On this episode of The Six Five Insider, hosts Daniel Newman and Patrick Moorhead welcome IBM’s Rob Thomas, SVP and Director, Software and Chief Commercial Officer, along with Dr. Dario Gil, SVP and Director of Research.
Their discussion covers:
- The dominating topic of AI and how it’s changing the way enterprise gets things done
- How enterprise leaders and decision makers need to respond to the opportunities of AI
- The transition from +AI to “AI-First” and how businesses can best utilize AI to transform how work gets done
- IBM’s new strategy to helping enterprises become “AI-First” including an overview of its WatsonX platform
Be sure to subscribe to The Six Five Webcast so you never miss an episode.
You can watch the full video here:
You can listen to the conversation here:
Disclaimer: The Six Five webcast is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded, and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors, and we ask that you do not treat us as such.
Transcript
Patrick Moorhead: Hi, this is Pat Moorhead and we are live at IBM Think 2023. Big news, big discussions are AI. I mean AI is pretty much everywhere, not that the hybrid cloud wasn’t important, but it’s all about AI this week.
Daniel Newman: Well Pat, I think we’re all seeing where this is heading. And so over the last few years, we’ve seen all the development with hybrid cloud, that Red Hat acquisition for IBM has really come into conscious of how big and important it’s going to be. And now, we’re seeing what’s going to get built on top of it, and that is in fact AI.
Patrick Moorhead: Super exciting. And we’re here for a Six Five Insider. But hey, let’s introduce our guests. Rob, Dario, great to see you again.
Rob Thomas: Nice to see you both.
Patrick Moorhead: I feel like it’s been a threefer, you’ve been very generous with your time, with the industry analysts, saw you both on the big stage, but things are really happening here at the show, and it’s super exciting. And by the way, there were no doubts in my mind that IBM was going to show up big with generative AI, it’s not like you haven’t been working on it for years and years and years, but love the perspective that you bring and welcome to the Six Five Insider.
Rob Thomas: Thank you for having us, really glad you guys are here. And it’s been an amazing, I’d say, past few years. But if you think about our history with Watson, it goes back nearly 13 years.
Patrick Moorhead: Right.
Rob Thomas: I think we learned a lot in the era around machine learning, deep learning. It’s really hard, you spend a lot of time annotating data. Some things are successful, some things are not.
Patrick Moorhead: Right.
Rob Thomas: Dario and I started the work, I guess, 2020, a little bit before that, on foundational models. And we saw that this is something that could really make AI more accessible.
Patrick Moorhead: Right.
Rob Thomas: And so it’s pretty exciting now to be presenting this to the world and giving everybody a sense of where we’re going.
Patrick Moorhead: Absolutely.
Daniel Newman: Yeah, that brings up a really good question for you guys, at least I think it’s going to be a good question.
Patrick Moorhead: Oh, we’ll see.
Daniel Newman: Early and often in this episode, Pat. But AI’s the top of everyone’s mind right now. Over the last six months, it went from something I think that people were starting to feel into their lives in little ways to something that’s right out in front of you. I think you could talk about whether that’s ChatGPT, you could talk about the AI arms race that’s going on, but that’s a lot of upfront consumer stuff. The enterprise hasn’t necessarily been addressed as much, but I feel like it was really addressed here at Think. Talk a little bit, Rob, about how this moment, this inflection is also really, really big for the enterprise.
Rob Thomas: I mean to the credit of OpenAI, they did inspire a lot of interest in this topic. Since that’s happened though, every CEO I talked to, the question is we’re unsure of what to do. So we’re convinced there’s something here, but how do we do this in a way that it’s going to have a positive impact for our business, we can avoid any of the pitfalls that get written about frequently? Our focus with Watson X and what we’ve announced this week is how do we deliver something that’s trusted, that’s scalable, that’s adaptable? And part of the adaptable piece is we have IBM models. Dario will talk a little bit about the models that we’ve built, how they can be utilized, but we also want to be open with clients and say you can build your own model. You can work with open source models. I think this is going to be the key for how businesses start to adopt generative AI and foundation models.
