Latest / CXOTalk / Former Accenture Tech CEO Reveals Top AI Mistakes to Avoid | CXOTalk #895
Transcript
- 0:00Paul Doherty has had an incredible career at Accenture.
- 0:04He was group CEO for technology. He was their chief technology
- 0:09and innovation officer. He's written two books.
- 0:13He's advised many startups, sits on boards and is now AI advisory
- 0:19chair at the large private equity firm TPG.
- 0:23Today on CXO talk number 895, we take an AI reality check with
- 0:29Paul to examine what works, what fails and what's next.
- 0:34I'm your host, Michael Krigsman. Let's get into it.
- 0:37I spent a lot of years at Accenture.
- 0:39I was there for 38 years until I retired last year.
- 0:43And it was an amazing vantage point from which to see the
- 0:47transformation, transformational power of computerization and
- 0:51digital technology and everything that happened over
- 0:53those 4 decades. And to put it in perspective,
- 0:56you know, in a way that some of you will appreciate, the first
- 0:58production system I wrote was in was in punch cards on a
- 1:02mainframe. So it's a, it's quite an arc
- 1:04from there to where we've gotten to today.
- 1:06And, and I've always been passionate about technology and
- 1:10my role at Accenture is always about leading what's next and
- 1:13leading us into what's next. So I'm still an advisor to
- 1:16Accenture. I still do some work there.
- 1:19I took on a role, as you said, with TPG, which has been really
- 1:22exciting. TPG is a great firm.
- 1:24A lot of partners are based in in the Bay Area at the heart of
- 1:26technology. They've been very tech forward
- 1:29PE firms with, you know, large PE firm, 260 billion of, of
- 1:34assets across a lot of different platforms.
- 1:36And it gives them a way to invest in big companies, the
- 1:40large companies that are impacting the direction of
- 1:42technology as well as through some of their platforms involved
- 1:46in the the leaders of tomorrow. And it's been a, it's been great
- 1:48to, you know, come on board and work with TPG and we can talk a
- 1:51little bit more about my role there.
- 1:52Where are we today? What's the state of AI?
- 1:55Are we in a bubble? Is all of this hype real?
- 2:01What's going on? This isn't a bubble.
- 2:03It's real. Like in terms of the real
- 2:05transformational power of the technology and what's already
- 2:09happening is real. We're just at the early stages
- 2:12of it, which is where you see some of these bubbly things
- 2:14happening. Let me talk about, you know, a
- 2:16little bit about, you know, kind of what's real and you know,
- 2:20what's hype, what's real and what's missing.
- 2:22I think what's, you know, one example of hype I think is
- 2:24around AGI, artificial general intelligence.
- 2:28And one of the problems with AGI is there's not really and agreed
- 2:31upon clear definition of it. But generally speaking, people
- 2:34talk about AGI in the sense of general purpose AI that can
- 2:38exceed the capability of anything a human can do.
- 2:42And I think we're far, far, far from that in that there's a,
- 2:46there's a group of people in a group of eminent people that I
- 2:48respect that claim we're, we're either they are pretty close to
- 2:51it. I, I, I, I'm with a different
- 2:53camp of people who, who would disagree with that.
- 2:55I think, you know, what we have today are really amazing
- 2:58technology that approximate a lot of the things that people
- 3:01can do, but that are far from real AGI as I would think about
- 3:05in terms of human capability. Mustafa Suleiman, who you, many
- 3:09of you may know from DeepMind and then inflection now leading
- 3:11AI at Microsoft. Mustafa Suleiman has, has had a
- 3:15series of posts that you should go look at if you haven't.
- 3:18The recent post about consciousness and the the risk
- 3:21of assuming of giving AAI agency of things like AGI and
- 3:26consciousness because it can lead to very bad human outcomes
- 3:29by misinterpreting what the AI really does.
- 3:31And I think Mustafa explains it really well.
- 3:34So I'd encourage you to read his post and this, but I think
- 3:36that's a, a bubbly aspect of lots of people like talking
- 3:38about AGI companies are claiming they're already at AGI, but I, I
- 3:41really don't believe that's that's the case.
- 3:43We've had AI around for 70 years.
- 3:46The the learning branch of AI with the transformer technology
- 3:49in large language models is real and it's amazing.
- 3:53If we just froze AI right now, no more innovation, no more
- 3:56products. Freeze it where it is right now
- 3:58with large language models. I think you'd have 5 if not ten
- 4:01years of business adoption of what the technology can do
- 4:04today, much less the massive amazing arc that it's on.
- 4:09If we were to freeze it today, that implies that companies have
- 4:13seen benefits, they're getting benefits.
- 4:15Just what are the what are those benefits?
- 4:17Generally speaking, there's benefits in the obvious areas
- 4:20you see like coding and software engineering, lot of advances
- 4:23being made. There are a lot of great tools
- 4:25there in customer service types of applications and marketing
- 4:29applications. Again can go into detail.
- 4:32One organization from working with that had 30% increase in
- 4:36productivity and 60% increase in Net Promoter score and
- 4:39satisfaction by introducing new forms of, of generative AI into
- 4:44their customer service organization.
- 4:47And then in, in different, in different industry kind of
- 4:50functions and, and, and such that we can kind of unpack a
- 4:53little bit. So, so there's real, you know,
- 4:55real opportunity in, in there in there already.
- 4:58But just circling back, maybe I just want to finish one thought
- 5:01on the, on the what's missing from AI, because it's really
- 5:04important to where things are right now.
