Latest / CXOTalk / Why Your AI Strategy Is Failing: Fortune 500 CEO Advisor Explains | CXOTalk #900
Transcript
- 0:00Most executives see AI as just automation.
- 0:04That's wrong and explains why their AI strategies fail.
- 0:09Today on CXO talk #900, Sangeet, Paul Chowdhury, advisor to over
- 0:1640 Fortune 500 CEOs and author of the book Reshuffle, explains
- 0:22why AI is actually coordination infrastructure and what that
- 0:28means for your organization. I'm Michael Kriggsmann.
- 0:31Let's get into it. There's a lot of hype about AI
- 0:34at the moment and there's, there's some fairly, you know,
- 0:40consistent patterns that we see in how strategy and systems
- 0:44change when new technologies come in.
- 0:46And when we're in the midst of hype, we sort of miss that
- 0:49whole, you know, understanding exactly how some of these things
- 0:54will play out. So the reason I wrote Reshuffle
- 0:56was to kind of extract us away from how we think about AI in
- 0:59terms of the hype today, but really to think about how
- 1:02endealing change happens because of new technologies.
- 1:04And that's really what inspired me to kind of bring this book
- 1:09together. Your book describes a different
- 1:14paradigm or a different way of looking at AI, so tell us about
- 1:20that. I work with a lot of executives,
- 1:22and when they say that they're implementing AI or they're
- 1:28thinking about AI, they typically mean that they're
- 1:30automating tasks. They're automating workflows in
- 1:33their organization, they're deploying technology towards
- 1:36automation. And my key argument in the book
- 1:40is that the real value of AI plays out not through automating
- 1:44tasks in the system that you're playing in today, not through
- 1:49just automating what already exists, making it cheaper,
- 1:51better, faster, but really reimagining your business around
- 1:54the capabilities of AI. Because the real advantage or
- 1:58the real impact of any technology plays out when
- 2:00economic activity gets reorganized based on the new
- 2:04capabilities of the technology that is made available.
- 2:07So even when we say, you know, when executives today say we
- 2:11need an AI strategy, when I ask them about it, very often what
- 2:14they mean is we want to figure out what should we do with AI?
- 2:18And that sort of traps them within thinking about, here's
- 2:22our business. How do we apply AI into it to
- 2:24get things cheaper, better, faster.
- 2:26But that's not really what strategy is about.
- 2:28Strategy is fundamentally about answering 2 questions.
- 2:31Where do we play? How do we win?
- 2:33And so when we put AI into the mix, the question that we should
- 2:36be on, you know, the questions we should be thinking about is
- 2:40how does a world where AI comes in?
- 2:44How does that change our playing field?
- 2:45How does that change, you know, who we compete with, how we
- 2:48compete, how we create value, how we capture value?
- 2:52And then how, how does it change advantage within that playing
- 2:54field? How does it change how we win?
- 2:56So it's really about reimagining advantage and your business
- 3:02given that the world is changing, rather than thinking
- 3:05about how to apply AI within the confines and assumptions of your
- 3:09current business. OK.
- 3:10So take that down another level for us because you're describing
- 3:16the limitations of viewing AI as a technology that can make
- 3:21things essentially better, faster and cheaper.
- 3:24And I think that's how most of us view technology in general.
- 3:30So so you need to unpack this. In my book Reshuffle, I make a
- 3:34distinction between thinking about the impact of technology
- 3:37at the level of tasks and activities, which is, you know,
- 3:41you do a certain set of activities and run a certain set
- 3:44of workflows to create value as a business.
- 3:46You're bringing AI in to speed those things up.
- 3:48So that's the automation view of thinking about how you think
- 3:52about AI or technology in general.
- 3:54But if we look at, you know, previous technology waves, and I
- 3:57take the example of the shipping container, I use that example
- 4:02because today we are very obsessed with whether AI is good
- 4:07enough or not, whether AI is intelligent enough or not.
- 4:09My point is that even with it's current capabilities, AI is
- 4:13massively underutilized because we are still thinking about
- 4:17speeding up things in the existing system and that
- 4:20prevents us from seeing what's possible around the capabilities
- 4:23of AI. So if I very quickly just take
- 4:25us through, you know what happened with the shipping
- 4:28container because it's, it's an interesting parable to
- 4:31understand what's possible with AI today.
- 4:33When the shipping container was introduced, it's fullest order
- 4:37effects were seen as automation because prior to that we used to
- 4:42live in a world of brake bulk cargo.
- 4:43So you needed people to, you need needed dock workers to go
- 4:47up and down a ship, moving cargo off and on a ship.
- 4:51And that lack of standardization made port operations slow.
- 4:56So when the shipping container was invented, people assumed
- 5:00that the port would get automated because cranes could
- 5:02move, you know, cargo on and off a ship.
- 5:05So the first order effects were certainly that of automating the
- 5:09task and making it better, faster, cheaper.
- 5:12And the first order impact of that was in the loss of the jobs
- 5:15of the dock workers. And that's where we, you know,
- 5:18if you think about AI today, we're always thinking about that
- 5:20what can AI do that a human is currently doing?
- 5:23And hence those jobs are going to get lost.
- 5:25And what are we doing right now, which we could do better,
- 5:28faster, cheaper. But the parable of the, you
- 5:31know, the story of the shipping container really plays out
- 5:33through the 2nd and 3rd order effects.
- 5:34Because what happened next was that trucks, trains and ships
- 5:39agreed on a common standard for the shipping container.
- 5:42And that totally unlocked logistics at a global scale,
- 5:46because now you could move shipment from source to
- 5:50destination completely seamlessly.
- 5:54And that made logistics reliable.
- 5:56And because shipping now became reliable from source to
- 5:59destination, the logic of manufacturing change because the
- 6:02previous assumption of manufacturing and I'll come back
- 6:05to this, you know point about assumption, because we had to
- 6:08rethink our assumptions with AI. The previous assumption of
- 6:11manufacturing was that shipping and and movement of goods is
- 6:15unreliable. So you want to keep either
- 6:17manufacturing internally vertically integrated or you
- 6:19want to keep your suppliers Co located.
- 6:22But once shipping became reliable, suppliers could be
- 6:25anywhere in the world. And so global supply chains came
- 6:27up and jobs created a lost were not because of what happened at
- 6:31the port, but what happened because of these global supply
- 6:34chains. Manufacturing moved from
- 6:36vertical vertically integrated models to component based
- 6:40manufacturing. So new jobs were created because
- 6:42of that component based manufacturing and competition
- 6:44that emerged and eventually even countries rose and fell based on
- 6:49how they plugged into this new system of global supply chains.
- 6:52So my point is, a lot of our knowledge work today is
- 6:56structured around certain assumptions of scarcity of
- 6:58performing knowledge work. And just like there was an
- 7:02unreliability of shipping or the cost of shipping was very high,
- 7:06today, the cost of performing knowledge work traditionally has
- 7:09been very high. But with AI, certain forms of
- 7:12knowledge work, not all, but certain forms of knowledge work,
- 7:15the cost of performing them dramatically collapses.