Dr. Dario Gil: Yeah. I think that element of the strategy is really important. If you look under Watson X, the platform has been very carefully thought through to do the AI piece of training, validating, tuning, and then doing inference right at the moment of deployment. But what is the data store associated with enabling you to build the data piles that are necessary to be able to train those models or fine tune those models? And then the governance framework, how do you have an end-to-end lifecycle management of the models? Because as we know in this field, the job is never done.
Daniel Newman: Right.
Dr. Dario Gil: Right? You have a family of models, you have it and then what happens? You have either new data that you want to incorporate, maybe some data you are using you no longer want to have in the training set, right? Or there’s a new regulatory framework that allows you to have to comply by doing X, Y, Z. So that governance framework is really important. And then the companion idea to all of this is a data and model factory that you always have in production where Watson X is the engine that allows you to do that. And we operate one. So we create a family of models across natural language, around code, or geospatial data, et cetera. And then our clients can also operate their own data and model factory and you can bring open source models. Another thing we’re really big believers on, the pace of innovation of what is happening in the open AI community is extraordinary.
Daniel Newman: Right.
Dr. Dario Gil: And I think a myth busting thing that we can do right away is if people believe that the future is this one company or two companies that have the magic model that everybody’s going to use, that is going to be flat out wrong. And if you look at just what has happened in the open field, it’s phenomenal of how creative people are, how quickly they compress the models and the new novel thing. So we want to bring all that innovation to the client.
Patrick Moorhead: So with this new technology, maybe it’s three-years-old, right? How should enterprises be looking differently at AI? I mean it seemed to be a little bit of a bolt-on, which is okay, I have my enterprise and we’re going to add it. I think you’re referring to it as the show as AI plus.
Rob Thomas: Yeah.
Patrick Moorhead: And I’m curious, how are you recommending to clients that they look at AI in this new generative, excuse me, more foundational models?
Rob Thomas: If you go back to maybe 2017, AI adoption was starting but not very material. It’s now doubled in the last five years. And I think we’ve started the transition from companies think about let me do my normal business and then maybe I’ll do a little AI on the side. I’ll do some experiments.
Patrick Moorhead: Sparkle.
Rob Thomas: Right.
Patrick Moorhead: Yes
Rob Thomas: To what I would call AI first. And I think every company now is at a fork in the road, they have to decide am I going to be AI first?
Patrick Moorhead: Right.
Rob Thomas: Or am I going to let my competition do that? I don’t think many people want their competition to do that. So I do think there’s a unique moment in time. What’s different and I’d say special about this time is there is also a macro catalyst, which is the focus on productivity that’s happening in every country, every industry right now. And I think the early applications for foundational models are going to be about doing more with less. It’s going to be about productivity. It’s not going to be betting on can I find the next revenue growth vector? That’s important, but there’s a lot that can be had in terms of automating repetitive tasks.
Patrick Moorhead: Right.
Rob Thomas: Running your core business more effectively. So there is a macro driver here around productivity.
Daniel Newman: Yeah. It feels like every company has the opportunity to participate. I know I had some conversations here and actually I spoke to the Wall Street Journal the other day and we were talking about is this just a big enterprise thing? And even the word enterprise starts to change because every company can get bigger and still be smaller.
Patrick Moorhead: Right.
Daniel Newman: And that’s one of the biggest probable inflections of this AI era is that you no longer have to have the same resource volumes to hit the same revenue volumes. It’s like hedge funds, you can get really big or very small in other industries though. But I think that’s really the question that people are asking, whether it’s the industry asking, whether it’s CEOs that you’re talking to asking, whether it’s the media is asking is what does this mean in practicality? What is the practical process here whether you are a 500 person company or a 50,000 person company right now, what is the practical next steps to think around AI? And Dario, I’ll let you go first on this one.
