บทถอดความฉบับเต็มของ Stop Piloting AI. Start Leading It With Charlene Li

Well, AI is a technology and a strategy is about a vision of where you want to be. I don't think you should ever lead the word strategy with a technology in front of it. You want to plan for what you want to do with AI, but it should be in support of your existing business strategy. It may even change your business strategy and influence the direction of it, but you have a business strategy. So be really clear what you're trying to achieve. and figure out how to use AI in service of that strategy. And so I think we conflate and we get this bright, shiny object syndrome when it comes to AI. Look how cool it is. Let's focus on what we can do with AI. And you can do a lot, but strategy is about making choices about what you do. Sometimes very tough choices, and you need to do the same thing with AI. What are you going to do with AI to support all the strategic choices and options and challenges that you face? Morag Barrett (04:19.533) So what are some of those choices that you, that leaders are wrestling with right now that you are helping them to navigate through? Charlene Li (04:30.232) Well, the reason we wrote the book is because people wanted to know, executives want to know, where do we start? How do we create value with AI? There are so many things being thrown with us. How do we actually begin this process? Because we've given people Microsoft Copilot or even ChatGPT or Claude, and we're not seeing value. We're seeing some incremental wins here and there, but we're not seeing that promise of transformational value. Where is it? Where do we? How do we do this? And I think that the biggest issue is we're focusing again on the technology and not the value that it can potentially create. Morag Barrett (05:10.797) So you touched there on the perennial challenge that this new technology has been thrown over the fence at leaders, at team members. I've been reading the headlines about thou shalt use it and then crickets. So what's the biggest barrier then between the pronouncement of thou shalt use it and people actually having the confidence to do so. Charlene Li (05:28.27) Bye. Charlene Li (05:37.038) And so again, we're throwing this out there in the context of tremendous fear and anxiety. In the US, only 39 % of people believe that AI is going to be more beneficial than harmful. That means the vast majority of people think it's going to be a bad thing. And why should they think otherwise? Look at our history with technology, especially social media. Why should we trust these people? They're the same characters. And so we're throwing it out there and saying, they'll also use it. But for what? What are we going to use it for and how do we make sure we're safe? And you're asking us to use it and to train these agents even sometimes we're training it to take my job, right? Is that what you're asking me to do? And so in the absence of a strategy of how you want to use AI to achieve your objectives, these executives have said, okay, here's the

technology, use it, but they're abdicating. the responsibility to be leaders in this space, to create foundations and direction and scaffolding to help us through this bridge of fear and anxiety, to say, this is what we're going to do. This is where we're headed. This is the path we're on. This is how you know you're being successful. We're just giving them nothing. And instead of saying to our CIO, make something happen with AI and the CIOs, what do want me to do with that? So this is significant disconnect. We know what we should be doing. We're not leading and we're not telling people what to do. Morag Barrett (07:05.353) It's- Morag Barrett (07:09.485) So it's interesting the speed with which AI has infiltrated at least explicitly everyday life. It was there implicitly in some of the technology that we as consumers or I might have used when I'm phoning and it's helping to route me through a call tree, but right now it's front and center. I think would be hard push to find people who hadn't heard of chat GPT or Claude. maybe still hard pushed to find people in a corporate setting who haven't at least dipped a toe in the water. But what's surprising to me is two sides of the same coin, the number of sudden experts, the self-proclaimed experts on this topic, not you to be clear because you've got deep research here. And also on the other side of the coin, the leaders who are asking their organizations to adopt it, but they actually know nothing either. So how do we navigate the what do we believe, who do we listen to and how do we bridge that gap? Charlene Li (08:27.598) Such a great question. Such a great question. And this is something we struggle with. Again, where do you begin? There's so much potential here. We recognize it and see it's just out of reach, and it's through this gray mist. How do we get to that point? How do we clear out that point? We believe you have to start with strategy, but you also need to start with just a basic understanding of what AI can do and what it cannot do. Morag Barrett (08:34.957) Mm-hmm. Charlene Li (08:58.156) And we spend the very first chapter of the book just explaining what AI is, giving you some exercises, explaining what generative AI is and how that's different from the old AI and what is this new thing called agentic AI? How are they different? And what are their limitations? So I think AI fluency is what we need to first develop