- 7:18For example, the cost and speed of translating a document was
- 7:22very high just a few years back, but today it's completely
- 7:25collapsed. So when that happens, the
- 7:28assumptions on which your business is structured
- 7:30fundamentally change. And with that, you have to
- 7:32reimagine your business. You have to reimagine what kind
- 7:34of competition will come in. You have to reimagine what's the
- 7:37basis of advantage. And that's my key point, that
- 7:39unless we think in those terms, we're missing out on the real
- 7:42potential of AI. We have a couple of interesting
- 7:46questions that have started to come in relating to these issues
- 7:51that you've just been describing.
- 7:52So let's jump to those and I just want to tell everybody that
- 7:58you can and you should ask questions.
- 8:02If you're watching on LinkedIn, pop your question into the
- 8:05LinkedIn comments. If you're watching anywhere
- 8:08else, go to X Twitter and use the hashtag CXO talk.
- 8:14I urge you take advantage of this.
- 8:16When else can you ask Sangeet, Paul Chowdhury, pretty much
- 8:20whatever you want. So take advantage of it.
- 8:24So to begin, on LinkedIn, Huawei, Yi Lee says that when it
- 8:31comes to doing what you're describing with the least impact
- 8:35on structural unemployment, she believes that the right thing to
- 8:42do is not teach employees how to use Gen.
- 8:46AI for their existing roles, but to imagine how human agent teams
- 8:55will be structured and where employees will fit into this new
- 9:00framework and make these teams part of the test and learning
- 9:05process. How far off is she?
- 9:08She's asking. Learning how to use AI is
- 9:13important, but it's stable stakes.
- 9:15It's not going to give you an advantage.
- 9:17It's just going to help you run. It's not going to tell you where
- 9:20to run. And so you're not going to get
- 9:21anywhere just by learning how to use AI.
- 9:23So this whole, you know, a dig of AI won't take your job, but
- 9:28someone using AI will is grossly incorrect, correct?
- 9:32Because it's sort of forces the listener to assume that adoption
- 9:36of AI, learning how to use Gen. AI is sufficient.
- 9:39The, the key, I would say there are two or three things that are
- 9:42really important. The first thing that's important
- 9:44is that whether you're an individual thinking about your
- 9:47job or career, whether you're a team leader, thinking about your
- 9:50team, whether you're CEO, thinking about your company, you
- 9:53have to think through what is the state of the playing field
- 9:57going to be out there a year from now, six months from now,
- 10:01whatever time frame you feel comfortable with it, it can't be
- 10:04five years because the rate of change of things is too, too
- 10:07fast. So you have to, let's say, take
- 10:09a year from now, what's the state of the playing field going
- 10:12to be and what in that's, you know, given those assumptions,
- 10:16what's going to give you advantage?
- 10:18So you have to use some level of foresight, even if an individual
- 10:22and planning your career, that's the only way you can have agency
- 10:25in a time of rapid change. So the first thing is apply some
- 10:29foresight. Think about what you know the
- 10:31future map is going to look like and where you're going to play
- 10:34over there, how you're going to win, make some bets and then
- 10:38start executing, start moving in that direction.
- 10:40And as you move in that direction, you'll learn what's
- 10:42working, what's not working, which of those bets were made
- 10:45sense or did not. And maybe the models will
- 10:47improve and some of those bets no longer hold.
- 10:49So keep updating your bets and your map on the basis of that.
- 10:52But if you don't have a vision, if you don't have some stake in
- 10:57the ground, that this is what, what my game is going to look
- 11:00like, whether I'm an individual or the team member.
- 11:02And, and this is what, how my team will fit in the
- 11:04organization at that point. Unless you have that stake in
- 11:07the ground, you don't, you know, you, you don't have any thesis
- 11:11that you're testing and updating and validating as you move
- 11:14forward. So that's the first thing I
- 11:15would say. The second thing that's
- 11:16important is that we very often think about this duality of
- 11:22automation and augmentation, and we feel that automation is a bad
- 11:25thing for humans. Augmentation is good because
- 11:27it's helping you get better, but there's a there's a fallacy over
- 11:31there because augmentation is not inherently good.
- 11:34Automation means the human effort gets substituted on on a
- 11:38task. Augmentation means that it gets
- 11:40complemented. Now, complementarity does not
- 11:44guarantee that you will continue to retain your salary and your
- 11:48agency associated with that work.
- 11:51It just means that if AI complements humans, we will
- 11:56emerge in a system where the new division of labour is made
- 12:00between AI and humans, so that AI gets what is best done by AI
- 12:04and humans continue to do what they can best do.
- 12:07That's that's the only thing that it means.
- 12:09It just means that we're going to have new workflows, new
- 12:11division of labour where the two work together.
- 12:13It does not say anything about whether you will be able to
- 12:17retain your ability to capture value and command a premium for
- 12:21your skill with that work. In many cases you may not, and
- 12:24it absolutely doesn't say anything about whether you will
- 12:27have the agency that you enjoy today.
- 12:29So the second thing that I would say is that don't assume that
- 12:34just because in your job you're starting to see higher
- 12:37productivity because you're using AI, that is always going
- 12:40to be a good thing 369 months from now.
- 12:43You have to, you know, always be aware of the fact that there's a
- 12:46new division of labour that's going to come.
- 12:48And that's the whole idea for a reshuffle in this in this
- 12:51instance. And so the final thing that I'll
- 12:54just point out over here is that somewhere in the question there
- 12:58was this point about, you know, should we be looking for
- 13:01guidance and direction from how the organization is thinking
- 13:04about implementing this new division of labour etcetera.
- 13:08I think that it's there's never been a more important time than
- 13:11today to have agency or our career choices or how we want to
- 13:17move forward. And I think we're at the start
- 13:20of a complete reshuffle and how we think about our careers.
- 13:24We're at a point where what we are learning through the
- 13:28traditional models of learning is getting decoupled from what's
- 13:30being rewarded. And that's only going to, you
- 13:34know, I won't say accelerate, but it's going to get
- 13:36increasingly decoupled and increasingly thrown out of
- 13:38disarray. So if you don't have agency with
- 13:41the first two things that I mentioned, relying on your
- 13:45organization or some other structure to tell you where you
- 13:50fit in is not going to be a winner solution.
- 13:53How do you recommend that business and technology leaders
- 13:58develop that broader perspective that you're describing and that
- 14:03vision when they don't know what's coming down the Pike?
- 14:08And we have a question relating to this from Twitter.
- 14:13And this is from Anthony Scrifignano, who's been a guest
- 14:16on. He's a great Day to Scientists.
- 14:18He's been a guest on CXO Talk a number of times and he says
- 14:22this. What are the blind spots that
- 14:25you see in the more mature organizations that lead as
- 14:30you're suggesting? The two questions are related,
- 14:34but they're slightly distinct. So I'll just talk about the
- 14:37first thing you know, how do you think about these issues?
- 14:39So there's there are certain clear ways and clear structures
- 14:44on how you think about it, which I explained as the framework for
- 14:47how you see the reshuffle. So the first thing is that
- 14:51technology always moves value. Value does not stay exactly
- 14:56where it was today. So the first thing that you want
- 14:58to think about is what are the assumptions on which my work and
- 15:02my business is based and what are those assumptions of
- 15:05associated with value. So, so typically when I say
- 15:08assumptions associated with value, it's about on what is the
- 15:12assumption about how I differentiate myself versus
- 15:14competitors and what's the assumption about what helps me
- 15:18capture that premium against my competitors.
