✦数字化转型需要重新构建组织，并且不能外包。
00:04
这很困难，需要最高管理层来推动变革。
哈佛商学院的 Karim Lakhani 博士讨论了技术、创新和商业的交叉点，以及数字化、数字数据科学以及新商业和运营模式的设计的指数效应。
✦技术作为一种赋能工具贯穿于整个组织。
02:58
现代公司及其基本模型的历史视角。
从部门和职能结构过渡到基于生态系统的组织。
✦数字化转型需要运营和组织的根本变革
08:58
云服务的集成可实现可变成本和规模经济
人工智能和算法正在推动以数据为中心的组织模式的变化
✦人工智能优先组织从人类瓶颈转向算法瓶颈
11:40
人工智能优先的组织面临着算法和分析能力的瓶颈，从而实现了巨大的扩展机会。
对企业的影响包括改变商业模式、通过人工智能和数字化旅程增强价值创造和价值获取。
✦数据科学和人工智能对商业领袖至关重要
17:18
了解数据科学和人工智能对于商业领袖来说至关重要，而不仅仅是成为数据科学家。
拥抱更广泛的技术堆栈并将其视为一项持续的任务也至关重要。公司需要持续投资技术而不是将其外包。
✦组织转型是数字化转型战略的重要组成部分。
19:45
仅仅专注于购买数字化转型技术往往会导致失败，因为它忽视了内部组织变革。
与不同人工智能采用成熟度水平的供应商、合作伙伴和内部部门达成共识是一项挑战，需要强大的领导力和支持需求。
✦通过人工智能功能进行个性化内容创建和营销。
24:53
人工智能能够以零边际成本创建按需、个性化的内容。
人工智能驱动的营销可以大规模创建个性化广告和内容，从而改变营销供应链。
✦人工智能驱动的实时销售辅导正在改变销售流程。
27:34
由于大流行，销售转移到 Zoom，从而提高了虚拟销售对话的舒适度。
与 Zoom 集成的人工智能工具为销售人员提供实时指导，彻底改变销售流程。
✦数字化转型战略需要自上而下的协调和项目排序。
32:32
人工智能项目的优先顺序应与公司的战略保持一致，无论是关注成本领先还是客户满意度和新业务模式。
成功的数字化转型通常始于高层的支持，并涉及按顺序扩展项目以满足总体战略。
✦公司需要将流程嵌入软件和技术中以实现数字化转型。
35:09
人工智能和技术将增强人类流程，而不是取代它们。资源分配决策需要反映这一点。
解决人工智能中的偏见至关重要，因为有偏见的数据和算法可能会大规模放大伤害。数据科学知识和法律问题是需要关注的关键领域。
✦人工智能增强而不是摧毁人类能力
40:38
人工智能带来的是能力和容量的提高，而不是适当的规模和成本效率
企业领导者需要投资于自己和员工的学习，以适应人工智能
✦系统地投资于学习和为您的团队构建框架。
42:58
随着组织变得更加数字化，数据安全变得至关重要。
数据标签和管道的安全问题对于算法的准确性也很重要。
It's a total rewiring of your organization. That's what's ahead for most people. Guess what. It's damn difficult. [Laughter] [Laughter] And it's something that the C-suite – this is CXOTalk – cannot outsource to other people. They have to become experts in what's going on and drive the change. That's Karim Lakhani from the Harvard Business School. My work is at the intersection of technology, innovation, and business, how digital technologies are transforming businesses and changing business models and operating models. On the research side of things, I run a lab called The Laboratory for Innovation Science where I'm the founder and co-director. We've done a lot of work on crowdsourcing: crowdsourcing for innovation, crowdsourcing for algorithms. That's what got me into this AI space more than a decade ago. We had partners like NASA, Harvard Medical School, The Broad Institute, and so forth. In light of that, I have (over the last year) launched a new institute at Harvard called The Digital, Data, and Design Institute (D^3) because we think that these three technologies (digitization and digital data science and also the design of new business and operating models) are having an exponential effect. We have launched with more than 30 faculty members at Harvard Business School (and some colleagues at the engineering school) with 12 different labs. We're trying to work closely with companies to solve problems and do great research as well. Your focus is on how companies can compete, so what is going on? What's unique about our present time with AI that caused you to need to look at this problem? The book title is Competing in the Age of AI. We're not even saying competing with AI but in the age of AI. In many ways, the consumer economy (with our mobile phones) has already put the vast majority of humanity in the age of AI. If you think about how you navigate your email, how you navigate your music selection, your viewing habits, your reading habits, your directions, all of that is already immersed through AI. Increasingly now, the tech giants have sort of brought that to us. But that whole world is now shifting into the rest of the economy as well. The book is really about we're not turning back with less data, less digital, less algorithms. We're going to be doing more and more of it. How does that shape what companies do, how companies compete above and beyond you being an AI native firm? The book really sort of starts with the fact that the technology is going to be an enabling tool and it's no longer a thing which sits on its own but is woven throughout the fabric of the organization.