Dr. Dario Gil: So look, one first element that we’ve been highlighting is this idea of digital labor, right? And in digital labor is we all perform many tasks in our day and there is a subset of them that have the characteristic of being well mapped to what AI can do today. They can be because of the repetitive nature or they can be things because the access to data that they have and implemented best practices better than any individual could do. So coding it could be implementing best practices, a good example of that. So what it means is that as a company in the associated productivity is you can be more ambitious on the number of projects that you can complete in time or you can be much more efficient and lower their rate, right?
For example, if you think in software it’s traditional 30, 50% of the time spent on the bugging, well could you imagine that number being half and the implications that that would have in terms of quality control? So those are all things that are immediate applications. And I think what is happening is people are starting to see, as Rob pointed out, that it is actually possible. I mean the implications for what has happened with the code, it’s amazing. I mean if you all get to experience it, it’s really impressive. You can improve 60, 70% of all the code that you ride is now suggested to you and you can just hit tap and adapt it. That has really profound implications on both quality and number of products that you can actually launch and improve.
Patrick Moorhead: So Rob, my guess is you’re spending most of your time with clients here and they must have a lot of questions for you in the green room or the pseudo green room. I guess we have green grass here. We were talking a little bit about where are these conversations because IBM offers a lot more than AI, but the conversations have been a lot about AI and I think that’s a good reflection of what you said and how it resonated with the problems that they have. But I’m sure you’re also in the question of how do I get started? How do I put this to work, right?
Rob Thomas: There’s two big trends happening here and I think to Arvind’s credit, when he took over as CEO back in 2020, he charted the next chapter for IBM around hybrid cloud and AI.
Patrick Moorhead: Right.
Rob Thomas: And you look at where we are now on adoption of containers, I would say the acceptance, the embracing of hybrid cloud, I don’t talk to any client that’s just running on a single public cloud anymore. I think people have gotten comfortable with that.
Patrick Moorhead: Yes,
Rob Thomas: There’s more to do, there’s always more work on application modernization. How do we get price for performance across multiple cloud deployments? So that’s continuing, but that’s understood at this point. AI is the topic where people are still a little bit unsure. That’s why we’re trying to bring clarity to what we think are the use cases for businesses, which is around digital labor, which Dario mentioned. The story I loved yesterday was Edward Logan who’s the CEO of Sport Clips, this shows you how this can scale up and down.
Patrick Moorhead: Right.
Rob Thomas: The fact that he’s using Watson Orchestrate now to automate how they do their hiring, how they’re going to bring on 50% more stylists in the next year, I don’t think anybody thought that a company like that, a franchiser of hairstylists would be an early adopter of AI, but that’s how accessible the technology is. So you’ve got this notion of digital labor, you’ve got IT automation, just make the jobs of the technology teams easier inside a company. Customer service, we’ve been going after customer service for a number of years. It gets better when you bring foundational models and cyber security. So those are I’d say the four broader domains where we see immediate applicability. Every company, those are relevant for. So it’s not industry specific per se. That’s why I think people are saying, “All right, we’re going to get over the hump of being unsure.”
Patrick Moorhead: Right.
Rob Thomas: “This is something we can actually understand and put into practice.”
Patrick Moorhead: As analysts, sometimes we have to simplify for the outside world. And what was simple for me was multi-cloud, multi-model. Dario, can you talk a little bit about what that is and why that’s important?
Dr. Dario Gil: Well I think a very nice synergistic aspect of the IBM strategy is that they build on each other. AI is an inherently distributed workload by necessity, right?
Patrick Moorhead: Right.
Dr. Dario Gil: Because where you do training in a data center and where you do inference, which could be all the way at the edge device, exploits the properties of it has to work in a distributed environment.
Patrick Moorhead: Right.