- 15:20So a lot of knowledge work, as I mentioned, was structured on the
- 15:23assumption that access to talent is expensive and hence access to
- 15:28performance of knowledge work is expensive.
- 15:29But if that performance of knowledge work becomes cheap,
- 15:32then suddenly that assumption changes.
- 15:35A second set of assumptions related to this is that, you
- 15:40know, traditionally knowledge was very siloed in the sense
- 15:44that if you had really good knowledge about chemistry, you
- 15:47would have that trapped inside certain industries.
- 15:50If you had really good knowledge about energy cells, you'd have
- 15:53trapped that inside certain industries.
- 15:56But with AI coming in, it's possible to now package that
- 16:01knowledge and make it available beyond industry boundaries.
- 16:04And that has implications both for firms and individuals
- 16:07because as individuals, we were siloed to the training and
- 16:11career path that we had. And today we could potentially
- 16:13access knowledge that we would never have had access to.
- 16:17If we had to access it, we would have had to hire a team.
- 16:19And that was prohibitively expensive for others
- 16:21individuals. But now you can have, you know,
- 16:24a team of agents working on your behalf, getting that knowledge
- 16:27for you so that you can then figure how to apply it to your
- 16:29work. So the the fact or the
- 16:32assumption that knowledge is no long, that knowledge is tied to
- 16:35a silo starts going away as well.
- 16:38So when these assumptions change, my key point is, and
- 16:41it's not just these two assumptions, they're
- 16:43illustrative, but when these assumptions change alongside
- 16:46that, what happens is that value used to be tied to these
- 16:50assumptions. So when the assumptions change,
- 16:52value moves somewhere else. When value moves somewhere else,
- 16:54you have to think about, you know, what's going to be newly
- 16:57valuable. And hence in order to capture
- 16:59that, what should I be doing? Should I be changing what I do
- 17:03completely and moving to a different part of the value
- 17:05chain in order to go there? You know, how do I differentiate
- 17:10myself if I've never been, if I've never played in that part
- 17:13of the value chain before? So these are the things that,
- 17:16you know, you need to think about.
- 17:17I'll just give a very simple illustration from a previous
- 17:22technological shift. So when digital cameras came,
- 17:25the cost of taking photos went down to 0.
- 17:28And when the cost of taking photos went down to 0 and you
- 17:30were no longer restricted by the capacity of the film, the value
- 17:34of storage initially and and storing and tagging your photos
- 17:37became more valuable. So you know, Flickr had a good
- 17:40valuation, then sharing of photos became more valuable.
- 17:43So Instagram got a good valuation.
- 17:45But my point is that it was not that Kodak did not become
- 17:48digital. Kodak was the number one digital
- 17:50camera manufacturer and seller in the US in 2006 and then went
- 17:54bankrupt in 2011. But that was because value was
- 17:57no longer in digital cameras or in selling cameras, and neither
- 18:00was it in managing and printing photos.
- 18:03It was in sharing photos. And so Kodak saw that value is
- 18:08moving, but it couldn't do anything about it.
- 18:10On the other hand, Fujifilm saw that value is moving and it
- 18:13realized that it's business had always been about chemistry,
- 18:17about chemicals and photos just happened to be a way to make
- 18:21money with those chemicals. So then it took the same
- 18:23capabilities and moved to other playing fields.
- 18:25It took the same capabilities, moved into pharmaceuticals,
- 18:28moved into cosmetics, and essentially used its
- 18:32capabilities in chemistry to start making money over there
- 18:35and survived that whole shift very well.
- 18:39So that's my key point that you need to think about the new
- 18:41playing field. You need to think about whether
- 18:43you'll go for new capabilities, leverage your capabilities to
- 18:46another playing field. You need to think about these
- 18:49aspects, even whether as an individual or even as somebody
- 18:52managing your own career. So this is from Arsalan Khan.
- 18:56He says AI requires imagination and reimagination of the entire
- 19:01organization. And here's the crux.
- 19:04If a CXO reached the C-Suite without AI, why should they
- 19:10care? Are there organizational
- 19:12existence issues but CX OS don't last in their position that
- 19:17long? It's a somewhat loaded question
- 19:19in the sense that there are certain the, the term CXO is
- 19:23very broad. So in certain functions, maybe
- 19:26if you're ACFO you can afford to not get affected by it a little
- 19:30longer than if you are say as you know, the CEO for instance,
- 19:35who who's responsible for driving competitiveness and
- 19:39growth and value for the organization.
- 19:41But that aside, even as a CXO, if you want to continue staying,
- 19:45staying relevant, it's not whether your organization is
- 19:51taking a certain posture to AI or not.
- 19:54The entire playing field is shifting.
- 19:56So within four to five years when AI or any technical
- 20:00technological shift, but more so with AI because the rate of
- 20:03change and the impact is more widespread than a lot of
- 20:08previous technological shifts. When the playing field
- 20:11constantly changes, the way companies compete change.
- 20:14When the way companies compete changes, the capabilities they
- 20:17value change. And so the point is not that you
- 20:21should be learning how to use AI.
- 20:22Maybe you should, maybe you shouldn't.
- 20:24That should be subservient to the question of which
- 20:28capabilities are going to be valued going forward.
- 20:30And will I continue to have those capabilities, the
- 20:33capabilities I have today, will they still be valued when the
- 20:36way companies compete in my industry will fundamentally
- 20:39change? So whether you're a CXO or not,
- 20:43the value of experience as crystallized knowledge will
- 20:47always stay. But the value of experience as
- 20:50the ability to just understand a lot of what you've talked, you
- 20:54know, just understand domain knowledge is going to collapse
- 20:57dramatically. Instead, you'll have need to
- 21:00have the ability to complement your crystallized knowledge with
- 21:03constantly learning new domain knowledge based on how the
- 21:06playing field is changing. So if you're ACXO and you're
- 21:09thinking of retirement, maybe that's different.
- 21:11But if you're a CX and you want to continue working somewhere
- 21:13else in order to be relevant, your capabilities should still
- 21:17be valued. And so, so it's less about
- 21:20learning AI and becoming a illiterate.
- 21:22It's more about build a very clear lens of what's going to be
- 21:26valuable. What, how is the, you know,
- 21:31changing playing field going to change my job, my job or my role
- 21:35or, you know, the capabilities that I have?
- 21:37Will they increase in value or decrease in value?
- 21:39Those are things that you need to think about.
- 21:42So if I can paraphrase what you just said, if a company or an
- 21:49individual does not expand their skills and broaden their
- 21:55perspective on how AI will affect the roles, company,
- 21:58markets and so forth, then they're basically screwed.
- 22:01I would say that that's pretty much on point.
- 22:03It's it's less about the skills, it's more about the perspective.
- 22:06If you have the perspective, you'll know which skills to move
- 22:08after, but you need to have that perspective.
