Latest / The Unpack Show with Anandh / Ep-8: Kai Feng - Head of Data & AI Strategy | Thoughtworks
Transcript
- TheUnpackshow: Hey, welcome everyone. Welcome to another episode. Today's topic is going to be an interesting one, because I have a special guest who is an author who has done a lot of work in data and AI space. So the topic for today is human side of data and AI. ⁓ So we often talk about data and AI as if they are purely technical things like numbers, dashboards, algorithms, automation, right? So you name it. But at the heart of it, data is about people, their choices, their feelings, their behaviors, and most importantly, the stories behind those signals, right? ⁓ And nobody brings them out, ⁓ brings that human side better than the special guest that I have, a great data storyteller, ⁓ a great musician. ⁓ We'll hear about his journey, ⁓ who blends curiosity, humor, empathy, and creativity into the field. So in this episode, we will go deeper into how we can bring more heart into data and more humanity into AI. So insights connect, stories matter, and technology truly supports people ⁓ rather than overwhelm them. So without further ado, ⁓ I want this special guest to introduce himself. So why don't you introduce yourself, Tiankai Feng: Yeah, thank you so much. And it's great to be here. Hi, everyone. I am Chiang Kai. And I work currently as a director for Data and AI Strategy at ThoughtWorks, which is a consultancy focusing on bespoke tech solutions. And I'm very passionate, as you mentioned, about the human side of data and AI, because I realized that over the 12 years of working in that area, that most of the failures in data and AI are rooted in human factors. So basically, it's usually a lack of competence or a lack of collaboration, ⁓ some misunderstandings and all of those leading to bad decisions that are finally then represented and have consequences in bad technologies that are not working. So my goal, I hope, is to evangelize a little bit more the human side of data and to not see it only as a weakness or a wild card of data and AI, but see it as a real superpower to be able to be empathetic and TheUnpackshow: Yeah. Tiankai Feng: pay into basically human growth and empathy to drive data to success. I'm also ⁓ passionate about making music. I think you mentioned that already, right? So I like to use music to make topics also around data and AI more fun and approachable. TheUnpackshow: Right. Yeah, we would love to hear a song, but we will get to that. ⁓ So without further ado, Kai, like I know you have done a lot of this work in this space and you've written two books, right? The Human ⁓ Data Strategy and Human AI Strategy. I read the Human Data Strategy book, reading the Human AI Strategy book. Thanks for writing those books. ⁓ When we talk about data, people often talk about like, okay, hey, let's look at the problem and bring in some tools and technologies to solve the problem, right? Tiankai Feng: Thank you. TheUnpackshow: It's always the case in our knees in my experience. That's what I've seen. So we treat it as something cold or mechanical, but you have always approached it as something emotional, right? Something connected to real people. So what does this human side of data mean to you, right? I would like to hear it from you because you wrote two books and you've been doing this for a long time. And most importantly, why do believe it matters in a world obsessed with metrics, technologies, tools, and you know, the new shiny objects? Tiankai Feng: Yeah, it's a great question. I think the really one key thought I have about all of this is that we work in data and AI to help humanity, right? I think this is something we should all inherently agree on. And if we agree on that, working in data and AI is supposed to help people, then why are we somehow dealing with it as if it's not important to help people, right? Why is it then about either only prioritizing one group of people, ⁓ compared to the others, or why don't we think about people at all and we only think about the cool stuff about technology of what is possible or not. And that leads me to then to think about why we actually should then be more human about it. And a lot about it is that ⁓ all around data and AI, we have people that work on it that are human beings, people that use it that are human beings, people that are customers from it that are human beings, and people that are leading and kind of guiding it that are all human beings. Then let's treat them that. TheUnpackshow: Yes. Tiankai Feng: Right? Everyone should be motivated to have purpose behind it. Everyone should be able to work with each other and understand each other. And everyone together should then also get the recognition and the feeling of achievement together by having done it. But too often it happens the opposite, right? Where one group of those people is feeling unheard, is feeling hurt by it, maybe feeling misunderstood. Other people getting a direct disadvantage from it and might even lose their way of doing things because of the other. TheUnpackshow: Yes, yes. Yeah. Thanks. Tiankai Feng: groups making a decision. So I want to bring that back. And this is why I wrote those two books, also ⁓ Humanizing Data Strategy, Humanizing the Asteroidity, to bring more focus onto the human side again. TheUnpackshow: Right, right. I think you well said, Kai. People often talk about the narratives, right? So for you to connect with a human being, you need to have an impactful story so that it can resonate well with people. know, oftentimes I've seen people have all the data, but they don't act on it because not because the numbers are wrong, but not because of the fact that they cannot tell the story. They cannot feel the story behind it. I would like to get your perspective on like, can you share a moment when a story really helped a team? truly understand a difficult insight or finally taking an action because that's a critical idea of like what we do in the data and insight space, right? ⁓ People talk about, okay, I have all the data in the world, but if you're not acting on it, if you're not telling the right story behind what we have, it's kind of a moot point. Tiankai Feng: Absolutely. I think ⁓ the most impactful example I have is from data quality management. So what happened basically there was that when I was working on this one project, we basically were talking about product master data and how the data quality of product master data was not well maintained, meaning that when it was showing up in an e-commerce store ⁓ for customers to buy it, TheUnpackshow: Mm-hmm. Awesome. Tiankai Feng: They couldn't really find it, right? It was tagged in the wrong way, categorized in the wrong way. So there was no way of selling it in this way. And ⁓ it would lead to, of course, the sales team that was working on the e-commerce job to manually change it. But we all know that manually cleaning it downstream is not really curing the root cause of the problem coming upstream, right? So it became that continuous, tedious effort to always clean up the data over and over again. TheUnpackshow: Okay. Tiankai Feng: And initially when we talked about the data quality of it all, we just said, ⁓ the data quality is expected to be this, but we're doing here. Can you please start now correcting the data? And they would say, that doesn't mean anything to me. Why should I do it? I have other more important jobs to do. Why do you want me to just clean up the data? I have other things that I need to work on. And we realized that when we actually put ⁓ human empathy behind it, meaning we show actually and let the people talk that are cleaning up the data. TheUnpackshow: Yeah. Tiankai Feng: and how much effort they spend invest into it and let them talk to the people that are producing the data. They were activated by the empathy, right? They realized that they are actual human beings that are because of that, wasting a lot of time in their day-to-day job. And by me just investing five minutes, they can basically avoid ⁓ three hours of work per week to have to do it. So maybe that is the right thing to do. And by just bringing people together and putting faces behind the stories. TheUnpackshow: Yeah. Yep. Tiankai Feng: the story itself was more impactful, right? And that was what led to finally data quality increasing and thereby also the business outcomes increasing in the right way. I feel like it's often the story that needs to be relatable, right? And empathetic for people then to actually get the call to action and to do something. TheUnpackshow: Yeah, it's. Right, I think it's a powerful message. Oftentimes people ⁓ kind of discount the fact that you need to talk to people, right? Data teams often work behind the scenes and rarely meet the humans representing the dashboards they build, right? How do you help these analysts? I know you've been in the space and you talked about this e-commerce example, but ⁓ what is the message you want to provide to the analysts or the people who are working behind the data? so that they can connect to the people, ⁓ their needs, their emotions, their journeys are going to be important, right? Rather than just creating a dashboard out of it. Tiankai Feng: Yeah, absolutely. I think my key advice would be to make the effort of understanding your collaborators, right? Because I know that most people are either introverted very much, so they don't like to talk to people, so they rather just do it on their own and hope for the best. But too often that doesn't work. Or they are very respectful, almost too much respectful of other people's time. They're like... TheUnpackshow: Yep. Mm-hmm. Tiankai Feng: ⁓ no, I don't want to spend one hour with that person. They have probably more important things to do. I'm not going to bother them. Let me just assume what they think. But none of them leads to the goal, right? Because by investing that additional time to understand the other person and asking how they make decisions day to day on how they actually use data day to day, then what you actually then create as a dashboard or report or whatever, it's much more tailored to what they actually want rather than you just hope that it's something that they want. So I would always say that TheUnpackshow: Yes. Yeah. Tiankai Feng: making the effort is worth it, right? Understand your audience and your collaborators first and then to the job. Never