Dr. Dario Gil: So therefore, you inherit the value of the property of a hybrid cloud architecture. So by natively building on top of OpenShift and extended it with other open source packages, we’ve done a wonderful collaboration with Meta, with PyTorch. We’ve done a fantastic collaboration with the Ray community and others with Hugging Face as well in terms of interfaces. So now, what we’re giving our clients is the AI workload itself can run everywhere and inherently it’s needed. So you no longer have to engage with a client about an intellectual argument about why the benefits of running in different places is good is like AI runs that way.
Patrick Moorhead: Right.
Dr. Dario Gil: So that’s one property of the multi-cloud nature. And then the second one is one model will not rule them all. Multi-model is way better. Why? I’ll give you an example, Watson Code Assistant. Watson Code Assistant, we have taken an umbrella foundation model that now is embedded in Red Hat Ansible Automation Platform. That model is 35 times smaller than the equivalent co-pilot model and is higher performant. It’s better. So it’s cheaper to run.
Patrick Moorhead: Sure.
Dr. Dario Gil: Right? At inference time, it’s higher performing. So the reality of it is when you get down to it, using the generic technology, which is the foundation model approach, but then applying it to model specificity and use case, you deliver more value to their clients. So that’s what people are going to do.
Daniel Newman: Yeah. Rob, you brought up digital labor as an application. I think that’s a really interesting one. And by the way, I do love that Dario brought up the smaller foundational models.
Patrick Moorhead: Yeah.
Daniel Newman: I can’t tell you how much time you and I are probably spending, I know in my advisories, I’m spending with people trying to deal with sustainability and the fact is this is a exponential moment for compute. And so deploying more GPUs, more horsepower, more compute, more data center, companies are trying to deal with sustainability. By the way, sustainability is an application. I would love for you to just run through for me all the applications that you’re seeing. You mentioned digital labor, I just teased sustainability, but there’s four or five applications that I think IBM highlighted here at Think that are really practical that all enterprises could be thinking about building right on top of Watson X.
Rob Thomas: Sustainability is a great one, which is as much about data as it is the application of AI, can you actually collect data and understand how to get to carbon neutral as an example? Cybersecurity is a significant one. There’s no amount of humans that can be thrown at the cybersecurity process anymore that will solve the problem. So at this point, it is all about AI. I gave the example yesterday of Novaland. They’re a real estate company that was doing a thousand manual interventions every day. That becomes impossible. We can’t hire enough people to do this. Using AI and Watson, they’ve now reduced that to less than a hundred that humans actually have to look at and address. So it’s just the nature of technology. As it gets more complex, more penetration, you’re going to have to have AI to make how we work just a lot easier. The other philosophy that’s changed for us is we want to be customer number one for everything that we build.
Patrick Moorhead: Right.
Rob Thomas: And we shared the example of what we’ve done in our human resources where we’re now getting 94% automation what was before had to be done by an actual human. This is about making our own employees more productive, but our clients also look a lot like us. So we’re confident if we get great results on IBM, then we’ll get great results with clients.
Daniel Newman: Yeah. I think the augmentation thing, I just want to double click on that because there’s a lot of confusion, Rob, in the market. I know Arvind’s words got a little twisted in an article about where he talked about the 7,800 jobs and he was really talking more about the jobs on a factory floor of yester year, they then got replaced because robots came in. But there were still jobs managing that and there was all kinds of new jobs that got created with every industrial revolution that came through. But you just mentioned that you take 94% of an HR process that used to be done. It’s not all bad though. And I want to make sure we point on that because a lot of it’s about upskilling, right? I mean you have to be talking to the customers about now that you’ve automated this, we’re going to upskill the workforce, we’re going to do more, solve more important problems.
Rob Thomas: This is about a shift in work. Let me give you another example. NatWest, the financial institution, we’ve done a lot of work with them on automating customer service. They’re now able to handle 80% with software. That doesn’t mean they got rid of their customer service representatives, they shifted them to the more difficult problems.