- 5:07I think everybody's a little too centered just on the large
- 5:10language models in the scaling law around large language
- 5:13models. And I think to get the next
- 5:14level, to get to anything approaching AGI and to get to
- 5:18the next level of business improvement, we really need to
- 5:21focus on other forms of AI. And there's a lot of things out
- 5:24there, large quantitative models, physical or sorry,
- 5:26liquid AI, a number of other things that are that are getting
- 5:30more at symbolic AI. Yan Likun has been talking a lot
- 5:33about this. And I think the combination of
- 5:36the amazing things large language models can do and
- 5:38bringing out that learning branch of AI together with the
- 5:41symbolic AI techniques is what's really going to propel us to,
- 5:44you know, to kind of a next, next level of what we can do in
- 5:47business and next level of what AI itself can do.
- 5:50What about the impact on organizations?
- 5:54You wrote a book and a year ago pulled out and sections from it
- 6:02that appeared on the cover of Harvard Business Review and and
- 6:08you said this. Most business functions and more
- 6:10than 40% of all US work activity can be augmented, automated or
- 6:16reinvented with Gen. AI.
- 6:19So as you look back a year later, how does that prediction
- 6:25stand up? Yeah, it was embracing Jenny I
- 6:28at work was the was the article on Harvard Business Review and
- 6:33yeah, that came out of research, as you said from our book and
- 6:351500 organizations, lots of lots of research that we did and I
- 6:40would a year later we were low the 40% was was low.
- 6:44I think we we we significantly understated it.
- 6:46We've done some additional research and you know you can,
- 6:49you can peg it above, above 50% now in terms and that's looking
- 6:53at work hours that people expend, you know that all of us
- 6:57do that can be impacted by or that will be impacted by
- 7:01generative AI where we are. So the coverage, so to speak,
- 7:05the surface area that it impacts, yes, you know, is, is
- 7:08increasing or the potential is increasing.
- 7:11The actual what's happening is still pretty light, you know.
- 7:14So let's just be be real about this today in terms of, you
- 7:17know, adoption, we're still in the early stages from some other
- 7:21research that that we've done the we would, we would take a
- 7:27different cut and say that about 15% of organizations have made
- 7:31significant, you know, 15 have made significant some
- 7:34significant progress on scaling AI somewhere in the
- 7:37organization, 15%, another roughly 1/3 of organizations.
- 7:41So the balance of, you know, to get you to 50% are close on the
- 7:45heels and you know, kind of doing things, working, working
- 7:47to scale. And about 50% are still more in
- 7:50the kind of experimentation, trying things stage and maybe
- 7:53individual use cases and parts of the organization, those types
- 7:55of things. So there's so yeah, there's over
- 7:5850% of the work can be affected, but that's that's not happening
- 8:03today. That's the potential and
- 8:04organizations are working to get there.
- 8:06And I think one of the myths about generative AI now is I
- 8:10think you'll hear a lot of the tech industry talk about it.
- 8:13It's like there's an easy button, just buy this and
- 8:15implement it and hit the easy button and you're you transform
- 8:18your business. For general AI, that's not the
- 8:20case. This is really hard, probably
- 8:22harder than the pre, the technology that come before.
- 8:25This is harder than cloud and the cloud transformation because
- 8:28it's invasive in your organization.
- 8:30It's changing processes, it's changing roles of people to re
- 8:33educating people in transforming your organization.
- 8:36You had to do some of that always of course with past
- 8:38technologies, but that's this is taking it to another level with
- 8:42AI, which is, which means it's we got a lot of work ahead of us
- 8:44to get there. Paul, you use the term invasive.
- 8:49What is it about AI that requires this invasiveness on
- 8:58organizations as companies adopt it?
- 9:02There's 2 dimensions of it, one of which is the invasive, the
- 9:06technology is invasive and organizations are starting out
- 9:11with legacy systems, with technology debt.
- 9:14That's been talked about a lot for years.
- 9:15So I won't go into that in detail.
- 9:17But you know, the average organization still spends, you
- 9:19know, 50% to 2/3 of their budget keeping lights on on legacy
- 9:22rather than new capability. It's much higher than that a lot
- 9:26of organizations and that creates, it's not just in that
- 9:29it's an investment challenge, but also creates a real anchored
- 9:32roadblock to to getting new capability introduced.
- 9:35And AI is significantly different.
- 9:37You have a lot of work to do at the data level if your data
- 9:39fabric and getting that in in shape, you have a lot of work to
- 9:43do at your architecture and application layer, at your
- 9:45middleware layers. And some of that technology is
- 9:48immature. We can dive more into this if
- 9:50you or your audience would like, but we we lack categories of
- 9:53products right now. There is not really, you know
- 9:56good mature agentic middleware out there for organizations to
- 9:59tap into and that's still an evolving product category.
- 10:02So on that technology dimension, this is hard work.
- 10:05The second dimension that makes it hard is the process
- 10:07dimension, because the, the unlike cloud, which a lot of
- 10:11benefits of cloud were technical, I can move to elastic
- 10:14capability and external data centers and things and reduce a
- 10:17lot of costs. And yes, there was business
- 10:19impact, but you know, you got a lot of benefits to get the
- 10:22benefit from AI. It's it's going invasively to
- 10:24processes and changing the way people work.
- 10:26And that's hard to do. There's change management
- 10:28issues, there's organizational issues often where things cut
- 10:32across silos, you know, for example, working with the
- 10:35organization, which that's looking at AI and customer
- 10:39service. But they realize the big benefit
- 10:41for them isn't just in customer service, it's integrating across
- 10:44product management and product innovations that they can really
- 10:47dramatically shorten this the, the, the cycle time between
- 10:50getting product, you know, customer feedback and product
- 10:53innovation out there. And that's really hard to
- 10:55compress those two functions into one and make it all work
- 10:57the way they want to with AI. And that's an example of the
- 11:00invasive on that process side. And then you can get into
- 11:03cultural change and everything else that's part of it.