- 22:11And this is from Maya Cunningham on LinkedIn, and she has a short
- 22:17question. So my hat's off to Maya
- 22:19Cunningham. By the way, folks, now would be
- 22:23an excellent time. To subscribe to the CXO Talk
- 22:26newsletter. Go to cxotalk.com right now,
- 22:30subscribe because we have incredible shows like this and
- 22:34we'll notify you of what's coming up.
- 22:37And on Monday you can go to that, to the site and see the
- 22:43edited version of this conversation.
- 22:45So it's being recorded. So subscribe to our newsletter.
- 22:48Do that now, please. All right, Maya Cunningham says,
- 22:52what is 1 early sign that an organization is truly shifting
- 22:58from task automation to institutional RE architecture
- 23:04with AI. And I would also add to that,
- 23:08what are signs that an organization is not making that
- 23:12shift? There are some leading
- 23:14indicators in terms of the types of directives that are being
- 23:18passed along internally. So let's take a simple example.
- 23:21You know, AI can be used to accomplish work faster, better,
- 23:24cheaper, yes. But AI can be used to redesign
- 23:27work as well, reorganize work as well.
- 23:29So even before we talk about institutional redesign, think
- 23:32about the average team inside an organization.
- 23:36Are they thinking about how to use AI to speed up tasks?
- 23:39Are they mapping out their tasks on a white board and saying,
- 23:42here's how our tasks flow and here's where we can use AI and
- 23:46move things faster. And instead of 10 days, we'll do
- 23:48it in three days? Or are they looking at their
- 23:51flow of tasks and asking themselves why does that flow
- 23:55exist? Because a workflow looks like a
- 23:58sequence of tasks, but a workflow is structured around a
- 24:02constraint in the system. A workflow is set up to solve a
- 24:05problem for the larger organization.
- 24:07So if you get distracted by looking at the tasks, instead of
- 24:10asking what was the problem in the organization this workflow
- 24:13was set up to to resolve, and does that problem still exist
- 24:17now that AI comes in, that's the way the team should be thinking
- 24:21about redesigning their work. They should be reimagining the
- 24:23workflow by asking first whether that constraint still exists.
- 24:27So that's a leading indicator in terms of if you're looking at a
- 24:29team and organization, how they're thinking about
- 24:31redesigning their work is, you know, one of the the leading
- 24:36indicators of whether the organization is taking a task
- 24:38first approach or an institutional redesign approach.
- 24:41I think the second piece is in terms of how the organization
- 24:46communicates internally at every level, how to think about AI.
- 24:51If the narratives are we need X person AI adoption by every team
- 24:57that does this and Y person by every team that does this, then
- 25:01you know, they're, they're, they're probably driving a
- 25:04certain mandate which will help with AI adoption, but they've
- 25:07not really thought through why. And if instead the CEO has
- 25:12taken, has made it a point to reimagine the future posture of
- 25:16the business and given that has asked his directives to think
- 25:20about what that means for their business units.
- 25:22And they have in turn, asked their teams to think about what
- 25:24that means for their team. You know, it has to be
- 25:28bidirectional every and it has to, you know, you need to have a
- 25:31constantly learning organization.
- 25:33It cannot be top down. It cannot be localized.
- 25:36If everybody's constantly thinking about given everything
- 25:39that's happening in other industry, what does that mean
- 25:41for my team a year from now, will my team still have the
- 25:43power it doesn't the organizational structure?
- 25:46Or does it need to rethink what it does?
- 25:48If they're thinking in that way, if they're thinking with, with
- 25:51foresight and future 1st and working back from it, then you
- 25:55know they're they're, they are on the path of institutional
- 25:58redesign. Kristen Wilson on Twitter asks
- 26:02another short question. So thank you for to Kristen for
- 26:05this and a really interesting one.
- 26:08What can legacy companies learn from startups and vice versa
- 26:12when it comes to AI strategy? Let me talk about what start-ups
- 26:18can learn from legacy companies. And what they can learn from
- 26:23legacy companies is really understanding why legacy
- 26:26companies continue to win and continue to hold advantage when
- 26:30new technological shifts come. And the reason they typically do
- 26:33is because they have existing sources of advantage and they
- 26:37bundle the new technology or, you know, whatever new value
- 26:41they're creating with that. So a very simple way to think
- 26:43about it is what happened with the Internet of Things when you
- 26:46know the Internet of Things are IoT first happened.
- 26:49There were a lot of start-ups that came out and that came into
- 26:54IoT with the consumer Internet mindset, which was that let's
- 26:58build a user base and we'll figure out how to make money.
- 27:01And while that works in consumer Internet and IoT doesn't work
- 27:04because what they were doing was they were embedding a chip
- 27:07inside a consumer device, a physical device, and giving away
- 27:11the device below cost, hoping that they would make money with
- 27:14data eventually. But it was a dramatically loss
- 27:17making model because every unit, you know, device shipped out was
- 27:22going out at a loss. What the incumbents did instead
- 27:25was they made the device more valuable by giving the service
- 27:28and data away for free. Any, any analytics around that
- 27:31data, any, you know, additional personalization and services and
- 27:37remote management. All of those things were bundled
- 27:40with the device. So the device became more
- 27:41valuable and the services got commoditized.
- 27:44So one of the things that start-ups should think about
- 27:46when looking at incumbencies, what's the logic around which
- 27:50the incumbents in my industry run their business?
- 27:52And does the technological shift reinforce that logic or does it
- 27:56dismantle that logic? So that's the first thing you
- 27:58should be thinking about. Pretty much the opposite thing
- 28:01happened when we moved from, you know, on premise software to SAS
- 28:06where the incumbents logic was dismantled because it was
- 28:10structured on a fundamentally different, you know, financial
- 28:13structure of upfront, upfront revenue collection, etcetera.
- 28:19And it was structured on a very different technological
- 28:21architecture. Now, what can incumbents learn
- 28:24from start-ups? I think one of the key things
- 28:26that incumbents can learn from start-ups with any technological
- 28:29shift, but more so with AI, is that the real opportunity with
- 28:33the technological shift is to ask what is fundamentally
- 28:40changing about the way businesses have been built so
- 28:44far? What, what is the, you know,
- 28:46architectural shift that's happening in the industry with
- 28:49this new technology. So a very simple example to
- 28:52illustrate that is that in the mid twenty 10s, Facebook,
- 28:55Instagram, YouTube were declared as the winners of social
- 28:58networking and all of them were structured on a, on a simple
- 29:02logic of the social graph. You needed to follow people in
- 29:05order to get your feet populated.
- 29:07And TikTok came along and three things happened around that
- 29:11time. You know, TikTok was mobile
- 29:12first, so was Instagram, but then in addition to that, with
- 29:154G connectivity, TikTok was also streaming videos and AI had
- 29:19improved. And so mobile video AI, TikTok,
- 29:22combine the three things made the video duration less than one
- 29:26minute. And with that, it was able to
- 29:28capture much more data with people scrolling videos on their
- 29:32mobile and just using that data, it was able to create an
- 29:35interest graph or a behaviour graph through which it was able
- 29:39to populate feeds. And that's why it was able to
- 29:41build a social network without ever having to build a social
- 29:45graph to start with. So that's the idea.