Patrick Moorhead: Right.
Rob Thomas: So in the same time that they’ve done this automation, their NPS, their net promoter score, client satisfaction has gone up 20 points. It was never going to go up 20 points if they had everybody dealing with things like help me set up account, help me change my password, really basic repetitive tasks. So this is about augmentation and ultimately, it’s about how do you better serve customers?
Dr. Dario Gil: By the way, I mean since your semiconductor backgrounds and the experience, I mean look at what happened with electronic design automation, right? And you had the designers and people were counting each transistors and timing it around it. And then at some point, there were very clever and sophisticated ideas. Imagine transistors are free, imagine we had enough automation that you could design any product that you wanted? All of a sudden, that freed up a different way to do things. So I think that there are many of these and that was an example of not even a simple process, it acquired a huge level of skill. And yet, we all as an industry said actually we’re not going to do that anymore. That is going to be handled by computers. But that didn’t stop either the growth of the industry, the growth of the number of people that could create products with it.
I think the same happens in a corporation, right? You have all these skills that you are doing and applying to different tasks. It’s actually we can do those skills better now this way, so now, let’s focus on doing this other one. And the history has been that we always find new ways to tap onto talent and exploit it. So I agree, there’s a lot of twisting that happens around that. And Arvind’s point on the 40% of the people were doing agriculture, right, 100 years ago, and I don’t think we missed the fact that we’re doing that and we only can do that with one and a half percent. There’s still more jobs and it’s still going. So I think it’s going to be a similar story on this that we’ve got to go and tackle.
Patrick Moorhead: Yeah. And by the way, I’ve never met an engineer who actually liked to go in to transistor level and optimize. It was always given to the folks, maybe the new people or something like that. And what did we see? We saw a democratization of semiconductor. There’s more semiconductor IP now, probably 20 fold than there was a few years ago because of this fractalisation and improvement in the processes that engineers didn’t want to do anyways, right? So I think we’re going to see that as well. Same discussion we had on even virtualization of it’s going to crater, it’s going to reduce jobs because we don’t have people do this, that just exploded the amount of industry that was out there. By the way, the one big ding, ding, ding bell that went off in my head probably 18 months ago, I was questioning why is IBM leaning into a hybrid cloud and AI? Wait a second-
Dr. Dario Gil: What’s the connection?
Patrick Moorhead: Didn’t these guys do this already? So the big thing was first of all, you saw this latest foundational model wave coming before it was cool, okay? One of the first companies to identify this. And the second thing is it’s very hard to do any of this generative AI foundational models without having a hybrid model, which by the way, Rob, to your point, I have yet to talk to a Fortune 500 company that doesn’t have more than one IS provider. Thank goodness, we’re like a teenager in cloud where we’re not debating this. So all these great things that are happening. Anyways, I just wanted to share that with the audience and with you. We’ve had a great strategic discussion on the needs, where this is coming from, how IBM is strategically approaching it, but we glossed over the products that you’re offering. I want to go back to that and maybe Dario, we could start with you. What were the three things that you announced? What is Watson X? What are the three areas and why do they matter?
Dr. Dario Gil: Yeah. So Watson X is our platform for builders of AI and what it allows us to do is we’re taking all the capabilities that we have built over many years to do machine learning and deep learning and govern the lifecycle all in there. But on top of that, natively now created to be able to do foundation models and generative AI. Watson X is composed of three elements, Watson X.ai, dot data and dot governance. AI allows you to train, validate and do inference on the models you create and it gives you a model family, dot data is your data store that allows you to do incredibly fast querying built on all open source technology. By the way, I’ll highlight which is incredibly important for the extension and compatibility of all of this.
And governance, it allows you to do the whole lifecycle management of the models themselves so that you know whether there is drift, whether you need to do updates, it gives you the fact sheets around the models. So that platform, which by the way, it’s a big deal. The only other horizontal platform that IBM was discussing was the hybrid cloud platform with Red Hat. We are building and we are delivering now a data and AI platform that powers all our software products, but on top of that, for clients and our partners to build on.
