk: Yeah, well, I do it on my phone again, just talking. I do it at least three times a week on some core ideas that I'm thinking about and I just go and pull the transcripts.

Richie Cotton: If all these jobs are changing, how do you think organizations need to change their approach to training their employees?

Mico Yuk: So first of all, I think they need to stop jumping on the hype train, I understand it's a lot of pressure from the top and you're going, everybody, we got to learn AI. I mean, first, let's just stop. Let's think. The first thing you should do is ask AI. Here are the profiles of all my teams, and here are our demands.

How should I approach training? Ha ha ha! Step one! Here's my budget. Give it a few parameters. I have a five person data team. I got a data scientist, a data analyst, da da da. Here are the core businesses we serve. Here is our budget. Here are the lists of classes we have. How should we be approaching training based on the demands that are coming?

I would literally let AI create my entire training plan.

Richie Cotton: I do like the idea that you should start with, like, what do you actually need rather than just picking something because it sounds cool. Um, That just like,

Mico Yuk: That's why I said based on the needs, right? Take the needs that you gathered from, recommend doing polls and stuff. I mean, I don't want to go back into like how we do this, but I did a podcast on this with the head of learning from LinkedIn with Steve. I did this a few weeks ago and you know, they have 2000 data AI analytics classes.

And he said, Hold your leadership, hold your end users, get that data. And I was like, dump it into AI, put the parameters of your team and say, how do I get my team from point A to point B with this amount of money or this amount of time? I let it create a learning plan. That's my whole point about the thinking process.

Like AI can accelerate

Richie Cotton: I like that. So you're using AI to tell you what you need to learn about AI in some respects.

Mico Yuk: a hundred percent. Correct?

Richie Cotton: I think generative AI is only one of the many things that are being hyped and buzzy right now. Do you want to pick next topic to talk about? 

Mico Yuk: there's one that was very interesting that you brought up to me around decision intelligence. And again, Richie, we're like millennial post age. And I have to tell you, right, there's some, when, when you sent me over, kind of like, here's the prep. There's some things that popped out to me and I was like, yep, I know Datacamp is having to, you know, cause you guys as a trading facility, I know you guys are having to like deal with like the era of data mesh and data fabric and decision, and you know, they're just words.

And so I don't actually feel like me, but I just kind of feel like I've heard these terms before.

Richie Cotton: Yeah, definitely. A lot of these things are new words for things that we've already encountered already. So as I was like, decision intelligence is very reminiscent of data driven decision making. 

Mico Yuk: A matter of fact, did you see the latest Harvest Harvard review? I was, I work a lot in the bookstore just because being surrounded by knowledge when you're working helps me, you know, I just get a spark. I was in a magazine area yesterday in Barnes and Nobles, and I picked up the Harvard business review at the front of it, it says, storytelling and good It's a framework for good decision making,

Richie Cotton: well, okay,

Mico Yuk: right? And I'm like, I was looking for buzzwords. I'm like, are they doing decision intelligence to your support? Like, you know what I mean? So again, those of us has been around for a little while. I know the younger kids are listening to this.

I'm sorry. Is that the popular bubble? It's just that we have seen so many iterations of this. And that's why I always tell my students I've trained over 20, 000 students myself, globally focus on the foundations. This in your lifetime, decision intelligence is going to become something else. It'll become decision AI intelligence, whatever it is, get the foundations right.

And so I believe in decision intelligence and here's why. Back in my day, do you remember decision support?

Richie Cotton: is it the same thing, but a different name? Or is this something slightly different?

Mico Yuk: No. So if you think about the maturity of data, right? Let's do the maturity cycle. Determine the maturity of data was, take a data warehouse, right? first, we had basically, oh my gosh, I hope I remember the whole maturity cycle. We had a lot of data. And the whole goal of the first level of maturity cycle was to get the data into one place.

Step one. When we got it into one place, everybody said, wait a minute, this is not telling us anything. We just have a bunch of information. So the next question was, how do we learn from this, Information turned into what we call insight. Insight was defined by things like trends and, these different elements.

Then we were like, oh my gosh, insight, like what is valuable? That's when we moved over into what we call the wisdom cycle. And the reason I bring up the maturity curve is because I really feel that back in the day, when all of these BI teams were tagging, they're tagging literally in signatures were like, this is decision support.

We're at a place where people had so much information, and so little insight, BI teams and data teams needed to be decision support. What I think has happened now is that as a lot of organizations that are crossing the barrier from having insight to wanting wisdom and wisdom on a maturity cycle is defined as repeated insight that provides.

So we have predictive date, right? In other words, it's not just telling me what's coming. It's actually helping me to understand how to prevent it or change the outcome. That's wisdom. It's insights that are trained. that can change outcomes and produce results. And so if you look at companies that are striving to get there, I think we see the progression of BI and data teams and analytics teams that are no longer decision support departments, they now become a decision intelligence.