because it's very difficult to trust AI to use it if you don't understand how to use it. And we're talking about more than just literacy. So understanding what it can do and what it can't do, understanding how to use it in a responsible and ethical way so you stay safe because you're not going to use it unless you feel safe. And then very importantly, learning how to use it to get your job done. And this is the most important part. Fluency just doesn't happen from a training or a presentation deck or listening to a bunch of YouTube videos. It comes from practice. And the analogy I have is that if you have Morag Barrett (09:51.053) Mm-hmm. Charlene Li (09:56.69) If you ever learned how to use chopsticks, you didn't grow up in a household that use chopsticks intuitively. Then the first time you use it, it's like, how do you use two sticks to pick up food? And then you practice it and it's awkward. And eventually one day you don't even think about it. You're just using it until the day you hit a slippery noodle and you go, I have to adjust my grip on how am going to do this? So I think AI is the same way. There's no substitute to practice. And most organizations are Morag Barrett (10:05.933) Mm-hmm. Charlene Li (10:25.11) sometimes giving us tools, oftentimes not giving us the training or if it is, it's superficial and definitely not giving us the time and the space to experiment, to learn. And that's very different from pilots and we'll talk about pilots in just a second. But I think our biggest mistake right now is that we are not, we're expecting people and ourselves to understand this without having invested the time to practice to develop that level of fluency. Morag Barrett (10:54.839) Well, you mentioned there, mean, AI, we use it as a collective noun as if AI is it, that there is one AI to rule them all. But you already mentioned that one of the biggest misunderstandings is that there are different flavours of AI. You've got the agenic, you've got generative and predictive. So let's just do the very high level. What's the difference between the three? Charlene Li (11:19.118) Well, predictive is something that's been around. It's the Netflix movie recommendation. It's the Spotify building. So it's taking past behavior and predicting what's going to happen. Generative AI is trying to figure out, it's guessing what is the best way to put different types of content together. So it could be words, could be images, music, code, video, and it's generating and creating it. So it's the creator. And agentic AI, Morag Barrett (11:23.957) Mm-hmm. Charlene Li (11:47.234) does things on your behalf. It literally is your agent out there in the world doing things for you, in many cases automatically, and you set the limits of what it can do. So they're flavors of the same base algorithms of artificial intelligence, but the way they show up and how we use them is very different. Morag Barrett (12:06.571) curious how are you incorporating AI into your day-to-day work? Charlene Li (12:13.474) Well, this morning I woke up and my Claude co-work and code had generated a daily briefing for me showing me this is what's happening. It had created notion pages for every single one of my meetings with notes, gleaned from the calendar events and the past emails and other notes that I may have taken and transcripts of previous calls. I went through my email and found all the things that I had not done. I looked at my to-do list and said, you have these three things. You don't have enough time today to do all

those three things, which are your priorities. Let me adjust your schedule. So that's how I'm putting it to work. And it's like having my chief of staff sitting here and understanding how I'm working and adjusting and helping me adjust and make decisions and draft emails. not necessarily executing. have limits on what it can do. It will never send an email on my behalf unless I have read it and reviewed it. But it can draft it for me, go through the past email trains, do the research, and generate it for me. So those are just some of the ways that I begin my day with my AI chief of staff helping me get ready and prepared for the coming days. Morag Barrett (13:27.545) well, you are my inspiration. Because literally yesterday I sat down and opened up Claude Cowork and was trying to understand how is this and I haven't yet got the mental map to understand, OK, how is this different to chat and projects and when and how would I use it? But this comes back to your comment around practice. It is also about having inspiration and others who in my learning style can help show me and talk me and give me a picture of how this could help, because then I can go and research it. But you're saying pilots can be the death, well of any new technology, but essentially of launching AI too. So tell us a little bit about, don't pilot, but practice. Charlene Li (14:09.558) Yeah, well, everyone has been doing pilots and a research study came out from MIT last year that show that 95 % of pilots fail. It's a very high failure rate. And by definition, pilots are designed to fail because they're limited in scope. They're one off. Nobody takes it seriously because it's a pilot. So in order to test it, I needed to connect it to data inside my company. So I go to IT like, can you connect my pilot? Morag Barrett (14:18.123) Hmm. Wow. Morag Barrett (14:30.477) Mm-hmm. Charlene Li (14:38.712) to our data, like, no, it's a pilot, we can't