- 11:05But those are at least a couple of the dimensions that make it
- 11:08difficult. So according to what you're
- 11:10saying, and I don't want to put words in your mouth, AI is a
- 11:15forcing function for relatively broad organizational change with
- 11:22tentacles and implications that extend in a lot of direct.
- 11:25Is that an accurate way of saying re saying what you just
- 11:28said? Your AI strategy is kind of your
- 11:30business strategy. And yeah, there's a big debate
- 11:33in organizations. Do I, you know, they want to do
- 11:35AI, Do they need an AI strategy? I'd argue generally speaking,
- 11:38what you're really up to, what you're really doing is thinking
- 11:40about your business strategy with AI as a significant enabler
- 11:43along with the whatever other things are in the mix that
- 11:45you're dealing with macroeconomic issues and
- 11:48everything else that's going on. So yeah, so so you have to get
- 11:52in in into the, you know, like what's happening in the business
- 11:55now, there's things you can do more easily.
- 11:57You can you get some additional incremental productivity by
- 12:00rolling out copilot tools and doing a variety of things.
- 12:03So yeah, you can do that. But to really get some of the
- 12:05bigger benefits and look at the impact and how you can get there
- 12:08faster than your competitors and some of these broad
- 12:11organizational and process transformation that that takes
- 12:13more work. This type of change, it sounds
- 12:18like you're describing BRP and business process transformation
- 12:23from, you know, 25 years ago. How are we different today?
- 12:27Client server technology is what propelled business process
- 12:31reengineering the famous Michael Hammer book and everything that
- 12:34some of the audience will remember from the 90s.
- 12:36And that the message there was similar that you can't, you
- 12:39can't create the Super highways of tomorrow by paving the dirt
- 12:42paths of of yesterday. And that that's similar to the
- 12:45message that I was just saying. I think it's just broader.
- 12:48If you look at what you look at the surface area of an
- 12:50organization that client server impact and it was quite low,
- 12:54whereas AI is impacting pretty much everything that an
- 12:57organization can do. So I think it's, it is a similar
- 13:00message in, in a similar need to step back and rethink about it.
- 13:05And that's why I call it rather than re engineering, I call it
- 13:07reimagining. You really have to step back and
- 13:09look at the art of the possible because you don't just have your
- 13:12human workers and customers in the mix.
- 13:13You have you have your, your, your AI agent.
- 13:16So you have AI customers and partners that you know that are,
- 13:19that are going to be part of your interaction that you're
- 13:21thinking about. So it does really require this
- 13:23kind of reimagine reimagination capability.
- 13:26Subscribe to the CXO Talk newsletter right now.
- 13:29Go to cxotalk.com because we have live shows and they are, we
- 13:33have amazing, really genuinely amazing guests.
- 13:36So subscribe to the CXO Talk newsletter.
- 13:39Do it now. We have a question from LinkedIn
- 13:44on LinkedIn from Larry, Larry Rubinacher.
- 13:48It's a very specific question, he says.
- 13:50What do you see as some of the key AI apps for the utilities
- 13:55industry? You know, utilities is an
- 13:57interesting industry because you've got services, customer
- 13:59service and billing and things like that.
- 14:01You've got manufacturing like process manufacturing, like
- 14:04capability generation, transmission, distribution,
- 14:07etcetera. You've got construction type of
- 14:09capability. So utilities kind of a microcosm
- 14:12maybe of, of a little bit of everything that you have across
- 14:16other industries. So I think there's clearly
- 14:18opportunities in that customer service area.
- 14:22But like we're seeing across industries and I, I know that's,
- 14:25that's an area that I've done some work in and I know that a
- 14:28lot of utilities have, have done, have done work in.
- 14:32Yeah. So that's, that's I think a
- 14:34clear area of opportunity in the mix in the, in the kind more the
- 14:39operations side. There's been some really
- 14:41interesting work and innovative work that some utilities have
- 14:44done around using AI for things like wildfire prevention.
- 14:48You know, the West Coast of the US in particular, where fires
- 14:51have been a cause of major outages and, and system issues.
- 14:55They're using AI in a couple ways.
- 14:57One is to detect fires more quickly using AI analysis of
- 15:00satellite imagery and all sorts of things you can get.
- 15:02But they're also doing, one utility in particular is with my
- 15:07working with my former company is doing, is doing some work to
- 15:10look at how you, how do you equip workers to go into very
- 15:13hazardous wildfire environments to, to, to preserve, you know,
- 15:18protect the, the infrastructure, which is very difficult to do.
- 15:21So a lot of implications like that in utilities as well.
- 15:25There's some interesting things going on in the in emissions
- 15:28management for utilities to to help utilities monitor
- 15:32emissions, for example, things like a meth, you know, tracking
- 15:36methane leaks or tracking methane burn off in different
- 15:40ways to allow utilities to operate more sustainably.
- 15:43So I think there's a lot of use cases in utilities because
- 15:46again, it kind of pulls in a lot of these use cases that happen
- 15:50across the across the other other industries.
- 15:52Utilities just as a challenging environment sometimes because of
- 15:54the rate based model and the way capital's allocated, sometimes
- 15:57hard to invest, you know in those opportunities, but I think
- 16:00there are a lot of lot of opportunities.
- 16:02We have another question and this is from Joy Tuhan and she
- 16:07says, Paul, it was great working with you at Accenture.
- 16:11What are some KPIs and metrics that organizations should use to
- 16:17show AI is creating real business value?
- 16:21It's a really important question.