Richie Cotton: So heard the same sort of thing, but with different terminology. So I've heard this is the data information, knowledge, wisdom pyramid, and it's a similar idea where you start off with this data and you gradually build layers until you get to the point where you can actually do something useful and make a business decision right at the top.

Actually, I have to say, I do like The rebranding a bit from data driven decision making to decision intelligence, because it's easier to say for a start. But also with data driven decision making, like the whole point was most decisions are made with no use of data at all. It's just someone like having an idea and saying it, but then a lot of people are like, well, you know, you shouldn't just use data to make a decision.

You've got to think about like what all of your colleagues think and the psychology

Mico Yuk: as well, hold on. That's being data informed, right? So on the data's transformation cycle, a lot of companies went from being data driven. Because remember, remember the whole era of data literacy, which is still a wrong. And I know we're going to get into that. The whole era of data literacy was like, when people see data, do they even understand what they're seeing?

And unfortunately, during a pandemic, we learned. They don't. The pandemic highlighted to the world, even how bad journalism was, like some of the charts that came up from journalism, like news TV was like, I, some of them, I was like, I need to get some people. I got DM saying, Nicole, maybe you should call this new station.

Look at this chart. And they knew better data. People were like, this is wrong, And so I think during the pandemic, we learned that, hey, data interpretation and literacy is a lot lower than we would hope. But kind of getting back over to data driven. One of the things I think that became clearly evident is that being solely data driven is not enough.

I think being data informed is important, but I have to be honest, I have a different angle. I don't know what the term for this is, but I think it's Data driven, which is the data in its purest form, combined with being data informed, which is knowledge experience from human beings perspectives, actually combined with AI, which is what I call the infinite knowledge.

Now, I don't know what maturity that's going to be, but I actually think that that's the right combo to drive true decision intelligence. What do you think, Richie?

Richie Cotton: Yeah, absolutely. I like that. It's not just data It's like it's all the rest of your organizational stuff as well. I think we said makes a lot of sense I think the tricky thing for me is how do you do this consistently? Cause it's like, sometimes you're like, okay, we've got a lot of data. We're going to, put it into a process and make a really good decision.

And sometimes it's like, oh, well, you know, it's got to make a decision quick. And we're going to rush through this. So. Do you have any advice for like how you do better decisions consistently?

Mico Yuk: I do. Sorry, Richie, I told you about me putting up my blood test. I like transcripts. I do transcripts of meetings. I feed it all to AI and then I question the hell out of it until I understand what I need to get from the conversation. So for me personally, when I go into meetings and I actually am transparent about this, I tell people, Hey, we pay for certain AI functions.

I said, I'm taking this transcript. I'll remove names. And I actually put it up and I say, Hey, this conversation happened around this topic. Here are the storyboards and visuals. I asked it a couple of core questions. Do a SWOT analysis. Where are the gaps? What decisions should we make? Give me different decisions based on risk levels.

Who's going to be a proponent of which decisions and who's going to oppose it and why? So I keep saying, like, I think the way I use AI just because I use it every day from everything from meal planning to date planning to, I go to it and say, just help me because I always, you know what I feel? and I've always had this about data.

I feel like. Infinite knowledge is a tool, meaning that like chatGBT has access to knowledge that I, in my brain and lifetime will never be able to comprehend. And I try to use that infinite knowledge to validate and become smarter about what I'm doing. So that's my approach. I know that's not for everybody, but that's my approach to really understand the depth of the conversation because I'm a storyteller.

Richie Cotton: That's brilliant. I have to say, I've only recently started using these sort of ALI meeting recorder tools. Been a complete game changer. Like I've had so many meetings where I've had a conversation about, yeah, I know what's going on. Come back to it couple of weeks later. Not a clue what we talked about, but having just this record of, Oh yeah, this is what happened.

Automatic summary generated is definitely one of the most important tools you can use.

Mico Yuk: Same here for my podcast. And also one more thing that I love to do is when I upload videos I like to ask it to read body language only because I myself, I'm, I'm fascinated with behavioral psychology, but I actually asked it to read body language and I went, once it gives me hints of like this person was apprehensive on this, I DM people.

I'm like, Hey, Richie. I remember in a meeting, like when we brought up this topic, you look, it was like cringy, you know, just like be sweet about it to some extent. And I'm like, I just wanted to make sure that you, how did you feel about that? Cause I just felt like I got a vibe and just break the ice, you know, and of nine out of 10 times, Chatty BT is right.

It's like, well, Miko, I didn't want to say anything. And once I start hearing that guys, like, I just keep saying like, use this tool. You're smart. Chatty BT can read expressions. It could real tones. It can, sense tension that we miss, don't sleep on these tools.