Patrick Moorhead: Love it.
Daniel Newman: So Rob, I just want to ask one more question before we take off and get back to Think. By the way, very excited for today. It’s going to be a great day to think and I want to talk about the speed just a little bit because I’ve never seen anything move this quickly. And of course, we always say that the law of diffusion of innovation with each new thing is the period gets shorter and shorter. But this has been incredible.
Patrick Moorhead: Right.
Daniel Newman: We’ve seen in six months literally companies that had AI nowhere on their radar to being the number one priority. You’re seeing companies that were doing AI as an experiment saying we’re all in. And even us as small enterprise owners are saying… I’m looking at every process in my business right now and saying whether it’s an internal operational process, Rob, or even just the way we might advise a company like IBM. Can I do it with generative AI? I’m asking that question. How fast do you think this goes? The conversations you’re having, are your customers doubling down? Is Arvind’s IT being the most protected is going to turn into a board focused initiative that’s going to be funded, that’s going to accelerate this deflationary value that that AI has on the future of enterprise?
Rob Thomas: AI has definitely become a board level topic. And if I think about our core business, it’s probably the second thing in the last decade, or maybe there’s been three. Cybersecurity has become a board level topic. Hybrid cloud has become a board level topic. Now AI is there and I think it’s time for that. I also put this in the category though it seems like it’s happened very fast. It’s actually been happening very slowly for a long time with a lot of iteration and now, the interest has spiked. And we were making enormous investments in training and model building when nobody really wanted to talk about it, sure. But we do think it’s what differentiates what we’re doing in Watson X which is we’ve already made the huge upfront investment, so clients can focus on bringing their proprietary data set focused on inference, which is not inherently as expensive as doing training. So I think it’s been a slow, constant way for quite a while. Now there’s a huge spike in interest, now people have to figure out how are we going to do something?
Patrick Moorhead: Exactly. Gentlemen, I can feel the excitement at IBM Think like I’ve never… I’ve been in and around IBM for 33 years and yes, that’s when I started working and watching, I’ve never seen this amount of excitement. I mean hybrid cloud is the right strategy, it’s what every customer wants, hybrid multi-cloud. AI, we just talked for 30 minutes about that. Oh, and let’s not forget about quantum, right, which is just around the corner and is a discussion we had before one to two years before we get this to this utility where you can do more with it. So gentlemen, thank you so much for coming on the show. You’re Six Five veterans and I really appreciate you doing this insider for us.
Rob Thomas: Nice to see you both. Thank you for having us.
Dr. Dario Gil: Nice to see you. Thank you for having us.
Patrick Moorhead: Yeah. So this is Pat and Dan at IBM Think 2023 here in Orlando doing a Six Five Insider where we have the most relevant companies with the most important executives. We hope you love the show. Tune into some of our other IBM Think 2023 coverage here. We appreciate you, thanks so much.
Author Information
Daniel is the CEO of The Futurum Group. Living his life at the intersection of people and technology, Daniel works with the world’s largest technology brands exploring Digital Transformation and how it is influencing the enterprise.
From the leading edge of AI to global technology policy, Daniel makes the connections between business, people and tech that are required for companies to benefit most from their technology investments. Daniel is a top 5 globally ranked industry analyst and his ideas are regularly cited or shared in television appearances by CNBC, Bloomberg, Wall Street Journal and hundreds of other sites around the world.
A 7x Best-Selling Author including his most recent book “Human/Machine.” Daniel is also a Forbes and MarketWatch (Dow Jones) contributor.
An MBA and Former Graduate Adjunct Faculty, Daniel is an Austin Texas transplant after 40 years in Chicago. His speaking takes him around the world each year as he shares his vision of the role technology will play in our future.
