Democratizing Generative AI – The Six Five Interview with IBM and GSMA

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: The Six Five is back at Mobile World Congress 2024. It’s the third year back. We’re here in the IBM booth. And Dan, this show is already on fire. I mean literally the hallways here are jam-packed. We’re talking network transformation. And as you would expect, a lot of AI. And in the run-up to this last week, what did we talk about? We’re going to be the themes here at Mobile World.

Daniel Newman: I mean, of course it’s not just going to be about AI.

Patrick Moorhead: That’s right.

Daniel Newman: But as we, even The Six Five this year, our summit is going to be focused on unleashing AI. We’re going to see AI as a theme that’s going to be part of every conversation. Look, it is a little bit cliche at this point. It feels like everything is AI, AI, AI. But the fact is, Pat, it’s going to change the world. It’s going to change the network. It’s going to change the way we work. It’s going to change every experience that people have. Maybe it is hype, but maybe it’s not.

Patrick Moorhead: Yeah, truly about what are the tools that are going to improve, not only the outcomes to the consumers, and this is very much a consumer show as well with all the handset and devices. But more importantly, the edge and also the network.

Daniel Newman: Absolutely. Pat. Well look, we’ve got a crazy busy week here, but I am so excited about this conversation that we’re going to have right now.

Patrick Moorhead: Let’s bring in our guests. Stephen, great to see you. Alex, great to see you. This is wonderful. The group that basically underwrites this entire event, GSMA and the Head of Industries at IBM. How you guys doing?

Stephen Rose: Fantastic. Wonderful to be here. Really, great.

Patrick Moorhead: Great. It’s our third year here in the IBM booth. It’s so exciting.

Daniel Newman: Yeah, it is really-

Alex Sinclair: I’m so happy you’re here on behalf of the GSMA.

Patrick Moorhead: Thank you.

Daniel Newman: Yeah, it’s great to be back. And it is really great to see the event at full momentum. There was a little bit of a comeback period I think where everybody had to be comfortable getting back out there. But this year I think it’s full motion, the event is full steam ahead. Let’s start off with full announcements that GSMA and IBM are making together. Alex, tell us a little bit about the news.

Alex Sinclair: Now that we are back, we are really back, we’re super excited about the announcement because we see this as a way to try and democratize AI, make sure it’s not just the big guys in developed markets that have access to it.

So we’re starting small. We’ve got two things that we’re focusing on. The first is around training. So together with the expertise of IBM, we’ve announced something called our GSMA Advanced Learning Platform. So we’re doing AI courses for Telco. And the second thing we’re trying to do is to challenge innovation in Telco Use cases. So that’s through our foundry challenge as well.

Patrick Moorhead: I love that. So Dan and I opened up the segment, it’s kind of a joke, but kind of reality in that it feels like all we’ve talked about the last 18 months has been AI. And like Dan astutely pointed out, sure there’s a little bit of hype there and you have to get people excited, but there’s a lot of reality and part of turning something from hype into reality is truly understanding what some of the challenges are. If I can hit you first with what are some of the challenges you’re seeing or some of the problems that you’re trying to solve your clients with generative AI?

Stephen Rose: Yeah, I mean I think it is very interesting. Look, it is a fascinating space. I think there are three kinds of challenges that we see that you need to overcome as a network operator or as an equipment vendor or whoever you are that’s using AI. The first one is a leadership challenge, in fact. If we think about it, there’s going to be massive systems of distributed innovation inside of these companies. First off, you’re going to turn all of your ordinary users into content creators. So those content creators are going to be using AI, but they need to make sure that that massive distributed innovation, that actually domain experts are now going to be the coders of the future in many respects. Because the barrier to entry for coding is so much lower. So how do you put the governance around that? So it’s two types of governance.

What’s the board that you’re going to run to be able to actually ensure that you’ve got the right kind of Use cases? And are they ethically bound within policies? And the second thing is, how are you actually managing all of the different Use cases you’ve got in the organization? Are you able to inventory them and are you able to have them scrutinized and inspected if necessary? So that’s kind of the first challenge.

Second problem is the economic challenge. What’s the return in invested capital? How do you actually understand what the cost of inferencing is and whether to buy, borrow, or build your own models. Then the third problem is technology. What’s the stack that you’re going to use? How’s that stack going to evolve over time and what’s the best way to incorporate a number of different opportunities to use different types of models within that stack? A number of challenges to overcome as an industry and to learn about, and that’s what we we’re trying to address.