That means that your operating model and your business model are going to change. That's what the book really tries to go after. How is this different from business as we've known it historically? A modern corporation really is maybe about 120 years old. If you think about the history of humanity, most of the time we've been sort of agrarian, small, little shops, and so forth. The modern corporation basically got set up 120, 130 years ago. If there were sort of seminal views of what happened in America, you look at Alfred Sloan setting up General Motors as a multidivisional company, Thomas Edison setting up General Electric as a multidivisional company, as models by which we have always run our organizations. The idea here was that you focused. You went after one thing after the other. You had visions set up. You had functional silos set up, and you were able to go and serve your customer needs. That model has done tremendous things. Our built environment, our built organization has been set up this way. Starting with the advent of computation and computers (with IBM and Microsoft), that edifice started to change where we thought that basically now what matters is not just the ways in which we organized from the top-down but the ways in which information flows across an organization. This information flow view of the world really first started with the tech industry and the software industry and the emergence of, let's say, Windows as a computational tool that allowed lots of people that power to analyze data and do things, and the spreadsheet as the way in which you would get work done. But everything was still very much in the model of divisional structures and functional structures set up. Every time you had to share data, you would be sharing large files, and there would not be things coming together. But what we saw emerge (even in the Microsoft era) was a new type of company was emerging. This company was set up as an ecosystem. Microsoft won the PC battle because they figured out how to build an ecosystem where they had lots of complementors and lots of consumers, and they were in the middle of it. This emergence of this ecosystem in the software industry then basically spread the tech industry, and more and more companies in the tech industry got organized this way. But an interesting thing happened along the way. As these ecosystems got built – and you can think about the mobile ecosystem with iOS and Google, and then, of course, Facebook dominating from that as well, and then, of course, Slack and Salesforce, and so forth, coming on its heels – what people saw was that the ways in which you would run an ecosystem platform-based company was very different than the way in which General Motors ran or General Electric ran. This meant, oh, all of a sudden we need data to cut across our entire enterprise. This meant that we had a better view of customer journeys and could personalize and create better offerings for our clients regardless of if it was a B2B setting or B2C setting or B2B2C setting, for example. The typical silos that we had in our enterprise were no longer the ways for us to organize. I think this is what's new is that what we're seeing is this pressure to de-silofy our traditional ways of organizing and to take advantage of the fact that we now have digital footprints and data across both our company operations and our interactions with customers and our suppliers. How can we put it to use more effectively and more efficiently? What we see really is, in many ways, sort of two models emerge. There is the traditional model in which all organizations, including Harvard Business School, has been in for about 100 years, which basically scales very fast and then reaches a plateau in terms of our ability to serve more and more customers and drive more and more value. You could imagine basically a concave curve of the number of users, your scale, and the value you're creating.