- 16:23If AI isn't moving your the performance metrics on like the
- 16:26Page 1 scorecard of the company that they track, whatever they
- 16:29did in organizations, they're going to track all sorts of, you
- 16:33know, key metrics to their business from from, you know,
- 16:38day services outstanding to accounts receivable, also to all
- 16:42sorts of productivity measurements of manufacturing
- 16:45and other things. And I think if if AI isn't
- 16:48making an impact on those things that I don't think you're
- 16:51looking at AI, right, in terms of of the impact it can have.
- 16:55So I think it's a matter of looking at the KP is that are
- 16:58critical in your business and understanding which of those can
- 17:02be dramatically impacted by AI in in my Accenture role, what we
- 17:06did to pick up throughout the on Joy's question, you know, what
- 17:08we're looking at with AI, even before generative AI, with just
- 17:12AI, you know, classical AI, so to speak, before we were looking
- 17:16at our, for example, our software development processes
- 17:18and looking at how do we improve the efficiency and productivity
- 17:21of those every year. And we were improving them about
- 17:2310% a year, every year. And we would we'd measure that
- 17:27to make sure we got those those kinds of outcomes.
- 17:29So I think that's I think the key is not necessarily a whole
- 17:32new set of outcomes, but to be very intentional about the AI
- 17:37projects that you're pursuing and be able to tie those to the,
- 17:41you know, the Page 1 metrics that you manage as a as a
- 17:45company. Of course you're going to have
- 17:46other metrics for your AI programs, other programs working
- 17:49right, other successful in delivering value and everything.
- 17:51But that's I think that's key. The one other element of that
- 17:54that is going to mention, but you know, maybe it's some other
- 17:57context, but that is critical is is I think organizations also
- 18:02need to avoid AI use case Itis. There's too many organizations
- 18:07that I walk into that very proudly show me 100 use cases of
- 18:11AI across the whole world. Do they get so many?
- 18:13They're so excited, they got so many.
- 18:15I'm like, you know, pick seven, pick five, pick a number, but
- 18:17focus and pick a few and, and really tie them to something
- 18:20that that makes a difference to your company and make them
- 18:22happen. I, I don't think it's victory to
- 18:24have a lot of use cases. I think it's a victory to have
- 18:26you, you know, a small number of use cases that make a
- 18:28difference. So you can, you know, show the
- 18:30value, show the impact on your company and then move to do
- 18:32more. I have to reinforce that
- 18:35comment. There are so many folks who
- 18:39describe who who have chopped up the organization, evaluated the
- 18:43organization and identified all of these use cases.
- 18:47But as you said, identifying use cases and putting people to work
- 18:52in so many different areas does not necessarily help you reach
- 18:57the finish line. I think that's right.
- 18:59I think you can solve small problems and miss the bigger
- 19:01problem. There is one large consumer
- 19:04goods organization that I worked with.
- 19:05We worked on AI in the sales arena and the initial use cases
- 19:09were around how do we get sales leads better to make a sales Rep
- 19:12more effective and all sorts of things.
- 19:14It turns out the bigger use cases were thing, the bigger
- 19:17productivity improvements. I should say we're cutting
- 19:20across use cases. So how do you combine a
- 19:22salesperson's productivity with things like inventory
- 19:26availability and, and, and delivery times so that you could
- 19:31cut across that and, and get things to customers faster?
- 19:34Like, and there's a whole different way of, you know, way
- 19:36of thinking, how can you move product to customers faster
- 19:38rather than just making an individual salesperson more
- 19:41effectively? You'll miss opportunities for
- 19:43those bigger reimagining, you know, if you just get caught up
- 19:46in use cases. This is from Bhavana Bhagat and
- 19:51Bhavana says is the talent transformation with AI for real
- 19:57or will it be another RP? AAI risks being another RPA if
- 20:02deployed the wrong way. I think RPA, you know, RPA was
- 20:08kind of had this problem of being a little fragile.
- 20:10You implemented use the RPA technology Robotic Process
- 20:13Automation for those who aren't familiar with it, which is being
- 20:15able to stitch together and automate processes across
- 20:18different systems. I mean, very powerful technology
- 20:21does a lot of great stuff, but I, I think it hit it hit limits
- 20:25in terms of what it could do and was wasn't flexible to adapt to
- 20:28changes on in the underlying systems.
- 20:30On a generally speaking, I think the, I think the AI could be
- 20:35subject to the same limitations if not deployed very well.
- 20:38So I think that I think the architecture of how you deploy
- 20:42AI is super important. The way you insulate from
- 20:45changes and underlying foundation models is super
- 20:47important. And I think, you know,
- 20:49architects are back for those of you listening, if you're
- 20:52enterprise architects, we need a lot more enterprise architects,
- 20:55you know, to, to shape these AI solutions going forward that
- 20:59really understand all these things because it's super
- 21:01important. And otherwise you can end up
- 21:02with brittle AI systems as well. Now, Bhavani, in your question,
- 21:05you said talent transformation. I, I do think so.
- 21:08I'm not sure exactly what you meant by the talent
- 21:10transformation related to RPA and and AI, but there is a,
- 21:14there is a, you know, I think a broad kind of rescaling of
- 21:18people, you know, develop software engineers, developers,
- 21:21architects, etcetera to learn AI maybe more so and will certainly
- 21:25more so than people needed to to advance their skills with the
- 21:29RPA era. So I think that is part of the
- 21:31journey ahead too. Sachin in Namdar says.
- 21:36He's from Accenture, he says. What are your thoughts on the
- 21:40future of the consulting business going forward?
- 21:44Services, generally tech services, you know, let's just
- 21:47include consulting, systems integration and outsourcing, all
- 21:50that stuff together. Tech services generally, I think
- 21:53is, is is is is is hugely impacted by everything happening
- 21:59with Jenner of AI. And it's probably the biggest
- 22:02impact from it, you know, more of an impact than any of the
- 22:05previous technology changes. And what that means is it's that
- 22:10if the industry's at the center of some of the changes that are
- 22:12happening, which means it's both a threat and an opportunity.