try to do it the other way around because otherwise you're going to waste a lot of time. TheUnpackshow: Yep, think it's a good segue for us to get into the book that you wrote, the Humanizing AI Strategy. So AI can draft, design, write, and summarize. In my view, it can be both exciting and unsettling for creative people. So I'm curious, when you wrote the book, how has AI influenced in your creative process? Tiankai Feng: Mm-hmm. TheUnpackshow: Before you release the book, often ask for like, hey, whether this title image makes sense, right? You give options, right? I see a lot of creative process and you're using AI to influence those creative process. I'm curious to know because you wrote two books, does it open doors? Does it sometimes challenge the way you express yourself? Like, is it helping? What does the journey looks like? Tiankai Feng: Right. Yeah, that's a really great question. I think there are actually two different stories ⁓ to each of the books because for humanizing data strategy, I purposefully decided against using any AI because I wanted to prove to myself that I can be a writer. Right. So it was like, I can do it without any help. I'm going to just push it through and write it on my own. And it worked. But the biggest challenge wasn't not using AI. It was more my own self doubts and TheUnpackshow: Okay. Yeah, yeah. Tiankai Feng: not being sure if anyone would actually read it even at all. Right. So it was a lot doubting myself and challenging myself and trying to do it. But it resonated so well with the community that I was very confident for the second book. And since the second book was about humanizing AI strategy, I felt compelled to actually use AI to showcase for others how it would work to be creative using AI. And specifically, it was very effective in two sides. One was proofreading. TheUnpackshow: Okay. Sure. Tiankai Feng: Right. So very good. You just put it in and it tells you all of the mistakes you made and how you can improve it. Grammar and writing and typos and all of that immediately gone. The other thing of it though, was ⁓ I call it removing writers block. So even if I knew what I wanted to say with this specific chapter, it's usually very hard to start writing that chapter. Right. So like what is the first sentence supposed to be? How do I get started now? So I asked basically to spar with Claude in my case, to ask TheUnpackshow: Yep. Yep. Yep. Yep. Tiankai Feng: What do you think could be a good starting point? Give me like 10 options. And then I would look at the 10, I would pick like some of the elements of the 10 ideas and I would come up with my own. So it removed that writer's box so efficiently that I was much faster in my writing process, right? Because I just had impulses all the time to continue. So although all the creativity, creative output was still mine, it did give me a lot of energy and a lot of motivation to just keep going. TheUnpackshow: Okay. Nice, nice. I think that's the humor and humanity part when we deal with data and analytics. I've seen you use humor to make complex data topics feel simple and human, right? And that's what I've seen in the conference that you attend or that you speak, even in the LinkedIn post that you post. Why do you think humor is such a powerful tool? I think there's a thin line, right? You know, can go either ways in terms of data storytelling, especially in the technical people background, right? Tiankai Feng: Yes. TheUnpackshow: What have you seen happen when people feel emotionally safe enough to laugh in a technical space? Tiankai Feng: Yeah, that's a great point. I think that humor is ⁓ not only a measure to laugh, right? I think humor, as you said very nicely, is creating an environment of safety too, right? Because by being able to laugh about something, you have to feel comfortable to laugh and show a little bit more of a vulnerable side of yourself even, maybe, because you are admitting that you find this funny and you let your laughter out. And for me, TheUnpackshow: Yeah. Yeah. Tiankai Feng: Humor is a really important one and also an indicator for confidence, right? Because if you are able to make a joke about something, you better be confident about what the joke is that you're making and the subject area that you're making a joke about. And if other people can joke with you about it, kind of elevates that subterranean expertise into something we all can share is, right? And by all together, I was making fun of it. It's more fun. And although it's a serious topic we work on, we can still have fun while working on it. TheUnpackshow: Hmm Tiankai Feng: And that creates then in hindsight also more motivation and team spirit, et cetera. TheUnpackshow: Yep, Oftentimes we talk about data as a bridge, data is the tofu, data is the new bacon, right? So there's a lot of ⁓ phrases that you can use. ⁓ Data should act as a bridge, not as a barrier, right? Data can unite teams by offering clarity ⁓ or create tension, right? When interpreted differently. ⁓ I'm sure you might have gone through a lot of these things, like when people talk about data, you might get multiple definitions for a single data point, right? Tiankai Feng: Yeah, absolutely. TheUnpackshow: ⁓ And it's hard, right, if you have multiple definitions. How can, you know, from your standpoint, how can leaders use data to create a shared understanding and alignment, right, instead of causing confusion and conflict? Tiankai Feng: Absolutely. Absolutely. I do think that data is having often the goal to be the source of truth, right? That's what many people try to aim for, that by creating reports or a database, this now acts as a source of truth. And we often compare it as the opposite of gut feeling, right? Like having known the truth and what actually happened is always better than what you feel might have happened or your own opinion. And ideally we can basically connect with you. So what I think leaders actually can do TheUnpackshow: Yeah. Right. Tiankai Feng: is to frame these data points that you have or any kind of data related work into something that, as you mentioned before, something that brings people together, right? As in like from ⁓ those data points that we see here together, we can derive that for you in the marketing team, this means actions ABC, for you in the sales team, that means actions XYZ. And for you here in the data team, that means we need to evolve in topics one, two, three. TheUnpackshow: Yeah. Tiankai Feng: And that is a nice thing to do, right? As in like this one data point showed us how we can all do better and how we can all elevate this into more business impact. And that is, I think the right way to do it. But I think the opposite of it, and that is also where the other side of the question is, is what if we use it ⁓ on purpose or ⁓ not on purpose in the wrong way, right? Where we use it to blame others or when we use it to actually police others, right? The worst thing to do is... TheUnpackshow: Right. Yes. Yes, yes. ⁓ Tiankai Feng: You are lying to me because the data says ABC, why did you do that? It's much worse than together looking at it and at least just clarifying it and agreeing on what the truth is. And so it's a lot about how you communicate it around too, right? You can use it to offend others or to question others. You can also use it to build trust and build a relationship actually. And this is what we should aim more. TheUnpackshow: Yep. Right. Right. Yeah. So, Kai, looking ahead, right, and I call this as keeping the soul in the system. As AI becomes more capable, the human role will change, right? It won't disappear. think definitely the human role will change. But if you imagine the next few years, what does a healthy human-centered collaboration between people and AI look like to you, right? What gives you hope for the future? Tiankai Feng: Absolutely. I do think that what ⁓ AI and specifically AI agents are really good at now is ⁓ being delegated the more tedious and repetitive tasks that human beings are doing. And ⁓ I don't mean only the most basic automation, but also where simple thought processes are flowing into it, where a human being had to categorize or make a choice and then do actions ARB, depending what happened. TheUnpackshow: Yep. Tiankai Feng: to then do it as a workflow. And those workflows can and should be all as much as possible agentic in the future. Meaning the question is what is left for human beings, right? And I believe in a very optimistic way that that leaves us human beings more with what makes us human, which is original thoughts and critical thinking, right? Coming up with the best ideas possible to really dive deep and connect the dots in different ways and then come up with the best solution that we can. TheUnpackshow: Mm-hmm. you Sure. Tiankai Feng: Because now that we don't have to spend all that time executing ourselves anymore, since we have agents to do it, we should all spend more time thinking and evaluating and actually analyzing and then decide for the right actions. And I hope that is the right way of collaborating, right? Where let's say, I heard someone really smart say once that AI agents can be like really smart interns, right? So basically you are still the boss, but you use AI agents as someone who is... TheUnpackshow: Yes. Tiankai Feng: learning from you and also continuously doing things for you. That should be the ideal scenario. TheUnpackshow: No, absolutely. think a lot of our time will go in designing and thinking and be creative in how we use the ⁓ AI and most importantly, how we build agent-ic AI. Any of the strategy and planning and execution of it can be handled by agent-ic AI, but the designing part should be done by human. ⁓ Kai, when I look at today's conversation, I think it is remaining ⁓ as something simple but powerful. Data doesn't become meaningful until people give it meaning. AI doesn't become responsible until humans shape it with intention. ⁓ As the technology evolves, our job is not ⁓ to become more mission-like. It's to become more human, more curious, more honest, more empathetic. I think you said it well. Tiankai Feng: Exactly. TheUnpackshow: ⁓ Thank you so much for helping us to see the future of data and AI, right? It is not just smart. It can also be a kind, creative, and deeply human. Thank you so much for taking part ⁓ of this podcast to show Koi. Tiankai Feng: Absolutely. Thanks so much for having me.