- 29:47You know, when a new technology comes in, it provides you the
- 29:49ability to completely challenge the dominant assumption on which
- 29:54the incumbents were working. So that's what I would say, you
- 29:56know, the two sides should, should learn from each other.
- 29:59Such an interesting point because also one of the things
- 30:02that TikTok did is it shifted popularity from the social
- 30:08graph. With Facebook, it's your
- 30:10friends. If your friends are posting
- 30:12something, then you respond shifted to interest.
- 30:17So if you see a short video that you're interested in, and as you
- 30:20said, because they were now short videos, Tiktok was able to
- 30:24gather all of this data. And that has changed
- 30:29dramatically how social media works.
- 30:32That's right. We have another interesting
- 30:35question. This is from Ronald Saldana from
- 30:39on LinkedIn. And he says, and you started
- 30:43addressing this, he says how can organizations look at AI
- 30:51holistically? In other words, OK, people are
- 30:56listening. They say, fine, you know, we
- 30:58need to do something to how do you do it?
- 31:00If you have to be holistic, you need to ask the what question?
- 31:04What is the future model going to look like with the AI coming
- 31:08in? So let me give a simple example
- 31:09to illustrate this. We talked about TikTok, right?
- 31:12And now what worked for TikTok was that because the video
- 31:15duration was shot, you could watch 10 videos in one session
- 31:20versus one or two videos on YouTube in one session that some
- 31:24some of users would have watched.
- 31:25And that gave it a lot more data and a much more scope of data.
- 31:31Now, if you are a manufacturing company, you would say, well,
- 31:34what do I do with this? This is TikTok.
- 31:36I can't really do anything with it.
- 31:38Well, what happened is that there was a manufacturing
- 31:41company that decided, well, we can apply the lessons from
- 31:44TikTok into manufacturing. And so you saw the rise of a
- 31:48company called Shane coming out of China.
- 31:51Now, when we think of Shane, we very often think of it as
- 31:55somebody polluting the environment and, you know,
- 31:58creating a lot of waste. But really what Shane did to
- 32:02manufacturing was what TikTok did to social media.
- 32:04So Shane, you know, traditional fashion used to work in the
- 32:09model where you would have experts go to Milan and Tokyo,
- 32:13sense the trends, build out a collection for the next season,
- 32:16put out that collection and then come back and then do the same
- 32:19thing for the subsequent season. Very slow, long cycles and you
- 32:24did not know what was going to succeed versus not.
- 32:26So not always a very good ROI on on the entire business.
- 32:31Now what Shane does is it does what TikTok did to social media.
- 32:34It constantly collects data from the Internet, from social media,
- 32:38from TikTok incidentally as well, but it uses all of this
- 32:41data to then determine what people are interested in.
- 32:46It finds micro styles and so it's again learning in small
- 32:49batches. It's creates these micro styles.
- 32:51Based on that, it creates new designs very quickly.
- 32:54It sends these designs to a network of manufacturing
- 32:58partners and ask them to run very small production runs of 50
- 33:02to 100 units. And then it takes them out, puts
- 33:05them in the market, test them, see if they sell off.
- 33:07And based on that it either double down, double s down on
- 33:10the design or it moves to the next design.
- 33:12But all through its algorithm is learning which designs are
- 33:15working versus not and hence improving its the ability to
- 33:18design. So my point is that a
- 33:20manufacturing company looking at TikTok would have said this is
- 33:23irrelevant. Tell me something about what's
- 33:25happening in manufacturing. Tell me what the biggest
- 33:26manufacturing company is doing. But if you're just stuck in
- 33:29that, you're not thinking about AI holistically.
- 33:32So it's less of a question of how initially, it's more of a
- 33:35question of what does this mean for me?
- 33:37You need to take off your blinders.
- 33:39You need to take off the, the assumptions that are trapping
- 33:42you and think about, you know, how does all of this apply to
- 33:47me? What, what are the assumptions
- 33:48that TikTok and log do there, which I can start unlocking and
- 33:51manufacturing. So I would say it's really first
- 33:53that before we do anything else to to do a holistic approach to
- 33:57AI. We have another great question
- 34:00from Greg Walters on LinkedIn. He says most advice about
- 34:05implementing AI seems to be on the same talk track from 1990.
- 34:10How should organizations implement AI today in light of
- 34:17the broader set of issues that you're describing?
- 34:20The challenge is that most talk about implementing AI is from
- 34:23the 1990s because a lot of people who are implementing AIS
- 34:26vendors are the people who cut their teeth implementing
- 34:29automation, as you know, robotic process automation and other
- 34:33forms of automation in the past. And they bring in that same
- 34:37automation hat over here and sell the same thing to to
- 34:40clients. And that's the real problem,
- 34:42because if AI is really about an opportunity for system redesign,
- 34:47you should be talking to fundamentally different people
- 34:49who are actually system thinking, who are thinking about
- 34:51what does this mean in terms of competitive forces, what does
- 34:54this mean in terms of our internal capabilities, and
- 34:56really rethinking on that basis. So the first thing I would say
- 34:59for organizations is think hard about whether yesterday's vendor
- 35:04is still the right vendor to help you tomorrow.
- 35:06If you're thinking hard about whether yesterday's talent is
- 35:09the right talent to help you tomorrow, and you're cutting
- 35:11jobs on that basis, then the least you should be doing is ask
- 35:14whether yesterday's vendor is the right vendor for tomorrow.
- 35:16This is from Chris Peterson, he says.
- 35:19How do we plan for for both technology and cost changes of
- 35:24AI services to reflect the providers actual cost?
- 35:29Is a one year ROI estimate even possible?
- 35:32Cost changes you probably can't figure out in a one year ROI,
- 35:35but system redesign and really the imagining your business
- 35:38model, all of that is not a one year ROI and you should actually
- 35:42be accounting for that separately.
- 35:45As you know, that's for us to stay relevant.
- 35:49It's, it's more of or rather an insurance cost of us staying
- 35:53relevant. It's more of if we don't do
- 35:55these other things, we may not even be in business tomorrow.
- 35:58So the whole point of cutting costs won't make sense if the
- 36:02game in which we are cutting costs is not being played
- 36:04anymore. This time it is from Gus
- 36:07Speckdash, and he says technologies don't create value
- 36:10quickly. In terms of normalized
- 36:13performance. Electricity took 70 years,
- 36:15computers took 40, Internet 25. How long will will this be for
- 36:21AI? The reason technologies don't
- 36:23create value immediately is because technology in itself
- 36:26does not create value. It's the largest socio technical
- 36:28system that creates value. What that means is every
- 36:31technology needs complements which all need to come together
- 36:34before value can be created. And I'll give like a very quick
- 36:36example of what how this plays out.
- 36:39I take the example of the rise of the barcode and Kmart and
- 36:43Walmart adopting this in my book and Kmart adopted improve the
- 36:48checkout speed at the register. But Walmart combined the
- 36:51technology of the barcode with compliments.
- 36:53It had invested in the satellite system.
- 36:55It had invested in its own internal supplier management
- 36:59system. It had invested in EDI and
- 37:02managing data across the supply chain.