Then we have these digital, AI-first native companies emerging which are growing, in many ways, in exponential rates. It takes a while for them to achieve scale. But once they achieve scale, they can keep growing exponentially. And so, there's a convex curve that shows up instead. These convex organizations, these exponential organizations, at their core are set up with data cutting across the entire enterprise. At the core, drive automation in their processes. At the core, are set up to basically use algorithms to make decisions and make predictions and drive pattern recognition. That shift, we think, is fundamental. We worked our way into it through these tech giants but, increasingly, more and more industries are facing that as well. In essence, what you're saying is the rise of ecosystems and then the rise of data becomes the underlying driver that forces organizations to change in some pretty fundamental ways. One hundred percent. Of course, we'd add in cloud computing and then the advances in algorithms in the last 20 years. Cloud, in many ways, made technology a variable cost instead of a fixed cost. You could then drive massive economies of scale, the cloud company provider, to then be able to take advantage of what you needed. All that have been these trends going in lockstep. But that has meant that the way in which you run a company and the way in which you organize production, operations, are fundamentally different in the ways you might have done things before. Subscribe to our YouTube channel and hit the subscribe button at the top of our website so we can send you our newsletter and you can stay up to date on these amazing live shows. Now overlay algorithms and AI on top of this ecosystem and data-centric model for how we must exist as organizations. Overlay the AI aspect on top. It's no surprise that the leaders in AI actually— Of course, universities were at the source of some of the breakthroughs in neural nets and so forth. But the adopters and the drivers of the changes in AI have come from industry. Why is that? Well, if you think about a company like Alphabet or Google, they face significant challenges in their infrastructure. I remember, in 2005 or '04, talking to people at Google. They said, "Oh, yeah. We built our own server farm with, like, 40,000 servers, and we have 5 people managing it." I was just like, "What?!" [Laughter] At that time. [Laughter] And they said, "Well, we had to make advances in algorithms to be able to make this all self-managed. We didn't want to hire thousands of people running our server farms." Remember, in 2004 or '05, this is novel, right? And so, what happened is that as these ecosystems grew, as they got embedded within the lifestyles of us – we search all the time for information – they were generating a ton of data. This data was laying fallow, and they were like, "Oh, okay. Well, we need to analyze this data, so let's use artificial intelligence and drive the advancement of these algorithms so we can better understand the data." But then they were set up in a very different way. If you think about it, there's no auctioneer at the backend of Google running auctions. It's all machine-driven. The human element is acquire the customer. Acquire the customer, then the algorithms take over. They work with you on your keywords. They work with you on your SEO optimization and the auctions. Then every step in your Google journey is mediated through algorithms. They felt the need to advance the algorithms themselves as a way to drive their own usage and their own growth. Then those spilled over into the rest of the economy. To your question, what's interesting is that the bottleneck in most traditional organizations are humans. Like, "Mike, answer my damn email, please. I sent you the spreadsheet. Can you analyze this for me," or "Can you please FedEx me the hard drive so I can go look at this data?" which is what happens in most organizations or stuck in some Slack conversation and so forth. In many of these AI-first organizations, the bottlenecks are not humans but algorithms and our capacity to actually analyze that. That then opens up scaling opportunities that are quite significant. What should people in business then be doing? Okay, you're running an organization and you're surrounded by this change. What are the implications of this for you? Let's be systematic about our analysis because I really think that the advantage is really technology folks now becoming business folks and also, by the way, HR folks becoming technology folks and technology folks becoming HR folks in thinking about what this change is. The first is, in the book with Marco Iansiti and I, we talk about business models are changing. When we talk about business models, we say you need to be clear about what a business model is. It's both the ways in which you create value, why do customers want to interact with you, and the ways in which you capture value, the ways in which your company makes money. Those need to be separate sets of analyses that you need to do. Now, you can create more value with algorithms and with AI and with digital. You can be more personalized. You can scale better. You can offer your customers much more variety and scope and so forth. Think about how your customer journeys and the value creation journeys that your company does can be enhanced through AI and digital, digital journeys. That's the first bit. You lay that all out. Then separately say, "Now that I'm creating all of this value, how might I capture all of this value as well?" The typical model, I would say, if I create value from you, I charge value, some portion of that value from you. When