Richie Cotton: Do your colleagues feel that they're being monitored somehow then? Because. They know that their body language is being monitored. I think they're going to start doing fake smiles at you. Just like, oh, yeah, that's a wonderful idea.

Mico Yuk: So keep in mind, I've been doing this for a living as a storyteller. So I did a lot of training on body language and stuff. So they know, like, I just, I read stuff. I'm very EQ driven, like just reading what's not said. I just use chat to get smarter. So I don't think anybody who deals with me personally do it, but I personally recommend using That it's a way to read the room, especially for EQ challenge people.

And decision making is so critical. You have to understand the unsaid things in the room because typically the things that are not being said in decision making are the things that are going to drive the decision.

Richie Cotton: All right, so I want to go back to something you mentioned before. you mentioned the ideas of data meshes and data fabrics. So, Gautam, do you just want to talk me through it, Gautam? Because they're both very hyped ideas. Very closely related, but slightly different. So do you want to talk me through what they are and what the difference is?

Marker 

Mico Yuk: Yeah. For this purpose, I actually went and looked at the Google definition, because I think we should start at the top, right? Because typically people, people are reading this and like, oh, shoot, let me go look what it is. I'll read out for you. Okay, so according to Google, Data Fabric is an architecture that facilitates the end to end integration of various data pipelines and cloud environments through the use of intelligent and automated systems.

So Richie, let me ask you a question because you've been in the game for a while. What does this song like in a previous lifetime?

Richie Cotton: Okay, so, this is like you've got lots of different databases all over the place, and you feed everything into one central database. So, it sounds like a data warehouse. 

Mico Yuk: You know, ,Richie for 400, like, I mean, it is ladies and gents. It's not again to be a Debbie donor here, but when you've been in the game for a while, welcome to like data warehouse with new tools, and rightfully so, you know, like data fabric, when I heard this, right, just my sense of humor, I was like, is it linen, a hundred percent cotton or viscose?

Right? I was like data fabric, you know, but at the end of the day, basically I feel like data warehouses were more static. There's a lot of challenges with ETL. let's not forget, I think data fabric is shedding the ETL process. to some extent, right? It's bringing in this automation, bringing in these new tools that we have access to and automating something that basically kind of was the black hole of my generation.

Data warehouses are known as a black hole. And why was it the black hole? I remember CIOs telling me Data warehouses are where I put money in and nothing actually comes out.

Richie Cotton: That's a harsh interpretation of data warehouses. But yeah, I can see how that's going to happen to a lot of organizations. You build this thing, and then the value isn't immediately apparent.

Mico Yuk: Right. And I think data fabric, on the other hand, now that we have access to new tools, we have data engineering, which focuses primarily on data pipelines. I think the focus of the rebranding is the fact that we are able to actually overcome that very statement and challenge.

Richie Cotton: It seems like the pendulum swings backwards and forwards between centralize all your data, and then it's like, well, actually, it's quite hard to govern and standardize everything, so maybe I'll do decentralize, and then people worry, well, I can't really find any of my data because I don't know which organization owns it.

So I'm hoping a data fabric has got like the best of both worlds and hopefully not too many of the problems of either.

Mico Yuk: well, let's talk about data mesh. The fun thing about data mesh, it's another like big data, non IT pipe dream. Right? Like, I don't have a,

Richie Cotton: is

Mico Yuk: lying, Richie? It's the non IT pipe dream. It comes up at least once every three years. The last non IT pipe dream was the data lake. And do you remember Forbes article that later said that 80 percent of data lakes and big data ponds and sand dunes or whatever they were called, fail?

Richie Cotton: interesting. That's a. Scarily high statistic. 80 percent failed. Okay, go

Mico Yuk: I

presented on this on

Richie Cotton: what happened?

Mico Yuk: Yeah. you remember data lakes were the big hype. And then most people don't realize I did a whole podcast on this analytics on fire. It wasn't just data lakes. It went from data lakes to data ponds and data puddles. Did you know that there's a whole O'Reilly book on that? 

Richie Cotton: Wait, what's a data puddle then? Is that a really small data lake? 

Mico Yuk: I have a whole, I have a whole podcast on this. Like there's a whole book at O'Reilly that talks about data lakes. And I think I'm skipping one data ponds and data puddles. And essentially it's exactly what you're thinking. A data lake is big. Throw all the data in that people access it. A data pawn is a little bit smaller with a bit more control from an IT perspective and data pawns are tiny, more focused data lakes, essentially they're tinier that are more subject matter focused with access.

Richie Cotton: It sounds like someone's getting like very happy with the the jargon there,

Mico Yuk: Yeah, let me, I'm going to make sure I look it up because I know your listeners are probably like on Google right now feverishly typing in data pawns. I'm going to find that book for you. Go ahead and talk. I don't want to type on my computer, but I'm going to find a data pawn book for you.