- 37:04So it uses the IT used the barcode data to aggregate data
- 37:08across its stores and used that aggregated data to negotiate
- 37:13with suppliers because before that stores used to individually
- 37:16negotiate with suppliers and suppliers, which were the brands
- 37:19had more power. Walmart was the first retailer
- 37:22that flipped that power equation because it created a new socio
- 37:25technical system around the technology.
- 37:27So you have to think about what's that largest system?
- 37:29You can't just keep saying, here's AI.
- 37:31When are we going to see the value with that AI?
- 37:33You have to think about what's the largest socio technical
- 37:35system that needs to be created in in in your industry for that
- 37:39value to play through, just like Walmart did.
- 37:42This is from Sasan. Kumbulpuri, who says if AI
- 37:46learns by sensing and adapting continuously, what stops our
- 37:51strategies from doing the same? That's on LinkedIn and on
- 37:56Twitter, Elizabeth Shaw says. You say AI will change the
- 38:02playing field or the holistic systems in which they act and
- 38:06that and that an AI strategy must address.
- 38:09So what kinds of change of systems are you talking about?
- 38:15So we're really getting to the heart here of the strategic
- 38:19shift. Yes, AI learns and.
- 38:21If your socio technical system does not learn, which is your
- 38:25larger system within which you're setting up AI?
- 38:27If that does not learn, and if it's only the model that's
- 38:29learning, then it's not going to be valuable.
- 38:31The reason, you know, the Shane model works is because it takes
- 38:35a digital physical complex system and ensures learning
- 38:39happens across the board. I'll, I'll give an example that
- 38:42also illustrates the second question, but illustrates the,
- 38:45you know, the, the first point again with that.
- 38:48So think of what's happening in the materials industry today.
- 38:52Materials industry are plastics, rubber, all of these different
- 38:55industries. And in these industries,
- 38:57traditionally what was scarce was the ability to come up with
- 39:00a new formula that you could patent.
- 39:02And then on that formula, you created a material which could
- 39:05then be used in in cars and doors, etcetera.
- 39:09With AI, new formulas can be created much faster.
- 39:12So the constraint around lab testing and all of that goes
- 39:15down. But whether that formula fits
- 39:19the end use case and whether it's valid and whether it's
- 39:22validated, that becomes a new constraint.
- 39:24And in order to solve it, labs, production lines and end
- 39:30channels need to work together to keep testing, gathering data
- 39:34from the market and keep reinforming the labs and the
- 39:36production lines on that basis. Just like Shane does it it, it
- 39:39learns from the market and then it, it improves its future run.
- 39:43So as a whole industry, these industries will have to change
- 39:46To bring back to the first question, that whole learning
- 39:48cycle in and in in order to link it back to the second question.
- 39:51What we're seeing is the reason you need the learning cycle is
- 39:54because the playing field has changed.
- 39:56Creating that initial compound is not expensive anymore.
- 40:00What happens when the cost of creating compounds goes down is
- 40:03you see patent trolls because anybody can use AI to generate
- 40:06compounds. File patents sit on that.
- 40:08But those patents become useful only if they can be applied to a
- 40:12certain end use case for which you need to learn at an industry
- 40:15wide scale. So that's the thing you you
- 40:17cannot in an in a world of AI, you simply cannot say that the
- 40:22learning will happen in the model.
- 40:23Every single part of the socio technical system, every single
- 40:26part of the industrial system will.
- 40:28Have to learn in order to stay competitive, so you're not
- 40:31optimizing. Just for that technology, you're
- 40:34optimizing the broader system where that where that technology
- 40:38fits in your organization, all of its capabilities and so
- 40:41forth. You're actually redesigning the
- 40:43system. So that it can learn because the
- 40:44traditional manufacturing system cannot learn.
- 40:46But when Shane redesigns it so that it takes a small batch test
- 40:50it in the market and then on that basis updates it's
- 40:53algorithm about which design world and which did not.
- 40:56That's how the system is learning by the by redesigning
- 40:59first on LinkedIn. Sufin Ben Sabor says where
- 41:06should AI not be applied? Does everything look like a nail
- 41:12or, you know, we're trying to boil the ocean?
- 41:14So when do we apply and when don't we apply?
- 41:16AIAI should not be applied. Whether it should not be?
- 41:18Applied. What I mean by that is you
- 41:20should never start with AI to start figuring out whether it
- 41:23should be applied or not. You should try to understand
- 41:25given not just AI, but everything.
- 41:28Think about all the forces shaping your industrial playing
- 41:31field, right? AI, datifs, globalization,
- 41:34every, everything that's happening.
- 41:35How is that changing how your industry will work a year from
- 41:38now? And you know, how are the
- 41:39boundaries of your industry change?
- 41:41How is so that's why I used the word playing field.
- 41:43How is that changing where you play and how you create
- 41:46advantage. Start over there because AI is
- 41:48one of the ingredients, even though a very important
- 41:50ingredient, it's one of the ingredients shaping that.
- 41:54Once you figure out what's changing, you have to figure out
- 41:56what your future posture looks like.
- 41:58And then on that basis, you have to ask yourself, well, which
- 42:00capabilities will be value if you become that, if you become a
- 42:03learning materials company versus just a compound creating
- 42:07company, what new capabilities will be value?
- 42:09And so on that basis, you then start to think about if these
- 42:13are the new capabilities, which of these can AI help me with?
- 42:16So you have to really work backwards.
- 42:17You can't just take AI and say we start going to, we're going
- 42:20to start applying AI because everybody else is.
- 42:23I mean, that's again, it's, it's, it's a bit like you want
- 42:27to run it on, you know, you, you, you want to run from one
- 42:32corner of the US to the other, but you're just hitting the gym
- 42:35and you don't have a map of, of which direction to start running
- 42:38in. So you need to know what you are
- 42:40doing before you start applying AI.
- 42:42In other words, don't just. Follow the hype.
- 42:45Absolutely. I mean, that's sort.
- 42:47Of the unsaid subtext to everything, Let's go to another.
- 42:51Question here. Really, this is a good one.
- 42:54They're all good. But this one Mayena K employ
- 42:59Empoyo says Could we consider AI an accelerator of ecosystems or
- 43:07the missing link ecosystems? Ecosystems had been waiting to
- 43:11rise up to their strategic potential.
- 43:14I'm delighted that he asked this because as I read your book, I
- 43:19kept thinking about this in terms of ecosystems.
- 43:23So tell us about that. This is a topic close to my
- 43:26heart because. Before I started writing about
- 43:29AI and reshuffle, for the last 10-15 years I've talked about
- 43:33ecosystems and platforms and, you know, written a very popular
- 43:37book on that topic, Platform Revolution.
- 43:39Now the way ecosystems have worked.
- 43:41So what are ecosystems? First of all, ecosystems are
- 43:45essentially actors with misaligned incentives that have
- 43:50complementary capabilities which need to be brought together so
- 43:54that value can be created. So the whole point of creating,
- 43:57you know, managing an ecosystem is to align the incentives just
- 44:00as the capabilities are already aligned.
- 44:02And so when you're thinking about how you you organize an
- 44:09ecosystem together, take the example of the container again,
- 44:12you know, the container shipping example.