- 22:15It's a threat because I think the existing way that consulting
- 22:18happens, the existing ways that systems integration and
- 22:21outsourcing happens aren't going to be the ways to do it and are
- 22:24already not the ways to do it in the jet AI world.
- 22:26So I think the challenge for any company, for Accenture, for, you
- 22:29know, any company in the industry is how to be first and
- 22:32ahead in transforming to the new processes.
- 22:34So on the consulting side, rather than being having smarter
- 22:38people and good knowledge capital and walking in with a
- 22:40better presentation to the CF OS office that was yesterday.
- 22:44You know, tomorrow what you need, you need to walk in with a
- 22:46foundation model with pre trained on the client's data and
- 22:49in the CF OS office be remodeling their business
- 22:51dynamically in the first meeting that and that's a that's a big
- 22:54transformation on a lot of levels.
- 22:55And that's emblematic of what we what needs to happen in the
- 22:58industry and what, what what organizations like Accenture I
- 23:00know are working on. What is that due to the
- 23:03consulting industry when you have a lot of people whose
- 23:06primary skill was listening and now they have to of course still
- 23:13listen, but there there are so many more dimensions and the
- 23:16talent pipeline has not necessarily prepared them for
- 23:19that. This stuff is hard, as I've been
- 23:22emphasizing a lot as I've gone through, clients need help in
- 23:24doing this. So I think there's a big role
- 23:26for services organizations to help companies do this faster,
- 23:31better, more effectively. So I think there's a big need
- 23:34for this, for services to do that, but they need different
- 23:38shape to it. So an example, you need more for
- 23:40deployed engineers, you know, in the business to more quickly
- 23:42drive change for the client. So it's a it's a matter of
- 23:46transitioning the talent, which I think, you know, companies
- 23:49have been, you know, century as an example, I've been very good
- 23:51at transitioning and retooling talent.
- 23:54It's about rethinking the workforce, the the degrees of
- 23:58leveraging pyramids and things like that will change, you know,
- 24:01depending on the type of work. As you look at this going
- 24:04forward, it's about moving services to software.
- 24:07So what of the services that are provided?
- 24:09Where are there opportunities to platform, platform a ties if
- 24:12that's a word or transition to services kind of the pound tier
- 24:15model in that way. So I think those are some of
- 24:18the, those are some of the changes and organizationally
- 24:20talent wise, you know, technology wise, Michael.
- 24:22And this is from ebru Bayer, who says there are still
- 24:26organizations out there who haven't done cloud
- 24:29transformation. They're lacking data quality and
- 24:33governance. Do you see any prerequisites
- 24:36that organizations should focus on before jumping into defining
- 24:41an AI strategy or executing that strategy?
- 24:44Those are impediments, but you can't wait like like to so it's,
- 24:48it can't be sequential where I'm going to do my cloud and data
- 24:51and then eventually I'll get to AI.
- 24:53So there's, if there's some companies, I've decided that
- 24:55there's some company I've worked with who have looked at and
- 24:57said, man, I'm, I'm in such bad shape on data.
- 24:59I'm just going to do that. And then I'll, I'll deal with AI
- 25:01once I'm done with that. That's, that's a, a small
- 25:04minority that I think most realize.
- 25:06OK. I got a lot of work to do on the
- 25:07data side, but I got to get some elements of AI work and I got to
- 25:10make progress on my journey and they're they're melding the two
- 25:14together. So I think, I think it's you got
- 25:16to look at your specific circumstances.
- 25:17Those are big impediments. Many companies, I'd say most
- 25:20companies have some variety of the challenges that you
- 25:23mentioned ebru, but it's a matter of figuring out how do
- 25:26you parallel track it and and manage that in an effective way.
- 25:29Let's talk about adoption, success and failure.
- 25:33What can go wrong? We have all heard about a recent
- 25:37MIT study that said 95% of Gen. Gen.
- 25:41AI projects fail. Now there are some issues with
- 25:44that study. Maybe you could say it was
- 25:46talking mostly about proofs of concept.
- 25:48Nonetheless, when it comes to Gen.
- 25:51AI, what does failure mean? What causes failure?
- 25:56How do we even recognize it? It came at a time in the market
- 25:59news and when people wanted to seize it and use the message
- 26:02that 95% of projects were failing like the headline in
- 26:06different ways. So I think it was unfortunate
- 26:08that it got the traction in the way it did because I think
- 26:10that's kind of a misleading conclusion.
- 26:13But and I don't think 95% of gender Gen.
- 26:16AI projects are failing. That's very, I have a surface
- 26:19areas of hundreds of AI projects that I see and it it's nowhere
- 26:22near 95% failure. That's generally speaking.
- 26:25People are making progress in different, you know, in
- 26:27different ways. And what is failure?
- 26:29What? How do we define failure?
- 26:30And then? I mean, the way I would define
- 26:32failure is you don't achieve the business objectives you set out
- 26:35to achieve, which gets into one reason for failure is sometimes
- 26:39they they don't, they don't there's no clear North star of
- 26:43what they're trying to achieve or no clear metrics from a
- 26:45business perspective. So that's, that's how I define
- 26:49define failure. There's other, I guess you could
- 26:51define in other ways, but I think that's, that's one way to
- 26:53see, I've seen people to develop great individual use cases that
- 26:56they couldn't scale. I'd say that's an element of
- 26:58failure too, because they realized that, that, you know,
- 27:01they, they developed it great POC on a small scale, but they
- 27:04realized they had, they were missing the data or whatever
- 27:07they might have needed to, you know, to go scale it more.