people come to Harvard Business School, we create a ton of learning value for them. Then we charge them tuition for our value capture. What's happened is that, now with AI, you could automate value capture. You can scale value capture. You can actually even be more creative in value capture, like for example, again, the tech industry has been based on the fact that they create value for us as users and they capture value from advertisers. There are just many more ways to capture value. Thinking systematically about how algorithms, AI, and digital can help you capture value is a separate conversation that opens up. That's just on the business model side. Then we can go, "Okay, now let's bring it to the operating model," which is what actually delivers the value, what happens inside the company. There, we think about three things to scale. How do you serve more and more customers through digital operations and so forth? Here again, what you can imagine is that you want to reduce the marginal cost of acquiring more and more customers through digital. You can impact scale this way. Scope, which you offer them. If you think about your experience now with tech industries, you do more and more things with these tech businesses. How can you improve the scope of things that you do? Then learning, how do you learn better, how do you innovate better as well through machines and the data being infused throughout your organization? We see the transformation tasks for business leaders is to systematically think about applying this technology to your business model and your operating model. Which then completely begs the question how to do it because it's very easy to describe this but the execution and practice is massively difficult because the implications, the tentacles extend through every part of the company. It's a total rewiring of your organization, and that's what's ahead for most people. Guess what. It's damn difficult. [Laughter] [Laughter] And it's something that the C-suite – this is CXOTalk – cannot outsource to other people. They have to become experts in what's going on and drive the change. I would say there are three things. One is the burden on our current leaders of organizations is to learn this new stuff and not be afraid of it. This is a new body of knowledge that you need to acquire not so that you're going to become a data scientist or machine learning engineer or a cloud specialist. The joke I make at HPS is people come to HPS and we have a required curriculum for the MBA program, for sure, and we teach them accounting. If we made accounting an optional course, nobody would take it – or very few people would take it. My dean was the chair of the accounting unit. It was like, no offense to my dean, but accounting, we make it required because we feel like this is important. We feel – we know that in order for you to run a modern business, you need to understand accounting. Our sense is now today, in November of 2022, that data science and algorithms is as essential as accounting for people in business to know. Here's why. We don't want you to become an accountant when you come to HBS, but we want you to be a good business leader. Similarly, when we teach you data science and we teach you algorithms, it's not so that you're going to become a data scientist. We want you to become a good business leader. That becomes the essential bit because if data science and AI is going to be infused throughout your organization, you better understand the ways this works and, in many ways, the downfall of not doing this properly as well. That's the first thing. The second thing is this embrace of the broader technology stack. What I mean by this is that too often technology has been viewed as edifice-building, like we'll go do this technology project like we're building a factory, and then we'll forget about it. I'm sure – in all of your more than 700 programs you have run – you know that this is an ongoing task. Nothing happens in companies without software, without technology, today. We might have it done really poorly but, in fact, that's what we need to do. We know, company after company, the tech companies, Amazon has written the systems 3 times in their last 20 years of existence. Many companies have to keep rewriting their systems over and over again. Leaders need to say that the technology build is an ever-going thing and we can't sort of have that be outsourced and put away. We need to own it and think about it and be thinking about this as an ongoing set of investments we'd be making. The last bit, which I think is the most critical bit. If you think about the first two bits, the data science and the technology stack, I will say that's like 30%. The 70% is the change management you need to do and the change in the organization you need to do. That is the hardest, hardest part. What I tell technology executives that I encounter here at Harvard Business School is that I'm like, "Guess what. You better become an HR specialist as well. You better become a change management leader as well. You can't outsource this to anybody else. This is a change process that you need to embody and lead as much as other business leaders need to as well." That for me is the 70% part of what lies ahead. I think too many people, too many boards, too many CXOs index on the 30% and not the 70%. I am convinced the 70% is necessary but not sufficient. You have to do that stuff, and you have to become good at it. But you as a leader now have to drive the organizational transformation as well with this. Well, of course, it's much easier to focus on buying technology. Let's buy a digital transformation. There's a great