- 44:13I started with the whole global logistics ecosystem.
- 44:16Tracks, trains, ships, manufacturers aligned because of
- 44:20standards. They agreed on common standards,
- 44:22they agreed on a unified contract.
- 44:24So that's what I typically think of as coordination, which starts
- 44:28with consensus. You need the consensus between
- 44:30the players to align them, and then the coordination starts.
- 44:34Now, in the case of AI today, I believe there's an opportunity
- 44:38to start coordinating and aligning the ecosystem even
- 44:41without consensus because AI can take information from various
- 44:46players, even unstructured information, and make sense from
- 44:49it and use that to then guide decisions and actions between
- 44:52different players. A very simple way to think about
- 44:55it is that when you, you know, book a trip, you get your
- 45:00flights from one place, you get your hotel booking from one
- 45:03place, you have your itinerary on a spreadsheet, and then you
- 45:07you manage your activity somewhere else in your e-mail.
- 45:11And there was no way for all of these things to talk to each
- 45:14other. But today you can throw all of
- 45:15them in into an LLM and ask it to create a structured
- 45:19itinerary. And that's obviously a very
- 45:21simplistic way of how that alignment happens.
- 45:24But just imagine if there was a way for all of these players to
- 45:28start using agents to communicate with each other.
- 45:31They would not even have to work on the same interfaces and align
- 45:34beforehand. You, you would have an ability
- 45:37for an agent to go out, identify the right information from
- 45:41different sources and manage your trip alongside you.
- 45:44And so that's the idea of a coordination without consensus.
- 45:47Once enough consumers or once enough demand is aggregated,
- 45:52because this coordination without consensus has happened,
- 45:54then the consensus can come in place.
- 45:56So the, the primary way I believe that AI will change how
- 46:01ecosystems work is that it will not have to wait for consensus
- 46:05before coordination can happen. And so a lot of a significant
- 46:09part of the economy where coordination is broken because
- 46:12that consensus is difficult, we will probably see solutions
- 46:15coming in over there. Let's talk about.
- 46:18Jobs describe the division of Labor between AI and people and
- 46:27the implications. There are two or three things
- 46:30that we need to think. About over here.
- 46:32So the first thing is that when people say AI won't take your
- 46:34job, but someone using AI will, that's a complete fallacy
- 46:37because it assumes that substitution of labour is a bad
- 46:40thing and complementarity is always a good thing.
- 46:42Now, we've obviously seen in the case of Uber drivers that the
- 46:46drivers who used to work before Uber came about, you know, the
- 46:49London cabbies and so on, they did not benefit from
- 46:52complementarity because once Google Maps came in, mapping
- 46:57helped the less experienced drivers get access to the same
- 47:01pay and reduce the pay of the more experienced drivers.
- 47:04So in general, complementarity can actually reduce your pay
- 47:08while retaining your job. So complementarity is not always
- 47:11a good thing. And the second thing is that you
- 47:14don't always need substitution or automation to remove the job.
- 47:17And I take the example of the typist in the book.
- 47:20When the word processor came in, people would have said the word
- 47:22processor won't take your job, but someone using a word
- 47:24processor will. And typist would have tried to
- 47:26figure out how to learn about, you know, how to use a word
- 47:29processor. But what that, you know,
- 47:32confuses is again, what I talked about, it focuses on the task.
- 47:35It assumes that the typist job exists because of the task of
- 47:38typing. But that's not really true.
- 47:40The typist job existed because of the constraint in the system,
- 47:44which was that before the word processor, managing edits on a
- 47:49document was very expensive. And so you the value of typing
- 47:53or, or good typing was higher because it reduced the cost of
- 47:56edits. Now the moment the cost of edits
- 47:58collapse to 0, go back to the point, you know, if the
- 48:01assumption changes, the job goes away.
- 48:04In this case, the job of the typist went away because
- 48:06suddenly inefficient typing was not expensive.
- 48:09Anybody could type with the word processor.
- 48:11And today the the, the task of typing never got automated, but
- 48:15the job of the typist went away. So my point is complementarity
- 48:18is not always good and substitution is not, you know,
- 48:22always the only reason for jobs to go away.
- 48:24So we had to think about this a little more, you know, with a,
- 48:28with a lot more nuance. And the, the one thing that I
- 48:31would say we need to think about is stop thinking about your jobs
- 48:35or jobs in general in terms of tasks and what AI can do to the
- 48:38tasks. Because if you think of it in
- 48:39terms of tasks, you'll always say, what can AI not do today?
- 48:43And I'll start doing that now. The the two fallacies with that
- 48:46is it assumes AI will not improve and it assumes the
- 48:48system will not change. And both those assumptions are
- 48:51wrong. So focusing on tasks and trying
- 48:54to these skill based on the tasks is actually a fallacy.
- 48:57What you should be doing instead is focus on the constraint.
- 49:01And the constraint is important over here because when AI
- 49:05improves execution, everybody goes into this execution mode.
- 49:09You know, everybody's vibe coding, vibe marketing, they're
- 49:11just executing with without thinking about anything.
- 49:14The value shifts to whoever can manage the constraint.
- 49:17I'll give another example over here.
- 49:19You know, think of the anaesthesiologist.
- 49:20I mentioned that in the book, an anaesthesiologist, You know,
- 49:25every single task he performs in the operating room is performed
- 49:29by a machine. It's automated.
- 49:30He's managing the machines, but the reason he's paid extra high
- 49:33is not because he's a really good machine manager, but
- 49:36because he manages the risk in the room of ensuring that the
- 49:39right amount of anesthesia is going out at the right point to
- 49:43the patient. And so a very common constraint
- 49:46that commands a lot of value is risk.
- 49:49If you can manage the risk in the system, you can command a
- 49:52lot of value. The final point I will say, and
- 49:55I'm not saying risk is the only constraint, I'm just saying
- 49:57don't look at the task, look at the constraint.
- 49:59When AI comes in, the old constraint goes away, but new
- 50:03constraints will emerge somewhere else.
- 50:05The final point I will say is that in addition to what we've
- 50:09talked about with AI, if we go back to the, you know, Google
- 50:12Maps and Uber example, the pay of the driver did not go down
- 50:17just because more drivers came in.
- 50:19The pay of the driver went down because the pay was now set by
- 50:23an algorithm. In the case of Uber, drivers are
- 50:25now being managed by algorithms. So there's increasingly A
- 50:28distinction between two types of jobs about the algorithms job.
- 50:31Those, you know, the Uber data scientist who creates those
- 50:34algorithms and is paid very well in equity and below the
- 50:37algorithm jobs who were this very little agency.
- 50:41And one of my key points in the book is that when once AI comes
- 50:44in, it will have the same effect that Google Maps had on driving.
- 50:49It will have the same effect on knowledge work.
- 50:50Because the more you complement people with AI, some jobs get
- 50:55increasingly hollowed to to the bare minimum.
- 50:58And at that point, they become amenable to being allocated by
- 51:02algorithms. They get pushed below the
- 51:05algorithm. And at that point you don't have
- 51:06agency anymore. Elaborate more on.