- 27:10So I guess there's five things that that I find that, you know,
- 27:17companies that are successful are focusing on, and I've been
- 27:20talking about these for a while. These aren't necessarily new, if
- 27:23you've heard me talk in the last few months, but but these are
- 27:25these five consistent things that I've been seeing.
- 27:27One is the value point that I just mentioned, Michael, you
- 27:30know, back to the definition there, you need to focus on the
- 27:33value, have a business case, tie it to metrics, understand what
- 27:36that business case is and in in value generally will mean really
- 27:41looking across the use case artists and looking at things
- 27:43that, you know, reimagine the process that are doing something
- 27:45significant. So it's this value piece is
- 27:47important. Sounds obvious.
- 27:48Why wouldn't everybody do that? But you know, not everybody
- 27:50does. A lot of people are starting in
- 27:52different ways. The second point that you know,
- 27:57reason for success or if you know, obstacle is, is what I
- 28:01call the digital core in data. And, and that was at the heart
- 28:06of the question we just answered.
- 28:07But sometimes companies don't have their their digital core of
- 28:11the systems in place. If you're not on SA PS S4, for
- 28:14example, their new generation of SAP, you don't get access to
- 28:16their AI. So you got it.
- 28:17You got to get your digital core in place and upgrade your SAP,
- 28:20the latest version to get access to the AI as an example.
- 28:22So I think getting a digital core in place and getting the
- 28:24data in place is key and, and companies are generally speaking
- 28:27behind on the data piece, but that's, you know, that's that
- 28:31can cause that can cause failure or limit the for the
- 28:33effectiveness. The third piece is talent and
- 28:37under investing in the people, and this is the talent to do AI.
- 28:40Like you have people who can do the AI, but it's more more
- 28:43broadly the talent across the workforce.
- 28:45Are you training the workforce with what they need to do?
- 28:47I guess there's some best in class examples of companies that
- 28:50are doing a great job of educating their entire workforce
- 28:53on AI, what they're doing, what their objectives are and
- 28:55involved in their workforce. Bottom up.
- 28:57These tend to be an enlightened leaders who who kind of get it,
- 29:00you know, bottom up, you know, in in a corporate, their
- 29:03workforce, great things that companies like like Merrick,
- 29:06like Walmart, I think are examples of companies doing this
- 29:08that that, you know, that are really building the talent base
- 29:13they need going forward and the learning systems they need.
- 29:16The 4th point is responsible AI and really putting the guard
- 29:19wheels around it. So you know, you're not getting
- 29:21into trouble as you do AI that leaders leading companies, I
- 29:23think are doing that well. And the fifth point is viewing
- 29:26it as an ongoing multi year change programs, you're building
- 29:30the the, the capacity for technological and process and
- 29:33organizational change into your approach rather than viewing it
- 29:35as getting one use case or one project right.
- 29:39So those are the characteristics that I see are companies that
- 29:41are they're doing well. Let's talk about private equity
- 29:46and venture capital. Just give us some insight into
- 29:50your role at TPG as AI Advisory Chair.
- 29:55CPG large private equity firm 260, you know, billion in assets
- 30:00and they spanned all these different platforms.
- 30:02It's not just, you know, buy out, you know, capital, it's
- 30:06growth. They have growth capital, they
- 30:07have credit, they have real estate, they have impact funds,
- 30:10they're climate fund and others that are focused on impact
- 30:13investing. And it's, it's really an amazing
- 30:15opportunity, you know, amazing opportunity to combine across
- 30:17those. And the reason they brought me
- 30:19on to the team is it's a very tech forward team.
- 30:22They, they really understand and they've been investing, they've
- 30:24been very successful investors in a lot of domains, including
- 30:27technology for many years. But they really wanted to look
- 30:30at, you know, how tech, you know, where technology is
- 30:32creating opportunities and how to accelerate some of the
- 30:35opportunities that they see. So the role as AI advisory chair
- 30:37is to help with the, with the overall strategy and investment
- 30:40themes and work across those different capabilities that I
- 30:44just discussed, looking at where there's opportunities, helping,
- 30:47you know, on deals that they're doing, you know, new deals that
- 30:51they're doing. And also work on pulling
- 30:55together, you know, expanding the network, pulling in other
- 30:58expertise on AI, building other connections in AI that'll help
- 31:01them, you know, do be be more even more effective going
- 31:04forward. And it's again, been a lot of
- 31:06fun. And given their scale, they can
- 31:07have impact of, you know, kind of on a impact on on a big way,
- 31:11which has been very, very exciting to see.
- 31:14So again, that combined with the VC angle has been a lot of fun
- 31:17to see it from both perspectives.
- 31:19As you're looking at startups and investments, of course the
- 31:24traditional evaluation points have been the product, market,
- 31:28fit the team, and so forth. How does AI now fit into
- 31:36investment analysis and decision making?
- 31:39There's a strategic thing. I'll go through this first.
- 31:41There's a strategic thing you have to look at that's a little
- 31:43different. And I'll talk about, you know,
- 31:45technology related companies first, but it's not just
- 31:49technology, but this conscious and technology terms.
- 31:53I think there's like five ways that AI can impact a technology
- 31:57company. First, it can be no impact.
- 32:01AI doesn't impact it at all. That's number one, rarely the
- 32:04case, but it could be the case that that that's the case.
- 32:08The second is AI enhances, you know, the what, what the, the
- 32:12the SAS model or whatever the company does today and look at
- 32:15work day as an example. You know, the new agentic, they
- 32:17just had some new announcements that rising this week and AI
- 32:20enhancing, you know, the the capability of of what what they
- 32:23do. I think that's that's the second
- 32:25level or second example. The third level is the AI is
- 32:29becomes more important than the underlying SAS product or
- 32:32whatever itself in the AI is the folks.