Okay.

Richie Cotton: okay, while you're searching for things, maybe it's worth talking about how the data mesh works. So in theory, this is supposed to be an idea of it's just a process, there's actual technology involved, it's just each person's department owns their own data, and then they set up some kind of, agreement for how that data can be accessed by other teams, which in principle sounds very sensible.

It also sounds like an awful lot of work to set up just in terms of getting the processes right to do this. Do you have any sense of how organizations might actually be able to make that work?

Mico Yuk: Well, I go back to higher foundation. So I've been over the years just a little bit of background for me. I was a top 1 percent consultant. I started up in New York, work with , my core focus was fortune 500 companies globally. And so I've seen. the experiments of both centralized and decentralized approach to data teams, right?

So centralized being, we're going to bring everything into the data team, and then we're going to let people access, might have an analyst in the data team who's assigned to finance, assigned to marketing, you know, I kind of do this shared approach, but it's within. Then we started to centralize, which is, hey, data teams don't have enough hands.

We have to wealth of data and ownership. We're going to empower super users across these different departments and get them involved and get them to own. So data mesh to me is a swing and a pendulum. And by the way, I've seen both approaches succeed and fail, right? So know, it's not a pop.

Anybody's bubble again. But did you see the recent Gartner where they talked about data mesh height being over and the failure rate of data mesh?

Richie Cotton: It's already over, is it? No, I haven't seen this. Go on, talk me through. What's the failure rate?

Mico Yuk: just go on LinkedIn and look at data mesh dead or dying. I want to make sure I do not butcher this because I know this is sensitive and I do like Zamac a lot, so I'm just going to go ahead and make sure I quote this correctly, but just type in data mesh is dead in Google, but let me get the Gartner study.

It was done. A few years ago, innovation trigger. is data mesh obsolete? So Gartner predicts that a data mesh concept would be obsolete before it plateaus. That was 2022.

Richie Cotton: Okay, interesting.

Mico Yuk: Right. And I wanted, there's a number behind it. So I am here trying to see if I can get the number, but I don't know if they cited exact, yeah, they have a whole diagram here.

And what they're saying is that it doesn't imply that the paradigm will be absolute, but it might be supplanted by a different approach before reaching full maturity. People in Gartner, right? I used to sit on with analysts from Gartner, Forrester, Bark, all around the world, because I was considered to be an analyst, even though I wasn't, because it didn't know what to do with me.

Cause I asked a lot of questions. I think that , we've seen this before. I think Gartner has, and I think what they're saying is if you decentralize. And this again goes back to the IT freehook dream. If Teams is centralized and they don't do it properly. It's simply going to fail. There's just no good way to put it.

And the reason it's going to fail is you have three things that you can guarantee in our data life. Y'all going to love this, richie, you always hear that like, what are humans guaranteed, right? Debt and taxes. I tell my data students, I said, there's three things that data people are guaranteed.

Excel, SQL, and change. Okay, and you're not going to get away from this, and so the reason I bring that up is that with decentralization, one of the things that I've seen, and I appreciate what DataMesh is trying to do, is that decentralization works until it doesn't. So what does that mean? I gave it to the business, now the department changes.

People leave, the knowledge is gone, We didn't document stuff. All these things happen. And so, thinking that you're outsourcing it and handing ownership back, the problem is that when it breaks, guess who owns it?

Richie Cotton: Go on. Is that gonna be the data team?

Mico Yuk: Absolutely. That's the problem, is that the minute it breaks, Which it does a lot of times, what I've seen is it comes right back into the data team. So I think that while data mesh is powerful, I think that there has to be an approach where it's a shared ownership to where whether changes happen internally on the human side to the data team or changes happen externally, there's a level of accountability and ability to transfer, do KT, right?

Transfer ownership and be ready to adapt to that. I think if that change management is addressed, then these things can succeed. But until then, to me, unfortunately, it's a pipe dream.

Richie Cotton: Yeah. I can certainly see how if you've got a very decentralized system, you're gonna end up with different cultures across different parts of your organization.

Mico Yuk: People problems.

Richie Cotton: different, yeah, and then you're gonna end up with different Data formats and styles across your organization.

It's going to make it very hard to run cross team initiatives then.

Mico Yuk: and that's why I think AI is powerful. Once it can get security behind AI, if you have AI simply consuming every meeting, everything that you're doing, when someone leaves, , you just questioned AI, I have Richie's every call, his thought, his emails, everything. I'm like, Hey Richie is no longer here.

I got a couple of questions about this data model. So what, why was this created? This dimension, what day was it created? Why was it triggered? Oh, hi Miko. It was triggered because he had this meeting with Joanne. She requested this and this was the outcome. And then she said she didn't want this. And by the way, here's the artifacts.