Patrick Moorhead: So Alex, I mean, you’re not an analyst and you’re not a vendor, you’re a CTO. You actually have to make this stuff happen. What are some of the challenges for your point of view, maybe incrementally from what were discussed?

Alex Sinclair: Well, some of the problems are the same. Stephen already mentioned ethics so we’ve done a lot of work around that as well. We have our own AI ethics playbook, but I think part of it is just with any new technology, you’ve got to remove the fear factor. And with AI, that fear factor’s been dialed up to 11. So the fact that you can make, not a data scientist, not a machine language expert, but a regular Joe into one of the content creators that Stephen’s talking about, I think that’s really important. By giving people a little bit of training, but also access to the right tools, things like WatsonX, I think you can go a long, long way.

Daniel Newman: It’s very interesting. I love that you said the fear. I mean, look, there’s a lot to learn still. I mean, IBM, for instance, Pat, you and I have both given a lot of credit for their governance and their focus on governance early on, the ability to deploy a model, understand how it’s grounded, understand how it’s developed, understand how it works. I mean, we’ve seen recent news, and I won’t be specific, but recent news, just about major concerns about how AI models are developed and how they work. We’ve got this kind of continuum of trying to address bias, and you’ve got this continuum where you can over-rotate to it. You got this, we’ve got to make sure that we find balance. This has to be able to add and work in the real world, both for businesses and for consumers, and we have to trust it.

And that’s one of the things that’s really, really interesting. Now, you started off Alex talking about democratization. And so democratization as I see it would be something like GSMA building so that every Telco, every service provider that you work with has access to the information, the learning, the skills, the tools that they need to be successful. Why is it so important in your mind? Because frankly, it would be easy to just maybe focus on a couple of the biggest companies. They’ve got the money, the resources, but you seem to want to bring it to everybody.

Alex Sinclair: Well, look, in our industry, we’ve been dealing with a bunch of gaps since the early days. Remember when you couldn’t get a signal tone in lots of places?

Patrick Moorhead: Yes.

Alex Sinclair: That was the coverage gap. Then we have the inclusion gap, the gender gap, the usage gap. We don’t want an AI gap. We don’t want just the top 25 Telcos in the world to use this. There are 700 Telcos in the world and all of them are our members. So we have a duty of care for those people, but it’s also to make sure that anybody can get involved because it’s wrong to concentrate this sort of stuff in the hands of just a few people in a few countries. We are really excited about it and we think through the use of the tools and the training, that’s a great start. We’ve got other plans, which we won’t go into yet, but it’s a really good start. We want to avoid an AI gap.

Patrick Moorhead: Anything to add to that?

Stephen Rose: Yeah, I think the one thing to remember is again, if you turn users into content creators and you make the technology as pervasively available as possible, then you know that for a fact that what you can guarantee is that the greatest ideas will not come from the biggest players in the world. They won’t necessarily come from the smallest players, but they’re going to be somewhere where you don’t expect it. The beauty of the program that we’re actually releasing, which effectively gives GSMA members an opportunity to run a very extended trial of what’s an next platform, which enables them to come up with different ideas on how to use the technology in any particular different domain.

The most important thing is you get that technology into domain experts. The domain experts are the ones that actually understand the problems that they’ve got in their ordinary ways of working. They’re the ones that say, “I wish I could do something”. And whatever that is, finish that line, no pun intended. So at that point there, what you’re able to do is give that technology to those people and they’re the ones that are going to come up with a great idea. So we believe that if you put it out there like that, that’s the best way to make the industry turn towards an AI first type industry sector.

Patrick Moorhead: I want to do a double click. We’ve been talking pretty at a high abstract level here, which always is a good place to start when it comes to AI. There’s a lot of segments to the overall value chain here. You have the end device, which is connected to the edge network, which is connected to a different network, which is typically connected to the core, and then everybody’s talking to each other. Where is the force multiplier? Where’s the biggest value in AI going to hit? And we’ll start with you.

Alex Sinclair: Well, I think the simplest answer and the glib answer is all of these and more.

Patrick Moorhead: Yes.

Daniel Newman: It’s my favorite way to answer any question. It’s always “And…”

Patrick Moorhead: And analysts, we always answer it like that.

Daniel Newman: Safe.

Alex Sinclair: But look, obviously this thing starts in the super end where most of these models are trained in data centers and everything else. But you mentioned Edge, right? And I think for us, that’s pretty exciting to take some of this path to the extreme edges on device. If you walk around this show, I think practically every major terminal is vendors talking about what they’re doing. But yeah, if you want an AI that’s highly performant, highly interactive, you’re going to have to use the Edge and that’s good for Telco. We’re super, super okay with that.