do that, we can't expose the data, come back when it's a real project. I go to marketing and say, hey, I need to go and test this with customers, no, it's a pilot, we don't wanna expose customers to this. You're trying to get people to work across silos to give you permissions, and because it's only a pilot, no one's going to put their effort into it, they're gonna do the real work that they're being paid to do. So pilots are designed to fail. And so our belief is, Morag Barrett (14:49.805) Mmm, yep. Charlene Li (15:07.598) Pilots are, again, an application of leadership. We don't really want to commit resources because we're not sure it's going to be important. So we'll give you permission to do a pilot, but we're never going to put real resources against it. And so it's a way to procrastinate making these decisions. Instead, we say, figure out what are your top strategic goals? Really, which are the must-do, must-succeed, mission-critical things we have to do? And then look and say, How can AI help us achieve it better, faster, cheaper, even safer? And then don't make it a pilot, say, we're going to do this. And the challenge is leaders don't like to commit to things that they don't know whether they're going to be successful or not, which is why they do pilots. But pilots will never win. And if you think that this is going to be successful, we will find a way to persist and make it happen, or we will pivot. We'll figure out this is not the right way to go. We'll find another way to make this work because it is mission critical to us. It is strategic to us. We won't get this done. So no readiness assessments because what does readiness assessment show? You get a score of 62. So what, right? Feasibility, it doesn't matter because nothing is feasible at this point. You've never done this before. Of course it's not going to be feasible, but if it's strategic, Morag Barrett (16:20.397) it would not. Charlene Li (16:32.735) then you will put the resources behind it. You will find the people internally or externally. You'll figure out what technology needs to happen, and you will make it happen because it is very important to the organization. Morag Barrett (16:45.197) So you've given some great examples of how AI exposes poor leadership. So what does good leadership actually look like in an AI era workplace? Charlene Li (16:57.752) Well, we actually have an entire chapter, the last chapter about leadership because we believe leadership is the accelerant to AI transformation. And the number one thing is that the leader is able to expand the scope and understanding of what they are doing with AI. They're using AI themselves to become better leaders. One of the biggest challenges that leaders have is just Morag Barrett (17:17.325) Mm. Charlene Li (17:26.112) understanding what is going on, your scope of understanding of influence, what's the truth out there, whatever that truth could be. And having it just at your fingertips compiled in a way that you can understand so you can make decisions on is a huge advantage that AI has. And you can ask other people to use AI to create that for you, but it is the expectation. that you set as a leader that people are using AI to help them and to help you be the best leader you can be. So the little tip we give to leaders is in your meetings as people, it can be one-on-one or in group meetings, how did you use AI to prepare for this meeting? So instead of saying you must use AI, specifically, how did you use it for the particular agenda that we have in front of us? And even better, if you're modeling the way, you can say, this is how I use AI to prepare for this meeting. And that's an open question, but it sets the expectation that we will use AI in our work in a very specific way to help us achieve the goals that we have set in front of us in a very concrete, practical way, prepare for the decisions we have to make in this meeting to prepare for the discussion. And again, AI can also help you just think. very strategically, can

help you just look around the corners, be that challenger, be your thought partner and all of these things. Again, being a leader can be very, very lonely. Who else can you talk to about these things? Is it this, is it that, are you always just weighing the options? And AI can just help you think through these things in the way that a board of directors can do, but it's available 24-7. And then I, yeah. Morag Barrett (19:06.103) Mm-hmm. Morag Barrett (19:15.273) Yeah, I mean, using it as a sounding board is a is a powerful tool, but only just scratching at the surface as to what it can do. And what I'm seeing, obviously, in the headlines is it's already having an impact on the workplace, on potential career structures, as it inevitably removes some of those entry level roles. So what are you seeing in terms of those workplace dynamics and relationships within teams and within organizations? Charlene Li (19:44.974) Yeah. And again, I think this is a key thing that leaders have to be aware of. If you were to say to somebody, don't worry. AI is not going to take your job. First of all, they don't believe you. So why do you even say that? And you've never been able to guarantee somebody a job. So why are you doing that now? I think a more rational and a good point of leadership, because people just want two things from their leaders. They want you to be honest and they want you to be fair. Morag Barrett (19:55.458) know. Morag Barrett (20:13.261) Mmm. Charlene Li (20:13.538) So show up in that way. And the better way to respond to that is, say, I'm going to take my job. And you go, I really don't know. That's the honest answer. I do know it's going to be very important. And hopefully you can say the following. Because it's going to be very important, we are investing in our people. We want you to be AI fluent. We're giving you the tools. We're giving you all the training from the top down and also the ability to learn from your peers and from outside experts in your field. And we're giving you the time, especially the time to set aside all the demands of your job and to carve out and protect time for you to learn and experiment, to fall on your face, to share your successes, but just as importantly, the things that didn't work out and how you learned. Because that's the only way we're going to proceed. And if we do that, you will be a fully AI fluent person. You will be able to contribute in the fullest way to our organization. And no matter where your career takes you, that will be an asset to have. It'll make you just, just more employable across the board. And will you join me in this? And that's an honest question that person has to honestly answer back because not everyone will. I was speaking with a bank and they said, we usually have 90 people just reading checks. that our machines couldn't do. The AI has gotten so good now, we only need five. That means 85 people no longer have a job. And this is the reality. But we keep saying AI is

going to create more jobs. But the reality is the people whose jobs are being impacted by AI, they're not necessarily the ones who are going to take the new jobs. So this bank was able to retrain and reskill about half those people, but the other half, they couldn't find a place, so they weren't willing to take those kinds of jobs. So they left the company. This is the responsible way to be thinking about this, to say, what are we going to do? How are we going to prepare our organization, our workforce, our people for this seismic change that is coming? AI is very good at doing very specific tasks. It's not very good yet at doing actual jobs. We still need people. And my favorite case study in the book is a company called Connecta, a call center company, 130,000 people all over the globe. Charlene Li (22:37.25) And you're thinking call centers. Well, the people there are toast, right? Some of the predictions are 60 % of call center jobs that go away. And they said, you know, our goal here is to not use AI to replace people. Our goal for AI, first of all, is to find ways to enhance our people, to reduce the error rate, to increase the ability for people to answer questions in a coherent way, and to reduce the training time to get people from the entry level into that. that excellence level. The next thing they did is what are the ways that we can enhance customer engagement? How do we use AI to work alongside our agents, our human agents to have better responses and look at all these long lists of things that we haven't been able to do as a company? Now we can because we have more time and efficiency because of the things that we've done with AI. And the best part is they started reinventing themselves. We now have these AI capabilities. What are the new businesses we can get into? They were working with law firms and they said, hey, you got all this information around non-performing loans. We can take all of that data because we have access to it already. We can run AI and analysis and things that used to take you weeks, we can do in hours. Would you like us to do that for you? And they knew that these law firms were never going to get AI in place in any time soon to be able to do this. so this is a company that thought much more strategically about AI, not just... to look at it as a way to replace people, but to enhance people, to say, how do we take these workers who know so much about our business, about our customers, about the way we work, and what are the things we can do with them? So I think that's a much more strategic and wise way to be approaching it. And that's what we need leaders to do today, to look beyond the surface and say, and we talk about this in the book, how do we make our people superhuman? where they can take the best of what AI does. And instead of just using that capacity to just find ways for them to do more work, how do we work differently? How do we enhance the things that are inherently human that AI cannot do? How do we increase their empathy, their intuition, their self-reflection, judgment, and wisdom? And in the call center, for example, Charlene Li (24:58.59) is empathy fatigue is a real thing. How do we preserve that energy so they can take the human aspects and handle the really important cases where AI can't do it, but human empathy is needed? And again, that's how you create super humans. And I believe it's a leadership trait because it's a choice at that point of what do we do with AI? Are we just becoming more efficient or becoming better? Morag Barrett (25:12.063) Mm-hmm. Charlene Li (25:28.31) are re-enhancing our people in maximizing their potential. Because in the end, that's what really differentiates us. That's what allows us to. Morag Barrett (25:37.527) And it's interesting listening to you talk there, insecure leadership is going to keep us stuck in what was before. Insecure leadership, as we know, can result in, I'll hoard