So again, I go back to like, Where does AI come in handy? I hate to say it. It's not human replacement. I think it's human knowledge sustainability. So that's one creative use of AI that could fill the gap. But as it stands, this experiment of decentralization, It has been done before.

It has succeeded with mixed results, and I think that's why Gartner was trying to hype. And by the way, again, don't do it right now, but just go type in Google or LinkedIn, Datamesh is dead, and judge for yourself.

Richie Cotton: Yeah, that's absolutely fascinating. And I like the idea of having like the ghost of employees past knowledge management. So hard. TikTok

Mico Yuk: to do it. IBM charges people millions of dollars just for the documentation portion of a project. I remember sitting down and thinking, this is nonsense. Years later, in my mid thirties, they came out in early thirties and were like, documentation is dead.

And I was like, yes, yes, because think about it. How many documentation, nobody reads it. We can't even get people to finish reading emails that scroll on their phone. People do like this. They do two thumbs down. And after that, they're like, Oh, this is still going. TikTok next. Am I lying?

Richie Cotton: is very persuasive in that way.

Mico Yuk: Yeah. But they just keep swiping. Right. If it keeps going far less documentation, well, this isn't working. Make sure you open that word doc. That's like 50 pages long. Come on, man. That average attention span, Richie, is like three seconds. Absolutely. Absolutely.

Richie Cotton: And it's. Always like, oh, we've got a meeting in an hour, here's a 50 page document you need to read. So I can certainly see how AI summarizing that document is essential at this point, yeah. so you mentioned something very funny earlier that there are only three important things in life, Excel, SQL, and change.

Let's cover Excel because It's been around a long time. I think we can both agree. Well, at least my opinion is that for any given data task, Excel is never the best tool. It's just that you can do kind of anything in it. I think you're more of an Excel enthusiast. 

Mico Yuk: Yes or no, but the numbers speak. Richie, let's turn this to you. How many users does Excel have today?

Richie Cotton: Oh, man. I would guess it's got, it's got to be 100 million.

Mico Yuk: Richie, that was 10 years ago.

Richie Cotton: Really? Okay, go on.

Mico Yuk: One more try.

Richie Cotton: It's lower now, is it? Or, ooh. I'm going to say it's got a decrease in popularity in the last 10 years. I'll go 50 million.

Mico Yuk: It's almost a billion.

Richie Cotton: What? It's gone up.

Mico Yuk: Do you know, don't you realize that Excel is not on any BI quadrants? It's because it's galactic,

Richie Cotton: Interesting.

Mico Yuk: right? So it's 800 plus million users.

Richie Cotton: Okay.

Mico Yuk: Excel is basically the first level tool where you could do anything and I don't want to lie. The other day I was pivot table ing myself. Like I, you're going to laugh about this, complete side joke.

For 10 years I've talked about users that go to pivot table heaven and experience a dopamine release on like anything. I was in that heaven yesterday, right? It's just powerful when your pivot table works, your index matches work, the answer pops out, it's there, you feel glorious, you put a chart on it, and boom! Okay, so let's, as you can see, the dopamine is reflective. Anyways, let's get back to that. So, the reason why I think this is interesting, I'm going to turn this on you in a second, is that Back to that interview I did on my Analytics on FHIR podcast recently with the head of LinkedIn Learning Content Management for Data AI Analytics, I happened to ask him, what's the number one skill set that is requested by recruiters?

And he said, I know it's 2024 and you're not going to believe this because we couldn't believe it. Freaking Excel. Excel is back to what I said about data people is not going away. But I think what's interesting is I asked you this question during prep for data camp, right? So let's reflect back on data camp.

So data camp, 450 classes. 12 million people have been trained over nine years. What are you seeing as one of the number one requested skill sets?

, 

Richie Cotton: the largest audiences for DataCamp are people who are working in data analysis. In that case, SQL, and then also data scientists, and then it's a mixture of Python and SQL. So, these things, , they're not new technologies. They've been around a while. Python's been sort of more recently ascending to the crown of most popular language.

But yeah. These are the technologies that are requested by almost everyone. The thing is that for data analysts, it's like you need some kind of business intelligence platform skills as well. Power BI is the most popular, but there are a lot of these things, and it feels like most companies don't really care exactly which BI platform you have, just as long as you can make a dashboard.

Mico Yuk: Thank you for saying that. One of the things that I knew was going to go away years ago was , you need to be a specialist. I'm like, of what? I remember this, this guy it was a colleague in Australia, and he said something to me one day, because, you know, I keep storytelling and data visualization, he said to me, Miko, the bar chart is the bar chart in every tool, he said there's a line and there's bars, he said the only difference is in one tool, it takes three clicks, in another tool, it takes one click, He said, basically, there's a core set of charts for any visualization and report that doesn't change, it's just the number of clicks. And so I think, , the fundamental principle of people understanding, more importantly, what charts are, how to use them, how to put your data, is way more important than a technical skill set.