vendor. We pay our money, and they just do it, and it's done. But it fails because it never took inside the company. [Laughter] [Laughter] Right. Exactly. So, as I have interviewed so many business leaders, without a doubt the common theme is just as you've said that the hard part about any kind of transformation (whether it's digital transformation or the kind of next evolution of digital transformation that you're describing), it's always the people. But we have a really interesting question from Twitter. This is from Arsalan Khan. He says, "AI needs business process optimization along with integration of data (inside and outside the organization)." Here's his question. It's a great question. He says, "How do you reach consensus with vendors, partners, even internal departments who are not at the same maturity when it comes to AI adoption? How do you make this happen?" An interesting case study that we should think about doing later is the transformation at Disney. If you think about Disney and Disney+ and how they are actually now beating Netflix at their game is an amazing technology and business model transformation story as well. I had a chance to interview Bob Iger. Last year, I led an effort here at HBS to drive our own digital transformation, and I had a chance to interview Bob Iger, the former CEO and chairman of Disney. He said you don't just ask for buy-in. You demand buy-in. [Laughter] This is Bob Iger, the icon of the entertainment industry and so forth. But he said, "Look. Leaders have to demand buy-in. You can't just say, 'Oh, I need your buy-in.' No, no. 'Hey, you're in or out.'" There's a hard answer, for sure, which is like, you've got to drive buy-in. The second thing I would say is, look, I think, in many ways, we as the people driving the transformation have to become good teachers. We have to make sure that people come along with us. And the way to do that is to take on a teaching role, to take on a learning role for them. That's our job. They won't be able to do it themselves. You have to be taking on the responsibility to say, "How do I show you that A) this is approachable and B) that this is doable?" My great colleague Tsedal Neeley, she's at Harvard Business School too and a professor, she has this great thing called the hearts and minds matrix. You've got to change the hearts, but you've got to change the minds. The minds are changed by training, by learning, by making people see that, yes, I am doable. The minds you do through motivation and by showing the relevance that this has. You have to attack both sides simultaneously. Change the hearts and change the minds, and invest in both of them. Again, that's part of the transformation journey that many companies get stuck at because they don't think about the hearts and minds collectively together. Karim, if we think about the kind of changes that AI and algorithms drive across a company, can you maybe give us some examples? For example, you mentioned the business model. You mentioned relationships with customers. There's talent. There's sales. AI changes relationships across all these different processes. There's been a massive explosion in these diffusion models and large language models. Some analysis shows that the rate of improvement is 10x Moore's Law, 10x Moore's Law in these large language models and in these image-generating diffusion models, and so forth. Somebody showed me this Twitter thing which sort of blew my mind, which was like you can now autogenerate videos saying— Mike, let me ask you. Are you a dog person or a cat person? My wife loves cats, and so the right answer is I'm a cat person – and that's for sure. And is there a particular breed of cat that your wife likes that you have? Oh, we love all cats – and I hope you're listening. We love all cats. All right. Great. Now that we know this about you, we can custom create on-the-fly content for you and your wife that always has cats in our promotional videos at zero marginal cost. Now I'll say, okay, we're going to sell Mike some microphones, but we should have little cats floating by because his wife will see it and say, "Oh, definitely those microphones are more fine than the other ones without the cats." Right? That level of personalization is kind of incredible. The fact that I can now generate, on demand, at zero marginal cost, these videos and fine-tune it to you is kind of mind-blowing. But that capability is here today. That capability is here today. What OpenAI is doing, what Google is doing, what Facebook is doing with these kinds of technologies is mind-blowing. Just think that I can now generate personalized ads for each person, tweaking based on their preferences, changes marketing. How would I run a marketing department now when I can create personalized content at scale for each individual? Think about the marketing supply chain from how ideas get generated, how campaigns get created, to how they get launched, to how they get observed and they get monetized. That whole function with these large language models, both in terms of text creation and in terms of content creation, blown away, blown away and rethought through. One example. I spent a bunch of time with Flagship Pioneering to think about how AI and biology are merging together. The same diffusion models that we see for ad creation can also be applied to creating proteins. The same capability for proteins. Now just think how the R&D process changes because now I can generate any protein I want. In fact, one of the companies that we have in our portfolio that I've been advising is Generate Biomedicines, and their view is that they're creating a platform that can generate any