information because information and knowledge is power. However, the data lakes that we have mean that with artificial intelligence, we're all diving into that same pool. And so my question is, as you start to think about the ripple effects, creativity, innovation, if we learn to rely on AI, what's at risk? Charlene Li (26:19.054) I think again, if we app again, just let go and advocate AI to do all that innovation and creativity. The problem is AI can only be as creative and innovative as you are. The questions that you ask it are the gate to it being creative and innovative. And so you ask it a decent question. Now they just write me a post about something and I'll write you the post about something. If you give it, want to combine this and that, help me think through this. What are the different ways to be thinking about it? Give me three options. I'm not going to take those and take it to another level. There's a conversation happening. You get a very different result. That's you inserting your humanity, your creativity, your ideas that are unique to only you. The questions that no one else has ever asked because they come from you. And that is the injection, I think, of that extra work that needs to happen because it gets so easy to just get placent and let AI do the work for you. And in some places, if it's just executing on routine admin things, thank goodness, take care of that. Let it do the repetitive data intensive tedious work for you. But reserve your energy and make sure that you are exercising that critical thinking, those points where creativity is needed. that curiosity. one thing we talk about very, very importantly, at the very end of the a sense of imagination, because we only get to the point where AI can do really innovative things. If we can imagine it doing that, I, I, I train my AI to do these things for me on a daily basis. And not because it knew how to do it, but Morag Barrett (27:52.087) Mmm. Charlene Li (28:09.088) I learned from somebody else and seeing how they're

doing it. And I adapted it for me and I'm trying new things and I'm sharing that with other people and people coming back to me. So I'm constantly learning, experimenting, curious, innovating. is not doing that. I'm doing that. AI is doing the execution, but it would never do the things unless I told it to. It doesn't exist just out of the box. And frankly, out of the box, it's fine, but I can make it so much better if I put a little bit of myself into it. Morag Barrett (28:39.565) I love that perspective and framing. And what I'm hearing is it's a call to don't be passive in this relationship with AI, the relationship I have with Claude, be proactive, don't take the first answer, challenge it. And in doing so, it actually accelerates and deepens our learning because it does allow us to connect potentially unrelated themes, different dots in new ways. and to think that through ahead of then sharing it with her colleagues and others. Charlene Li (29:16.206) Right. And again, I think what's amazing to me about AI and what gives me just awe and wonder are the things that you see just be super hard that are now easy. I mean, just brainstorming, finding somebody to brainstorm with, being able to just translate. I recently had it take a script that I had written for a webinar. that I was going to read over teleprompter and it just didn't feel right. And I realized it was because I was writing it with my writing voice. So I said to AI, go and look for interviews of me on the internet and use my speaking voice and rewrite the script. And it felt so much more natural. It felt like it was me. I could never have written that because I write and speak in very different ways. But AI knows the way I speak. I could never have done that on my own. Morag Barrett (29:47.309) Mm-hmm. Charlene Li (30:12.054) And so that to me is just amazing that it has that capability to do that. That it knows me better than I can know myself because this is something I could just never do. And so I deeply appreciate it, but also I know what its limits are. And I think this is, again, knowing what AI can do and what it can't do is such an important thing because it is changing so quickly. are so, so early in this space. So you can err in both ways, thinking that it can just do everything for you. We hear that from so many technology matters. It can do all the work of all your people. It can do a lot of the work for a lot of people. Can it do all of the work? We're testing it. We're seeing it. Some companies are having one person having AI do everything, but that's a very specific type of work that they do that doesn't apply to every single organization. So we are gated by how not Morag Barrett (31:01.77) Mm-hmm. Charlene Li (31:08.086) just how quickly we can adopt AI, but how quickly we can adapt our organizations. And that again, circles us back to this is not a technology problem, it's a people problem. It always is. Morag Barrett (31:18.261) It is a people problem. And so coming back to another theme I just heard you mention there then, it's the ethics and morals and guardrails. Cause the tool itself, I have no influence over the magic that happens behind the scenes that makes Claude what Claude is for me today. But I can make choices about how and when I use it, how and where. I shape and educate it to reflect me and my thought process. What does your research and the work that you're doing with leaders show for that part of