January 29, 2025 • 55 mins
In this episode of Smart Talks with IBM, Malcolm Gladwell speaks with Ric Lewis, IBM’s Senior Vice President of Infrastructure. They discuss how hardware capability has enabled the matrix math required to run large language models. Furthermore, they delve into some creative examples of how to put AI to work: from your bank to your local coffee shop. Ric underscores the importance of infrastructure in unlocking the potential of AI, helping businesses harness their data to drive transformative outcomes.
This is a paid advertisement from IBM. The conversations on this podcast don't necessarily represent IBM's positions, strategies or opinions.
Visit us at https://ibm.com/smarttalks
See omnystudio.com/listener for privacy information.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:00):
Hey everyone, it's Robert and Joe here. Today we've got
something a little different to share with you. It's a
new season of the Smart Talks with IBM podcast series.
Speaker 2 (00:09):
This season, on smart Talks, Malcolm Gladwell and team are
diving into the transformative world of artificial intelligence with a
fresh perspective on the concept of open What does open
really mean in the context of AI. It can mean
open source code or open data, but it also encompasses
fostering an ecosystem of ideas, ensuring diverse perspectives are heard,
(00:31):
and enabling new levels of transparency.
Speaker 1 (00:33):
Join hosts from your favorite pushkin podcasts as they explore
how opennes in AI is reshaping industries, driving innovation, and
redefining what's possible. You'll hear from industry experts and leaders
about the implication and possibilities of open AI, and of course,
Malcolm Gladwell will be there to guide you through the
season with his unique insights.
Speaker 2 (00:52):
Look out for new episodes of Smart Talks every other
week on the iHeartRadio app, Apple Podcasts, or wherever you
get your podcasts, and learn more at IBM dot com
slash smart Talks.
Speaker 3 (01:05):
Hello, Hello, Welcome to smart Talks with IBM, a podcast
from Pushkin Industries, iHeartRadio and IBM. I'm Malcolm Godwell. This season,
we're diving back into the world of artificial intelligence, but
with a focus on the powerful concept of open its possibilities, implications,
and misconceptions. On today's episode, our season finale, I'm joined
(01:28):
by Rick Lewis, the senior vice president of Infrastructure at IBM.
Rick has had a remarkable career focused around product innovation.
He was actually a few years into retirement when IBM
came calling with an opportunity he just couldn't turn down. Thankfully,
Rick came out of retirement and today he oversees a
(01:49):
vast portfolio from storage and software to global customer support operations,
and he's engaged in one of the key problems facing
companies today, an explosion of data. In talking with Rick,
I can see that this problem of having so much
data is also an incredible opportunity because if you're able
to leverage that data to get the most value out
(02:12):
of it, then you can use it to help bring
your business into the future. We talked about the serious
computing power needed to scale AI, as well as the
ways that infrastructure storage solutions can be essential to enabling
this new world of possibilities. It's a really great conversation,
so let's get to it. I'm here with Rick Lewis. Rick, Welcome,
(02:38):
Thank Here. We are in the IBM's New York City
headquarters at one Madison Avenue. I'm going to start with
you're a hardware guy.
Speaker 4 (02:47):
I'm a hardware guy. I grew up doing hardware chip engineering.
But like I tell a lot of people, a chip
engineering project is actually a giant software project with a
piece of hardware at the end of the project. I
think if you have that analytical brain, you like to
solve problems, you'd like to get things working. You can
do that in soetwork.
Speaker 3 (03:08):
But as being someone coming from a hardware background mean
that you think about problems in a different way.
Speaker 4 (03:14):
I think one thing that you do from a hardware background,
and especially a chip background, is a chip spin and
costs millions of dollars, So you're a lot more likely
to make sure everything has a great chance of being
perfect from the get go. Or if you start kind
of from a software background, your general mindset is I
don't know, try this, see if it works. I don't know,
(03:35):
try that is if it work and you kind of
it it iterate chip people are a little more uptight
about okay, if this first round of the chip breaks
costs us from building another new round of the chip.
Speaker 3 (03:47):
Yeah, so you're a little more You guys are spend
more time.
Speaker 4 (03:52):
Planning and planning verifying, tons of time verifying.
Speaker 3 (03:57):
So you began your career as you look back, yes, correct,
and you were there for how many years?
Speaker 4 (04:02):
I was there for thirty two years?
Speaker 3 (04:04):
Yes, And your last job there was I was leading.
Speaker 4 (04:07):
The software defining cloud business. I had grown up a
hardware guy. I had done all kinds of hardware projects,
big complicated Unix servers and things like that, and then came,
you know, grew out of R and D and more
into the business realm. And then I'm much an innovator
at heart. I really like innovating new concepts things like that.
(04:32):
And what I learned is I enjoyed innovating business models
and software projects as much as I did hardware products
and projects, and so getting teams inspired towards doing that
was really a deep fascination for me. So I ended
up doing a fantastic variety of experiences and had a
successful run and honestly retired, intending to retire and do
(04:54):
some of my outside activities and things like that.