protein, proteins that have actually not even existed in the world before, based on these types of technologies. Just think about the R&D function changing. Now I've looked at two very distinctive settings: the R&D function, which we've always thought requires this creativity and geniuses, massively augmented by AI. But then the marketing supply chain being completely turned upside down and fully automated this way. Now companies that will have access to data about you and your wife and can have permission from you and your wife to use that data to do that kind of marketing will be very differently organized. Companies that have an ad agency, creative department, they take six months to create a new ad. That ad is put on TV or even runs on YouTube but is nondifferentiated and so forth – examples. A cool thing I recently saw on this was in sales. Apparently, now lots of sales, because of the pandemic, a lot of sales moved to Zoom and people are now comfortable with having initial sales conversations on Zoom and so forth. Well, there are toolings that you can add onto Zoom that becomes, like an earpiece here, an earpiece for the salesperson to say, "You're talking too much. Slow down." Live, while you're in the conversation. "Pause for more questions. Ask a question this way. Your tonality seems to be more aggressive. Be softer." Realtime coaching for salespeople as to how to respond to a customer, and that's all AI-driven. Imagine how your sales, your face-to-face sales, process is now changing because you have this technology available. It's really augmenting capabilities that we just have not thought through properly before. That I think is the amazing thing that's ahead of us. For CXOs, then the question becomes, well, where do you begin? Do I start in marketing? Do I start in R&D? Do I start in sales? Do I start in operations? Where do I begin? That's why these guys that you bring on your show get paid the big bucks. That's part of the judgment that they need to have to say, "What are the high-value opportunities for me to start to do this? Then as I begin the transformation, how do I bring everybody else along in this way?" Of course, there are innumerable software companies now who are selling products and each one promises that it will be easier than the next. Yes. "And we all have incredible capabilities because of the data," and blah-blah-blah. We've all heard these sales pitches endlessly. Lisbeth Shaw asks a question on Twitter that is directly related to this. She asks, "How can established companies become AI companies while they run their existing business, because you don't want to go out of business while you're transforming your company?" A thousand percent. Lisbeth has it right, which is that's the biggest challenge. We don't have the luxury to be greenfield. We actually have to transform ourselves. What we've seen is there is a joint top-down and bottom-up approach. Declarations by the C-suite to say, "This is the journey we see ahead for us, and this is the way we need to go towards." You need the C-suite, the CXO buy-in, and belief, and a painting of a vision of what that means. Then what I would say is – that's the first thing – in that vision is, how will my customer value get enhanced; how will my clients be better off if I imagine this world to be? This is part of the top-down strategy around this. Then it's a question of saying, "Okay. Which are the problems that we should go after?" What I would say is it's easy for you to say, "I've got to rebuild everything," and it's like you're never going to rebuild everything. You don't want to be in this world of, like, I'm going to pause for five years and rebuild everything. You want to say, "All right. There are two things I need to do. I need to deliver value but also build capability so that I can do this more and more often and do it along the way." You then look around, either on your business model side or on your operating model side. Again, on value creation and value capture or on scale, scope, and learning. Say, "Where are some high-value problems that if I solve and I demonstrate that these get solved that I can then take that and then scale it across my enterprise?" I start with a prototype. I start with a POC. But the POC doesn't sit by itself. The POC is designed to scale. You say to the folks the green light to the POC that if this works, what is our plan to scale, and you have the plan to scale agreed upon before even the POC starts. What we've seen over and over again is that the POCs actually work. I've seen amazing hit rates for POCs working. But then they all are dead zombie projects in many organizations because there's been no commitment to scale. The commitment to scale then means, "Oh, I've got to change my operating mode, the ways in which I do that," but you need buy-in. It's the bottom-up identification of use cases, bottom-up identification of POCs, top-down agreement that we're going to do this and that, as these POCs start to scale, you prioritize. We will then make them go across the enterprise. The thing I learned from some colleagues – you know I was just spending a bunch of time at Boston Consulting Group (before I became academic), and I've reacquainted with them since I wrote the book – they had some very interesting perspective that oftentimes people get into this prioritization game, like, "Oh, which projects am I going to prioritize?" The reality is, in a top-down transformation, you'll need to do everything. And so, the question is one of sequencing. The sequencing of the projects and the scaling actually has to be based very cleverly on your strategy. Is the strategy to blow away your