Richie Cotton: Absolutely. going back to your example with COVID, it's like, yeah, you need to be able to understand a line plot. Is this trending upwards? Is it downwards? Is there some kind of cyclical effect? All that same kind of stuff. This is just a fundamental life skill, being able to interpret like plots.

Mico Yuk: Well, data literacy, right? Back to what we were saying, data literacy.

Richie Cotton: Absolutely. And yeah, I think it's something everybody needs to be able to do. Whereas on the business intelligence side, well, yeah, knowing how to build a dashboard is brilliant because don't want to have to, wait for your data analyst or your data team to help you out, you can do it yourself.

The dream of service analytics is sort of realized that,

Mico Yuk: Well, again, self service analytics, which we didn't get into and won't get into is another pipe dream, right? Listen, anytime people, and I am not even an IT person, right? I have a technical background, a computer engineer by, thing. But I noticed that again, every , two to three years, there is a non IT pipe dream.

And I just sit and wait for it to fail. I'm like, Oh, self service, we're going to hand it to the business and go away. Until it breaks, and so I'm not trying to knock people's dreams. I know that data teams cannot keep up with the requests, but I think there has to be a different approach because all these rehashed approaches are not working.

And that's where I think things like AI could come in. But I know we're going off. I actually think the most successful self service experiment that ever worked is Excel.

Richie Cotton: yeah, you may be right there because Pretty much everyone can use spreadsheet, you can do simple analysis in there, and so. It sort of set the bar flat, like first line self service analytics.

Mico Yuk: I think the second most important experiment is the iPad. 

Richie Cotton: Go on, how so?

Mico Yuk: Kids, you give kids that device, Nobody has to train a three year old on iPad.

 You come back and, Mommy, they're showing you stuff. You're like, wait, what? Do you have enough teeth to be talking? What are you doing? They don't have to talk, right? I got, yeah, I got to YouTube. I see the video I want and now I'm dancing. I can't even talk. I got two teeth in the front. So we've seen it succeed in data world.

Listen, could get so deep into this. Data is emotional. one of the things I tell my data students before we start is I do this like kumbaya thing. , I start my classes by saying, if you do not want to be in the hospitality and sales and marketing business, particularly the TLC business.

Get out of data. You know why? Data is an emotional transaction, and you are a therapist. People get emotional about data, and if you're in the visual part of it, which is where I spent my career, it is hyper emotional and scientifically proven to trigger areas of your brain, like your cortex, that drives people to do irrational things.

Colors change emotion. It's like music. So I don't want to get too deep into it, but really implore people and say, listen, data is emotional. Make sure that you understand that that's what it is before you get into it, because if you think you're just going to feed people data without a story, you're not going to put any energy and emotion behind it.

You're not going to have to focus on how you craft what you're saying. You are going to be sitting in a corner for a long time and wonder why the data Promotion train as fast as you buy,

Richie Cotton: Interesting. , so, I guess you're saying that you can't just, show some stats there and be done. It's like you've got to build it into that story. And so, , the story on the data can really trigger people's emotions, and that's how you get into having some kind of impact.

Mico Yuk: which AI can help you with. So if you have a low EQ, which there's a number of data people that have low EQs, emotional quotients, meaning their AI needs work, which you can work on. I was one of those people. I've worked on my EQ for, I get tested every couple of years. I've worked on it for over a decade now.

then I wish I had AI because I go back to AI, which is like, Hey, have all my stats. I know what I want to say for this weekend. Here's my audience. I got a VP of research, a VP of this. Help me craft this narrative. And AI can help you to do that. And then you can practice it.

Richie Cotton: I think that's certainly something that's very common in the data world, people struggling with that sort of emotional interactions and understanding what other people want and their desires are and things like that. I'm actually curious to hear more about this emotional quotient, training.

, how do you get better at this stuff?

Mico Yuk: So for people listening before I say it, so people are not intimidated. When I first did my testing this is a cute short story. I was excited. So I have an amazing coach called G again. I interviewed G on my podcast. He met me, I was at a celebrity event, whatever, but he met me and we sat down and talk and he said, well, let me test your EQ.

Richie Cotton: It's intimidating.

Mico Yuk: Some of my scores may have indicated I was close to a sociopath. I was lacking in areas of empathy. I was lacking in areas of self awareness. There's basically areas where I just didn't care, partially because I was like a young person in New York. We're like walking on the streets with the construction workers being harassed.