the human connection piece? Charlene Li (32:01.88) Right. Again, I want to draw a distinction between responsible AI and ethical AI. Responsible AI just says, how do you use it? Here is the black, here is the white, here's some gray. This is how you figure out the right thing to do. You know, don't put secure private information into AI, just all those usual things. And then also about transparency. And it's all about building trust. That's what responsible AI is. It's about building trust. Ethical AI says, just because you can, doesn't mean you should. Morag Barrett (32:31.767) Mm-hmm. Charlene Li (32:33.358) And this is where values become very important. I know one person who took his values and put them into the instructions and his AI. And he says, anytime you are looking at any problem that I have, look at it through the lens of my values. And I thought that was a really interesting way to inform AI. This is how I want to operate and think about things and imbue it. It doesn't always have to come to the forefront, but just imbue your thinking with my values. around connection, community, and fun, enjoy. And I thought that was an interesting way to build in really hard code values into your AI. But as an organization, I would hope that a discussion around values comes up quite frequently around your use of AI. Because again, should we be using AI in our communications with our customers? When we're communicating with our customers, when... does the AI do something? When does a human do something? When do we do something together? And how do we disclose that? And again, all of these conversations should be taking place. Value should be coming up all the time if you're using AI in the right way. And I really encourage people to have somebody who is looking at AI ethics that someone can go to. And some organizations even have AI ethics offices where people would say, can go to somebody confidentially say, I don't know if we should be doing that. Or a customer can have and raise questions where there's almost like an ombudsman who can, on behalf of employees of anybody, say, are we using AI in an ethical way? So again, there are safeguards we can build in. And so many organizations think, well, these rules are going to slow us down. We have to have, like, these things. I'm like, no, we need to think about the right level to have them. I think of it as Goldilocks governance. You want to have just enough and not too much. Enough to make sure that chaos doesn't ensue, but not so much that you're creating bureaucracy. But when things are clear, when you have these guardrails, you have ways to make decisions, because that's what governance is, then it is clear who has to be a part of that decision and how you make it and criteria for that. Things go faster, not slower. Morag Barrett (34:56.653) So as we come to the end of our time together, what's one message that you hope that the people watching and listening take when it comes to AI and human connection? Charlene Li (35:08.398) I think again, it's about what your role is going to be. And I keep saying that we're in a leadership crisis here when it comes to AI. leaders are needed when there's a lot of change happening and there's a tremendous amount of uncertainty. So what can you do as a leader to create some clarity around things? And it may not always be about you having an answer. because so many things are unanswerable. It could just be you simply asking some great questions. And I think that's the shift that we have to have. Again, it's not about knowing everything and being on top of, this is what I get asked the question, how do I say on top of AI? It's impossible. So instead, shift that desire to have all the answers, which is the natural leadership thing to do, to asking. Morag Barrett (35:52.844) Yeah. Charlene Li (36:01.836) really great questions so that we know where we want to be headed. That is the leadership act that we have today. And especially very, it helps us be very people-centric because it helps the people and ourselves answer the question, what should we be working on? What should we be focused on? How do we bring ourselves and our humanity to answer those questions? Again, the more we can center it around people and not so much around the technology, the better off we will always be. Morag Barrett (36:32.886) So how can people learn more about you, your work, your research? Listen to those interviews that you've got your AI to go and listen to for your speaking voice. How can people learn more? Charlene Li (36:37.386) Charlene Li (36:44.428) Well, I post pretty much everything on my LinkedIn page, so you can follow me there. I also have a charlinglee.com. It has all the ways that you can connect with me and work with me. And then please reach out and contact me. My contact information is there. I love hearing from people. What are your questions? Because that's how I learned also as well. And it's a way for me to continue doing my research. So please get in touch. I would love to hear from you. What's working, what's not, what your questions are. place, I thrive on that communication and connection with people. Morag Barrett (37:19.053) Well, Charlene, thank you for joining me on People First. For everyone listening, make sure you get a copy of Winning with AI, the 90-day blueprint for success. It's a fabulous resource and read, and I look forward to riding the AI wave with you and all of us together. Thank you.