- 32:35I'd argue that when you look at GitHub, when you look at maybe
- 32:38ServiceNow now, when you look at the future of Salesforce, the AI
- 32:41you know becomes the defining capability around the product.
- 32:45The next level would be AI goes even further and commoditizes
- 32:49the underlying technology. Look at, you know, Klarna and
- 32:53the way, the way this, the CEO of Klarna has talked about the
- 32:56impact of generative AI and how he works with different
- 32:58enterprise software companies. As an example, maybe Zendesk,
- 33:02you know, others, you know, that that fall into this, you know,
- 33:04kind of fell into this this pattern.
- 33:07And then the fifth level is, is AI just just compresses the
- 33:10whole spend in the category because you don't need the, the
- 33:13software. Chegg might be a good example of
- 33:15this. Chegg, the learning company who
- 33:17when the general VI models came out, people flocked to, you
- 33:20know, for free or whatever for very low cost to general VI
- 33:22models. And Chegg lost a lot of its, you
- 33:25know, a lot of its population there.
- 33:28So it goes. That's kind of a little bit of a
- 33:29framework to look at. I think you have to look at that
- 33:31from an investing perspective, really look at that and
- 33:33understand not just today, but if you look at the the move of
- 33:37the markets space in the next years, what the what the impact
- 33:40is. We recently had the CEO of
- 33:42Zendesk on CXO talk discussing their AI transformation.
- 33:47So if anybody's interested in that, we had an in depth
- 33:51discussion. So just go to cxotalk.com to
- 33:54check it out. All right, let's take another
- 33:56question very quickly now because we're we're just running
- 34:00out of time. And this is from Sufian Ben
- 34:03Sabour. And he says what is next after
- 34:07AI beyond AGI and really quickly.
- 34:09Please, Paul. The things to look at beyond
- 34:12what we currently consider is is AI and AGII think, look at
- 34:15physical AII know you said beyond AI, but not enough.
- 34:18Not many people are focused on physical AI, which is where
- 34:20really Elon's going with with X and XA on everything he's doing
- 34:24and where world labs and organizations like that are
- 34:27going. So think physical AI and AI
- 34:29meets the real world as one thing.
- 34:31Look at quantum, still general purpose quantum still number of
- 34:34years off, but some specific impacts of quantum happening
- 34:38fast and related to this world, real world blending thing is
- 34:41robotics and the the massive advances in robotics and these
- 34:45new physical world models that are coming, helping us make, you
- 34:50know, we'll predict on the show right here.
- 34:53We're going to have a a moment by bigger than a ChatGPT moment
- 34:56around physical and robotics of what they can do.
- 34:59That's coming relatively soon and maybe even bigger, bigger to
- 35:02us in, in, in, in realization than, you know, again, the
- 35:05ChatGPT moment. How soon?
- 35:07I think that's within a couple years, you know, in terms of
- 35:09really seeing these physical models and you know that what
- 35:13they can do, you know, say 2-2, maybe 2 years, you know,
- 35:18certainly about five years. Steven Redmond on LinkedIn says
- 35:22how can organizations get ahead on organizing and cleaning their
- 35:27data while not losing momentum on AI adoption?
- 35:31Really think about your business case, because what I see time
- 35:34and time again at organizations is people coming up with great
- 35:36data plans and data strategies and it gets shot down by
- 35:40management, whatever, because there's not a business case tied
- 35:42to it. So I think you need to think
- 35:43about how to link your AI benefits or whatever benefits
- 35:46you have an ERP transformation or whatever with, with getting
- 35:49your data state advanced and link that.
- 35:51Link things together so that your data transition, your data
- 35:55modernization, doesn't get cut off and shut down, which is what
- 35:58continually has been happening to companies for the last decade
- 36:00or more. Rajat Kumar says what will be
- 36:04the future of EMR electronic medical record applications
- 36:08using AI? Are some of the major players
- 36:11integrating AAI into their apps and what kind of integrations
- 36:14will that basically the future of electronic medical records?
- 36:18You know, if you look at the recent announcements from Epic
- 36:20at their event about a month ago, those of you that are in
- 36:22this space, they made a broad set of announcements and a
- 36:25really an expansion of what you think of as EMR.
- 36:28They're expanding epics role well beyond what you thought of
- 36:30as EMR to get into these care provision activities where AI
- 36:35will really have big impact. So I think if, if Epic's view
- 36:39vision of the future is right, you'll see, you know, the EMR
- 36:42providers critical play a critical role in automating a
- 36:45lot of aspects of patient care, revenue cycle management and and
- 36:49the like. There's other views of the world
- 36:50which say that'll build build up around the Emrs and you know,
- 36:54question will be which which vision plays out.
- 36:56Arsalan Khan says responding to your comments earlier about
- 37:01enterprise architect architecture, he is a recovering
- 37:04enterprise architect. He says all jobs will be
- 37:08affected by AI. Are there any that are still
- 37:11untouchable by AI? Physical skills, you know that
- 37:16the people do you go out and do things in the in the real world.
- 37:18I think those will take a lot longer for AI to, you know, AI
- 37:21to impact. I think anything, anything where
- 37:23you're dealing with knowledge and information, it it'll, it'll
- 37:27have impact. That's there.
- 37:28There's we have a lot of data. I can put give you a link to
- 37:31send around based on different labor categories of where
- 37:34generate Jenny, I will automate versus augment and where there's
- 37:37no impact, I can send that out. Maybe as a follow up, Michael,
- 37:40I'll give you the link. And that does it by industry as
- 37:42well as function. We need to talk about
- 37:45responsible and ethical AI. Everybody, every technology
- 37:49company says we are doing responsible AI.