You learn not to care, but that's a different story, which you live in New York. So, you know, I started in New York. Okay. So I think that affected me as a young person where I'm like, okay, I'm not being nice anymore, but either way with the EQ, what I learned that. What was helpful is Jay was like, but we have a path for you to increase these scores.

That changed my life. The Mikoyuk that's here today, that's on MSNBC, CNN, top 2 percent podcast, top keynote, Google, MIT, Meta, it changed my life. And so with the EQ testing there's a book called emotional intelligence 2. 0. It's been revised a couple of times. It is literally a Bible. I recommend that you read that book.

And then you can seek out Jay or another professional who has been trained and certified to test your EQ. Look at your scores and work with someone like Jay, which again, happy to put the information, to understand the path to increase your EQ. Because remember, one thing that I remember one of my mentors told me, and I appreciate this, I had two mentors, very intelligent, one dropped out of Harvard, the other one went to Brown.

And I remember them telling me one day when I was working in New York, there in Boston and Maine, They said, the person who knows how will always work for the person who knows why. Simple,

Richie Cotton: That's very interesting. Yeah, yeah. So knowing why that's your management role. Easy peasy. Okay.

Mico Yuk: right? And so knowing that why though comes at a level of emotion intelligence. IQ is more tied a lot of time to the how. EQ and IQ, and typically Harvard is on a study, higher EQ is actually tied to the why. Because on top of understanding why is being able to communicate it and then execute on it.

Richie Cotton: Nice. Okay. So, 

Mico Yuk: Sorry, rabbit hole.

I know. I love this topic.

Richie Cotton: No uh, it's a brilliant rabbit hole. And I think uh, this is like a good idea for personal growth for a lot of the audience. Do you have any thoughts on, like, what other soft skills are important then, related to that?

Mico Yuk: Absolutely. And it's not just my thoughts, Richie. I've interviewed I'm heading toward podcast episode 97. My podcast is primarily focused on data leaders and just through my own experience. I think there's a couple of things that data leaders ask and that I've seen through training. And I'd like you to also substantiate this with what is happening at DataCamp.

Number one is being able to ask the right questions. . Like I cannot even go ahead. 

Richie Cotton: I would say, yeah, it turns out like podcasting is good practice for this. 

Mico Yuk: Yeah. Like, and It's not just asking random questions or questions you feel, but legitimately being able to sit down, read the room and ask a simple question that opens up the door to the answers that you need. Because bad questions equal bad answers. And in our data world, it equals bad implementations.

Okay. Data pipelines of dashboards. I mean, it just trickles down into garbage can stuff. . And so I think number one, asking the right questions. Number two, I think is around storytelling. So the narrative. What we hear and what we see come together in our brain at different times, whether or not people realize this, right?

Like when I was younger, I remember my dad made me learn how to speed read. The whole premise of speed reading is you put your finger on a page and your fingers moving the speed of your eyes and your brain is catching up, but it doesn't miss anything.

I don't know if you know about speed reading, you can read up to 600, you know about that, right? 600 pages a day.

Richie Cotton: Yeah, the idea is, yeah, you can read a lot faster than you would

Mico Yuk: Right. And you know how it works. You put your finger and it's literally your finger is doing this across the page. And basically what's happening is literally when you study it, your brain is actually moving at a different pace from your eyes, but you don't miss anything. So if you take the premise of speed reading, I was like eight years old, my dad taught me that we take a premise of speed reading and you think of humans when you have a narrative and you can, it's an experience.

And so if you tap into understanding both the, what people hear and what they see, and you put those two mediums together, you can control any narratives. Being able to tell a story that not just drives decision making, but it drives impact, I think is one of the most sought out skills. Because at the end of the day, as a data person, you are always selling something.

Every time you walk into the room, if you're gathering requirements, you are selling the fact that you are trustworthy and can drive the discussion. If you're introducing a new tool, Power BI, whatever it is, you are walking in and you're selling them to trust you to implement this tool with data that they're accountable for, that you're not going to be in the room when they have to answer those questions for.

 The ability to garner trust through the story that you're telling is super valuable. So that's number two.

Richie Cotton: Absolutely. I like that. So, storytelling and being able to gather trust. Yeah, also interviewed a lot of executives about this, I think I've only ever had, like, three answers on, what's the most important, skill. One of them is communication skills, which I think, like, Very close to the storytelling idea.

Secondly, collaboration, which is close to this sort of gathering trust idea. And the third one is ownership. Just having people being sure that, you can go and do stuff without having to be watched over.

Mico Yuk: Confidence. It's the confidence, right? So even coming into the room, you know, I get into it because I coach executives as well. And I talk about just how they present themselves and their voice. And so case on point, I could tell you, for instance, when I was on CNN, I remember this is a good story for everybody, right?