competition and be the low-cost provider? Then the projects you would do for AI are very different than saying, "I'm going to be number one in customer satisfaction and new business model creation." That's a very different set of perspectives. The use cases get identified at the bottoms-up level. You need top-level agreement that this is the journey they want to go on. But then top-level agreement to say that as these POCs get developed, we're going to sequence them and scale them to meet our strategy. That's the way that these transformations will work. If you think about Disney as an example, where did they start? They first started by buying Pixar. Then they boat anchored Disney Animation Studios to create digital animation. That was 15 years ago, 17 years ago. Then in that journey, they've gone step-by-step to build their own digital capabilities and start to build a platform where then Disney+ launches just before the pandemic and can take advantage of people's home viewing, but then keep going that way by being able to actually beat out Netflix at their own game. It's interesting. Just as you were describing Disney on Twitter, Michelle Batt came in to point out that Disney's success is also related to leadership pushing from the top down. This is Bob Iger saying, you know, demanding buy-in. Lag [Laughter] It reminds me of the great leader of our time Elon Musk going to Twitter and saying, "You will now work 24/7. And, by the way, we're firing half of you today." You may disagree with his personality and his politics and his incessant use of Twitter, but his ability to change the space industry, the auto industry, the electric industry, you know, electrification with SolarCity, you can't— He's done things that we'd be lucky to do in one lifetime. He's done three of them already, and we'll see what he does with Twitter. I'm not a big fan of his management style but, guess what. One of the most important questions he asked at Twitter was, "How many people are writing code that ships versus managing?" It's like, "Oh." I think the ratio was 5:1. He goes, "Okay, that has to change," because, in the end, most of our companies are going to be embedding our processes in software and technology. That's the key thing that I think CXOs have to get their head around that everything we do is going to be embedded through software, through technology, through AI. That's where you have to then make resource allocation decisions and so forth. The technology and the AI is going to augment our humans. It's not going to replace them. It's going to augment them, but the processes you have would have to be very different. We have another really important point from Arsalan Khan. He comes back, and he asks about the bias question. He said, "With data and algorithms—" I'm paraphrasing his question but, essentially, he wants to know. He phrased it really well. He says, "How do we reduce bias in AI when the ultimate goal is increasing profit and not necessarily AI's impact? For example, changes on the workforce or in society." How do we balance these? It's a really important issue. Let's unpack this. One is, why is there a bias problem with AI? Well, because bias can exist because our data that we are using to train the algorithms is not representative. We just have one class of citizens generating the data instead of another class. Our labeling operations may not be representative as well. For example, lots of tech companies have problems identifying blacks in their image processing systems because the labelers weren't able to identify them properly or distinguish the features that way as well. One is a story of data and data operations. This is why data science is a critical skill for all executives because you have to understand the data generation processes and all of the faults that would happen. I think that's the first thing. To take it to the limit, just as I can scale the benefits of AI exponentially, I can also scale the harms of AI exponentially. Bias is one of those things. The second thing is that there is a real legal issue, which has been the thing that has been so interesting for me. Statistically, computer scientists and statisticians, when they look at the algorithms, say, "Is this algorithm fair?" Oftentimes, when we think about fairness in statistics and in computer science, we think about on average. Is this algorithm treating people fairly, on average. But the law doesn't say average. The law says each and every individual has to be treated fairly. There's a lot of risk that companies are facing today because their algorithms are, on average, fair but to the individual they're not fair. They're open to a lot of liability questions. How do we make sure that those things are addressed upfront instead of addressed after the fact? This is where I think is the new frontier for many organizations, which is the conversation about bias and fairness and transparency in expanding the algorithms should not be a computer science or an AI task. This is a cross-functional task that resides with business, with technology, and with legal and policy. This has to be done collectively. Importantly, we can't do this ex-post, after the algorithms have launched. We have to do them pre- in the design phase. I think Satya Nadella has done the most thinking about this because, remember, Microsoft had a crazy amount of cybersecurity issues in the 2000s. What they had to do was retrain their software developers to build quality software and security into their processes instead of doing it ex-post. I think the same thing is going to happen with algorithms and bias and AI is that we have to build in the awareness about bias