Patrick Moorhead: Anything to add on to that?

Stephen Rose: No, I think that, again, what we’re doing is we’re actually moving away from just thinking about large language models and we’re moving much more into small language models. Small language models are going to be able to use much more refined sets of data. Those that data is often going to be privileged types of data, which means that actually you can come up with more niche Use cases. And that’s great because actually if you think about it, that will be a small case. It will be at the Edge of the network. It’ll be actually working on more readily available hardware, for example.

And so again, what you’re going to see is proliferation of Use cases and a proliferation of models at different places in the network as well. So again, with those new architectures available to you, completely changed the way in which AI will be implemented across various different intersections in the network.

Patrick Moorhead: One thing with certainty that I think is really good for everybody at this event is generative AI capabilities. Quite frankly, machine learning is very much a better golf club to use for certain workloads. It’s going to mean more data and more data means goodness. And it has since we started with the first G, and even, well actually there wasn’t a whole lot of data before that was all calls, but the more data the better. It is interesting, I found it is how even the on-device AI capabilities and what they can do to optimize the network and then on the network side, what they can do when they have a certain type of a handshake. There’s a lot of discussion around that that I find incredibly interesting. Not just from a density point of view, let’s say on the Edge network, but also in the quality of service and the speed of answers that are going to be the expectation once we get two, three years down the road. Super exciting.

Daniel Newman: Yeah, it’s interesting, and I think what you were sort of alluding to in simplest terms is AI is not really new. I mean the machine learning capabilities, the Telco service providers have been using these for a long time. I’m sure you’ve been talking to them as the CTO at GSMA for a long time about how to implement generative AI, brought AI into the consciousness of the globe and has created this overarching trend that is now the theme of Mobile World, the theme of Davos.

But the thing is, in the end, this is all about productivity, it’s about efficiency. And one of the things that drives productivity to any industry is ecosystem. This is one example of a partnership, but Stephen, I’d just love to get your take. How do you see partnerships proliferating beyond this? Is the industry going to all come together? I know you had the AI Alliance, you saw the AI-Ran thing got announced today. There was another alliance, so partnerships seem to be in vogue.

Stephen Rose: Yeah it is. And again, it’s really important that we actually see those partnerships flourish, and they are. I mean, if you look just at some of the work that we’ve been doing in the last year, particularly around some of the Edge Use cases, real-time RIC, for example, where you’re going to have the X apps and the R apps actually driving different quality of service. Well, that’s a constellations between IBM, Juniper and Nokia. That’s another example of where you get domain experts in different areas. Security, of course, the network performance, and then the AI from our site. You bring that together, you certify and you can take it to market and scale. That’s a perfect example of the kinds of constellations we’re going to see. And then we’re going to see them much further back actually.

How do you actually understand where the relationship between a GSI, a partner like IBM’s technology and somebody that really knows the customer care space inside and out, for example. How do you transform that? What I’m really most excited about is that we are focused less on just basic use cases around cost control and cost optimization. We’re seeing growth and experience type cases coming to the fore. That’s where we get to change the industry and the way in which we can drive new revenues as well.

Daniel Newman: And Alex, I presume that GSMA is thinking endlessly about the collaborations and partnerships of the thousands of companies here.

Alex Sinclair: That’s basically in our DNA. We convene in the industry so it’s an ideal partnership. If you think about it, we have about 700 Telco members, we’ve got 300 others, and we’re getting more all the time. They’re spreading into every other sector. 5g is a great door opener. This whole industry 4.0 thing gives us permission to go and talk to the manufacturers, the coal miners, you name it, we’re doing it. But if you think about it, we’re not one company. We’re an organization of companies. So if you take IBM’s expertise and you partner up with 700 Telcos around the globe locally, then that gives you the kind of scale in our industry. That’s all we ever talk about scale. So it’s really important.

Daniel Newman: Well, Alex, Stephen, I want to thank you so much for being part of The Six Five here. We love having these conversations. You actually were the first for the event, but we’ll have many, many more. Stephen, Alex, let’s have you back again soon.

Stephen Rose: Looking forward to it. Thanks so much.

Alex Sinclair: A pleasure. Thank you guys.

Stephen Rose: Cheers.

Daniel Newman: All right everyone, thanks so much for tuning in here to The Six Five. We are in the booth at IBM at Mobile World Congress 2024. We’re talking about AI, but we’re also talking about the proliferation and the innovation that’s going to go into network transformation. For Patrick and myself, hit subscribe. Join us for all other shows here at Mobile World and beyond. But for now, got to go. See you later.