So I wear my glasses because my eyesight decided it wanted to decrease. I came on CNN and I had a brilliant idea one day, and this was my look. And I came on CNN like this. Now we have a 10 minute prep session. And I remember literally being told, wear your glasses. I'm like, I'm good. I can see without them today.

And they're like, no, you have a look and your look is a smart data commentator who is helping America understand how to save and spend their money during a pandemic. Wear your glasses, Richie. I had to pull those. I mind you, I'm a data storytelling professional. I put them on, you know what they said to me?

You have a look that people have been married to and your glasses are a part of that. We're not changing course right now.

Richie Cotton: Interesting.

Mico Yuk: Very interesting, right? And so, I sit at all to say data people, just even consistency in how you present, how you speak, how you listen, how you watch people also being able to read the onset things in a room.

There's the obvious things like this, right? What does this mean?

Richie Cotton: Let's check them down. That means we've been talking too long.

Mico Yuk: Exactly. And then what does this mean? 

Richie Cotton: Yeah. So that means I've probably come up with a terrible idea and you're uh, trying to find a polite way of saying no to me.

Mico Yuk: Correct. And so these are the obvious things. There's a book called You Can Read Anyone. It was by the former head of the CIA. That book I've read 10 times. It took a cheap 3 book because people who it was intended for don't need it. I want to give it out to every data person that exists. Like I think the onset skill that I would add to that two of them is being able to read a room and also curiosity.

So I think one of the things that happens to as well is . You know, data people have two extremes, they either go down too many random rabbit holes, or they don't go down enough, so I think striking the balance of where your curiosity serves your users is where you find data people that are super successful.

So I think hopefully we covered the spectrum, but I think again, going back to the EQ, I know for me, my journey is that once I got that check, It's like I almost, you know, my, my coach for me, it's like I checked in.

Richie Cotton: I like that. So is you getting some data driven feedback in terms of are you improving?

Mico Yuk: Absolutely. And the feedback is detail, right? There's scores and they basically analyze all of your results. And then my coach interprets it. And then there's a path that's created to see here. It is because you're going to work on, and it is hard work with you. I have to tell you, like being in New York and having to care about construction workers and be more empathetic when I really want to tell them what I think, because they're staring at my rear end.

It takes a lot of empathy, but you have to make sure that you practice this on a daily basis. And really just understanding that type of stuff and practicing it regularly. It's a skill set, but it works because it reflects in how you approach people.

Richie Cotton: This does sound like exhausting training, but also very rewarding if you can pull it off. Yeah, be nice to construction workers. Well, I'm sure for any construction workers who are listening, yes, we should be nice to you as well.

Mico Yuk: New York construction workers, Richie, come on. I worked in Manhattan. You know, come on. You know what I, you know what I'm talking about. They're not normal.

Richie Cotton: Yeah, I don't get wolf whistle that very often, but I understand what you're talking about all right then so before we wrap up have you got any final advice for data people?

Mico Yuk: Loving data is a start. Loving your users is more important. 

Richie Cotton: Nice. Okay back to the empathy. That's that's very good All right.

Mico Yuk: tend to find that if you love the users more than you love the data, it goes back to something that we all have heard, which is don't chase money, chase passion, and the money will follow. I feel like with data, that if we truly invest in our users, and we build that trust, that understanding, we invest in ourselves, that the data skills on the end, all that technical work that we've done, can actually truly be realized.

I don't think it's the other way around.

Richie Cotton: Yeah. Yeah, I think if you just start with let's do something cool with data You'll get somewhere but you don't get all the way. So yeah, I like the idea of care about users and Something good will happen. All right. yeah. It's been a pleasure chatting with you, Mika.

Mico Yuk: This is so much fun. Yeah. And one more thing, don't sleep on Gen AI. I know that we're not here to do hype cycle. I mean, me and Richie talked about that. we're going to be honest about it, but the reality is that it's coming. You don't need to, Think about a robot taking your job. Let's not get silly here.

But at the end of the day, make sure that you understand it because it's going to be one of those things where the users are going to show up at your department. And you do not want to say at that point, Oh, we're figuring it out. You know what that equals? It equals a consulting firm. Not knocking my consultants, but at the end of the day, what users do when they hear stuff like we're figuring it out, we're just learning, we're in training, is they go to Google and type in Gen AI specialist for financial department.

Richie Cotton: And then, yeah, they've got the answer immediately. Okay.

Mico Yuk: then they got the answer immediately. So don't sleep on it. Don't go crazy. Do again. I gave you guys steps on how to approach your learning. Just make sure that your answer is absolutely. I've already upscale Nico and Richie on it. A matter of fact, they have a little sandbox going.

Let's look at what you have. We absolutely could do it. Just be ready.

Richie Cotton: Excellent. Uh, Yeah. Scout's motto, be prepared. I like it. All right. Pleasure chatting.

Mico Yuk: This was fun.