in our processes upfront instead of ex-post when the algorithms are released into the wild. The example I use is that when you go to Toyota (as a manufacturing company), there's no quality department at Toyota. Why? Because they feel like if your processes aren't creating quality then a quality department is never going to fix it. They make quality the responsibility of all employees, and they've built processes to ensure quality is built into the systems instead of doing it at the end. I think the same thing is going to happen around AI and bias as well. This is from LinkedIn. Cezar Babes comes back, and he responds to you the following way. I'll ask you to just keep your answer pretty brief. He says the following: "Should AI use be more tightly regulated? Every now and then there's a new technology that becomes the catalyst for—" and I'm readying his quote "—profit-driven goose chases. This results in loss of jobs and resentment towards that technological advancement." He says, "It would be great if AI would be a driver for human growth and result in increased capability and capacity rather than right-sizing and cost-efficiencies." Is there going to be displacement because of AI? Absolutely. Do we need to retrain people? 100%. But my belief is that, in the end, AI augments human capability instead of destroying human capability. Just as prior technologies have been enhancing us, the same thing is going to happen with AI as well. Is there a displacement period and are certain occupations going to be displaced? 1000%. That's where governments and so forth have to come together. But regulation, who is going to regulate AI and in what way? It just doesn't seem tenable to us because it is so widespread. What advice do you have for business leaders who are listening to this and saying, "All of this is fine, Professor Lakhani, but my business is successful. We don't have to deal with this stuff. We're pretty much happy as clams, so this doesn't affect us"? Go talk to your customers, and not about your products but other things that they're doing. I tell you; you will be shocked with how they're thinking about the world and how much technology is driving their decisions. When you ask them about your own products or your competitors' products, you will never hear the right answer. Ask them about other things that they're doing in their businesses, and you'll be shocked. What advice do you have for business leaders who are listening to this, nodding their heads, and saying, "I know this is true, everything you're saying. We feel the pain. I feel the pain, and I don't know what to do. It's too big and complex and hard." There is a learning mandate for this for all organizations, which is, we have a generation of leaders that came in the old model. They don't understand the technology, don't understand data science, don't understand statistics, don't understand algorithms, don't understand cloud, and feel like that's for the IT guys. I think there's a learning mandate for these leaders not just for them to become better at this but then to also get their whole organization to change as well. And so, I would start with learning. Invest in the learning for yourselves and your folks. There's so much stuff available. What you have done through this amazing series, what we offer for paid, there's lots of stuff. There's no excuse for not learning. There are lots of books. Invest systematically in learning, yourself, and building a framework for your whole team. Then cascading that down so that everybody has the same reference point. That's the first step. I see too many people shirk on learning and say, "This doesn't apply to me," when they don't know what's going to hit them over the head with this stuff. Jose Kurian just wants to point out that security is the most important of AI and machine learning services, and so can you just say something about the security dimension of all of this? Security, overall, as we get into more digitally intensive organizations, which all of us are becoming, data security, information security is going to be key-key-key for all of us. Secondly, there are actually a bunch of very important issues about data security and our data pipelines being secure and not being tampered with. Just think about labeling operations that many companies have. Many times, those labeling operations are outsourced. They could be subject to attack where even just the slight bit of mislabeling could give you flawed algorithms. As we start thinking about this stuff becoming infused throughout our enterprises, the security side around data itself is going to be massively important. I 100% agree, 1000% agree. With that, we are out of time and over time. I want to say thank you so much to Professor Karim Lakhani from the Harvard Business School. Thank you so much. I'm so grateful for your taking the time to be here with us today. It was so much fun, Mike, and it was great for me to be on this side instead of just listening. Thank you for the invitation again. Well, I hope you'll come back again. Absolutely. A huge thank you to everybody in the audience who watched and especially to the folks who ask such amazing questions. You guys are such a great audience. I have undying respect for you. Everybody, thank you so much. Check out CXOTalk.com. We have amazing shows coming up. Before you go, subscribe to our YouTube channel and hit the subscribe button at the top of our website so we can send you our newsletter and you can stay up to date on these amazing live shows. Thanks so much, everybody. I hope you have a great day. See you soon.
本文由 xitibu 创作，采用 知识共享署名4.0 国际许可协议进行许可。
本站文章除注明转载/出处外，均为本站原创或翻译，转载前请务必署名。