- 37:55What's the reality behind that today?
- 37:57Less than 20%, and that's probably a generous number of
- 38:00organizations are really doing responsible AI.
- 38:02The way I would characterize the need for it, the to do
- 38:06responsible AI, right? I can't get into each one of
- 38:08these, but you need 3 things. You need principles.
- 38:10Everybody's got that you need. I want to be fairness.
- 38:13It's fairness, privacy, security, explain ability.
- 38:16Everybody agrees on the principles.
- 38:17Most organizations have that. The second thing you need is
- 38:19processes. How do you do it and make sure
- 38:22that you're doing AI responsibly?
- 38:24Is it baked into your software engineering practices, etcetera?
- 38:28Most aren't doing that. And the third thing you need are
- 38:30tools and measurement. Do you understand every AI use
- 38:33case across your organization? Do you know what the risk level
- 38:35of each use case is? And you know, what's being done
- 38:38to mitigate any, any, you know, bad outcomes, you know, from
- 38:42irresponsible outcomes, so to speak, Most organizations aren't
- 38:45doing that. So, so we need to, we need to
- 38:48get beyond the principles and lip service to responsible AI to
- 38:51implementing, you know, really rigorous industrialized
- 38:55responsible AI. And like I said, less than 20%,
- 38:57probably a lot less are probably at that stage.
- 39:00That was a harsh commentary. It's true.
- 39:02I've been preaching on this, so to speak for about 8 or 8 years
- 39:06now involved a lot of the organizations that have been
- 39:09advocating for responsible AI and it was preaching totally
- 39:11into the wilderness with 0% listening for a long time.
- 39:15So I think the 20% I view as an accomplishment, you know, we've
- 39:18got we've got more seriously, but I think too many
- 39:21organizations don't really understand there's a business
- 39:23issue because the biggest, the biggest risk of AI is eroding
- 39:28trust. If a is use deepfakes, whatever,
- 39:30think about all the different ways AI can tarnish your brand.
- 39:34Those companies that really develop trust in the way that
- 39:37they're using AI are going to have a competitive advantage
- 39:39over other companies. This is about business and
- 39:41strategic value of your brand. This isn't about a feel feel
- 39:44good thing. I think too many people still
- 39:46think a responsible AI is a feel good thing rather than something
- 39:48that's strategic and essential to your competitive survival.
- 39:51What's your view of government regulation of AI?
- 39:55I think it just needs to be balanced.
- 39:56I think we need a government regulation that encourages
- 39:59innovation, innovation, you know, leading innovation and,
- 40:04and, and has the right focus on preventing the real harm
- 40:09deepfake and some of the real, you know, obvious harm.
- 40:11If they're around. It's that's, it's a lot more to
- 40:13say on that, but that's the one sentence.
- 40:15And one sentence What technologies are you watching
- 40:18now? One new thing I'm watching that
- 40:20I think is in the underappreciated undervalued
- 40:23category or really under thought about category, maybe not as
- 40:26crypto. There's there's a lot of people
- 40:27thinking about crypto, but it tends to be the crypto converted
- 40:30that are thinking about crypto. But crypto as a general purpose
- 40:33technology for tokenization, not just a financial assets, but you
- 40:38know other things, data manufacturing, etcetera.
- 40:41I think is is poised with the recent steps the US has taken to
- 40:46really legitimize in a lot of ways to really have a much more
- 40:49profound impact on businesses and everything we do then then
- 40:53the non crypto community is aware of right now.
- 40:56What is your trajectory for the next steps in your amazing
- 41:02career? I'm going to keep doing all the
- 41:04fun things I'm doing now because this is like, this is kind of
- 41:06what I love to do. This, this arc of technology has
- 41:09been kind of my career and my my life.
- 41:11And when it's stay close to what's happening beyond, that's
- 41:14what more time with my family, which I love to do.
- 41:18And I'm having a great time with that and hopefully getting a
- 41:20little bit better at guitar, which has been my retirement
- 41:23project since I left Accenture. I'm trying to gear up and spend
- 41:26more time in guitar, but all this exciting technology stuff
- 41:29is taking time away. I did notice the guitar in the
- 41:31background. Yeah, I'm getting pitched on a
- 41:33lot of startups to that that can help me learn music faster.
- 41:36That's that's where I'm hopeful. And with that, we're out of
- 41:40time. A huge thank you to Paul
- 41:43Doherty. He is currently AI Advisory
- 41:47Chair at private equity firm TPG.
- 41:50He's a advisor to Accenture. He's doing a lot of other
- 41:53things. Paul, thank you so much for
- 41:55taking your time to be with us. It's always fun, Michael.
- 41:58Thank you and thanks to your audience.
- 42:00This is an amazing audience. That's what makes this so fun,
- 42:02in addition to all the work you put into it.
- 42:04And I love the questions. So thanks to all who are tuning
- 42:06in. Yes, for the folks who are
- 42:08watching, there's so much information that has been packed
- 42:12in today. By Monday, the edited video,
- 42:17we'll do some light editing. We'll be up on our site and then
- 42:20next week we'll create a summary.
- 42:23Watch this video. If you're interested in any of
- 42:25the topics that we've discussed, the replay will be there.
- 42:30You can take your time and you can unpack it.
- 42:34And right now, subscribe to the CXO Talk newsletter, Go to
- 42:38cxotalk.com. We really have amazing shows
- 42:41coming up. We'll see you next time.
- 42:42Thanks so much, everybody. Thanks to Paul Doherty and have
- 42:45a great day.