- 51:09The jobs above and below the algorithm because there are
- 51:14profound implications for all of us and many different kinds of
- 51:18roles. The first question came in from.
- 51:21Wei Yi, I think, and she's followed my book and once she
- 51:24read the book, she reached out to me saying that in her role,
- 51:27which was paid social and, you know, paid programmatic
- 51:32advertising and data data analytics as well.
- 51:37A lot of jobs that used to be about the algorithm have
- 51:39increasingly moved below the algorithm, partly because, you
- 51:43know, very sophisticated knowledge workers have
- 51:47constantly worked alongside these algorithms and trained
- 51:50them. So that a lot of the knowledge
- 51:52of allocating that work has gone into the algorithm.
- 51:55And what has remained with the human has pushed them below the
- 51:58algorithm. And so the, the, the levels of
- 52:02pay. I, I hope I'm doing her example
- 52:04justice, but my point is that's a very good example of a
- 52:07knowledge work that five years back was valued really well.
- 52:10But in the meantime, the workers were training the algorithm and
- 52:14work that they were left with went below the algorithm.
- 52:17The agency went away from them and into the algorithm.
- 52:20And this increasingly will happen with AI coming in,
- 52:23because the issue is not whether AI can perform a task or not and
- 52:29whether you know you have any tasks left to perform or not.
- 52:32The issue is whether it takes away so much from your work.
- 52:36It takes away it standardize, standardizes your work to such
- 52:40an extent that you become very commoditized.
- 52:42Anybody working alongside AI can suddenly deliver the output that
- 52:47you do. And there have been many studies
- 52:48over the past two to three years which show that less skilled
- 52:52workers, when complemented by AI, are able to move and deliver
- 52:56better, much better output than higher skilled workers when
- 52:59complemented by AI. Which essentially means that
- 53:02even though AI helps everybody, it leads to a flattening across
- 53:06the board. It leads to less of a divergence
- 53:09between high skilled workers and less skilled workers, which ACE
- 53:13compresses the skill premium. But then the more they get
- 53:16complemented by AI, the more the work that they are left with,
- 53:19which is uniquely performed by them and not by their peers,
- 53:23gradually starts shrinking. And if it reaches a point where
- 53:25you know like a delivery worker today is not doing anything
- 53:29which involves tacit knowledge because Google Maps is doing the
- 53:32navigation for them. When it reaches that point where
- 53:34all the differentiated work has been absorbed into the
- 53:39complementary technology, then you get pushed below the
- 53:42algorithm. What advice can you?
- 53:44Offer to individuals to all of us in our careers to address
- 53:50this at very fast. Please don't get into thinking
- 53:52about. New skills without first
- 53:54thinking about what's the new system, because everybody is on
- 53:56this reskilling Ted treadmill, but they don't really know what
- 53:59they're reskilling towards. So think about what is going to
- 54:03be valuable in the future and think about how you can reskill
- 54:06towards capturing that value. But if you don't think about the
- 54:09future system then you might just be reskilling in the wrong
- 54:12direction. Most people have no idea how.
- 54:14To think about a future system. That's why we focus on skills.
- 54:17So what do we do? You will intuitively know.
- 54:20That there are certain ways that you can play in certain ways you
- 54:22can't play. So the example I use in the book
- 54:25as well, which I find pretty interesting, is that when the
- 54:27Internet came, you know, happened, the value of a
- 54:32magician's skills went around dramatically because the
- 54:34magician is paid high, you know, well, and the audience pays him
- 54:38well because of the sense of wonder that their trick creates.
- 54:41But with the Internet, you immediately had these people who
- 54:43would deconstruct and push out the the secret behind any new
- 54:48trick. So the cost of creating a trick
- 54:50is very high. The cost of deconstructing and
- 54:52spreading it to everybody is very low.
- 54:54So the value of a trick dramatically collapsed.
- 54:56So what you see magicians doing today is they very rarely create
- 55:00fundamentally new tricks. They take the old tricks, they
- 55:03repackage them with new stories, with new spectacle, with new
- 55:07narratives to recreate the wonder for which they used to be
- 55:10paid. So a magician today is not
- 55:11necessarily paid for the wonder created by this skill at the
- 55:16trick, but for the wonder created by the spectacle that
- 55:19they are now packaging around the trick.
- 55:21So my key point is that when your main source of competing
- 55:26actually gets commoditized as well, you should really think
- 55:29about why your stakeholder was paying you in the 1st place,
- 55:32what about it was valuable for them, and how you can
- 55:35reconfigure that value in a fundamentally new way.
- 55:38Finally, two last questions. And sort of tweet length
- 55:42bundles. What advice do you have for
- 55:45companies? I mean you want it in 140 or
- 55:48two? 80 characters.
- 55:50Well, really, you know, it can be a don't, don't, don't look
- 55:55inside your company to figure out what to do with AI.
- 55:57Don't look at what you're doing today to figure out what to do
- 55:59with AI. Think about how AI will change
- 56:02your playing field and how it will change what advantage looks
- 56:05like. That will tell you what the new
- 56:06competitors are going to look like.
- 56:07That will tell you what new industries you're going to have
- 56:10to enter, compete in, etcetera. That will tell you which
- 56:12capabilities will be valuable in the future.
- 56:14So the more inward focus you are in a world of constant change,
- 56:18it's not about AI, but in a world of constant change, the
- 56:20more you focus inward and you say we this is what our core
- 56:24competency is, the more you are blinded to the opportunity that
- 56:28lies out there. And finally, what advice?
- 56:31Do you have For government policy makers and regulators,
- 56:36the most important thing is. To, to really understand that
- 56:42there are a lot of mechanisms of protection that used to exist or
- 56:48institutional mechanisms of protection and value
- 56:52distribution that used to exist, whether it was unions or whether
- 56:55it was taxation. I'm, I'm not saying that those
- 56:58are already efficient mechanisms, but what all these
- 57:00players need to understand is that the date at which those
- 57:05institutions move is, you know, a fraction and a very tiny
- 57:11fraction of the rate at which innovation moves.
- 57:13And so they have to, you know, if, if organizations have to
- 57:17create learning organizations and not just allow learning
- 57:19models, regulation has to create learning regulation, you know,
- 57:24it has to be more distributed, multiple points of understanding
- 57:29what's happening versus a single top down system through which
- 57:32regulation is being delivered. So we need a fundamentally new
- 57:35way to think about policy. The current model will not keep
- 57:38pace with what's happening. All right, well, this has been
- 57:41a. Very action-packed episode.
- 57:44An enormous thank you to Sangeet Paul Chidori.
- 57:49He's advisor to over 40 Fortune 500 CEOs and he's written this
- 57:54new book called Reshuffle. Sangeet, thank you so much for
- 57:58being here with us today. Thank you so much.
- 58:01This was so much fun. Thank you and a huge thank you
- 58:03to everybody. Who participated?
- 58:05Your questions as always, you guys are amazing.
- 58:07You got your questions are great.
- 58:09Before you go, subscribe to the CXO Talk newsletter and connect
- 58:15with Sangeet and with me on LinkedIn.
- 58:19Thank you so much everybody. We have incredible shows coming
- 58:23U. We'll see you again next time.
- 58:25Have a great day.