Speaker 3 (04:56):
And then how long did you stay retired before IBM
can close?
Speaker 4 (05:00):
Almost two years? And when I first got at ALL,
I thought, no, I'm having too much fun. But I
would say three things really got me thinking hard about it. One,
the industry that we're in, the IT industry. I think
it's the golden age. And what I mean by that
is for twenty years of that career, it is kind
(05:22):
of in the back office. Hey, make sure that stuff
doesn't crash, and can you please reduce the cost as
much as possible, because it's not that important to the
main business. It's just a back office function. You can
see it right now. It is at the forefront of
all business revolution. It happened with the Internet. It happened
again with cloud and how that changed every ounce of business,
(05:44):
not just IT business, but all business. And I think
it's happening again with AI. So to be in that
career that long and to miss the kind of this
age where it's like this is front and center. This
changes everything about all businesses, not just technology businesses. I
was kind of feeling like, gosh, you you trained to
be in these really awesome environments. Why wouldn't you do
(06:08):
that for a little while longer while you still can
do it. That combined with IBM and IBM seeing the
talent pool, the brilliant people at IBM, I worked with
a ton of brilliant people before I saw a chance
to work with even a larger staff of brilliant people.
And then the assets that IBM had, which is, you know,
they'd already been doing a lot of experimentation in AI,
(06:30):
they're working in quantum, the deep, rich heritage of successful projects.
I thought, who wouldn't want to kind of see if
they could be part of that next great wave of IBM.
And so I kind of decided, all right, I'm going
to put the outside interest on hold for a while
and get back in the game.
Speaker 3 (06:46):
Along between the phone call, the first phone call and
you say, yes.
Speaker 4 (06:50):
It was a while, It was probably six months. Arvin's
our CEO, teases me about that a lot. Yeah, he
was like, I.
Speaker 3 (06:58):
Don't think six months is that long? It took a while.
You're a retirement I know. Yeah.
Speaker 4 (07:02):
It's one thing to compare I'm working here and doing
this stuff versus working there. It's really hard to compare.
I'm doing exactly as I want to do every single
day when I wake up, and now I'm not going
to get to do that again. It took a while
for me to get over and I thought, I can't
miss this wave, and I'm really really happy that I did,
(07:22):
because we're doing some amazing, fun things and I'm getting
challenged in ways that I never did, so it's really fun.
Speaker 3 (07:29):
Talk a little bit about your job here at IBM.
You oversee a kind of massive portfolio.
Speaker 4 (07:36):
It's a big group, so I run the Infrastructure organization.
There's three main groups of products at IBM. There's the
Infrastructure group, which I run, the Software group, and the
Consulting group. And infrastructure is built up of mainframes, which
is called our Z portfolio, our servers which is our
power portfolio storage. By the way, those business include the
(08:00):
supply chain to build all of that stuff, so that's
in the group. Then I have the worldwide Customer Support
Organization it's called TLS Technology life Cycle Services, which is
a network of about thirteen thousand people around the globe
that make sure that everything runs and works when you
buy IBM products. And then also our IBM Cloud, which
is how we host applications and deliver as a service
(08:24):
products for our client base. So there's a lot. I
think it's about forty five thousand people total. Do those.
Speaker 3 (08:32):
Components of the Infrastructure group are they aligned in their
trajectory or are they on different paths?
Speaker 4 (08:38):
And I'm just curious what son little of both. It's
interesting you would ask that because I think of all
of the challenges coming to the new company, there were
things I expected, things that they didn't expect. But getting
that culture right in that group has been a big challenge.
IBM has a great culture toward quality products, toward emphasizing
(09:04):
passion for the client and making sure that the client
is happy, and for delivering innovation on a scale that
you know, for more than one hundred years has been
extremely powerful. But with success comes some challenges. And with
that success you can tend to get a little bit insular,
like you don't keep an eye on the competition. As well,
(09:25):
you can get more siloed, where you know, this is
my business unit, this is my business unit, I compete
with the other business unit. That's not a good thing
when you're a company, and you can get really risk averse,
meaning you feel like Hey, this is a successful business.
I don't want to do anything to mess it up,
so I don't need to try new things. Well, that's
exactly the recipe to kind of be shrinking, and infrastructure
(09:48):
had been shrinking for a little while, and so a
lot of what the challenge was for me was to
invigorate that risk taking and get to a growth mindset
where you're trying new things and seeing what works and
what doesn't work, and changing some of the models, like
investing a little bit less in hardware for some software
differentiation that goes into the hardware. So it's been very
(10:10):
successful so far, and it's been a good journey. It's
almost four years now.
Speaker 3 (10:14):
Give me an example of what was a really hard
problem that you've dealt with in those four years.
Speaker 4 (10:19):
So, boy, a really hard problem.
Speaker 3 (10:22):
An interesting and are you interesting is a better word
than art.
Speaker 4 (10:25):
One of the first things that I kind of chewed
on a little bit is I talked about how we
have Z power and storage. The Z and power product
lines are well known in the industry. Is is really
fit for purpose computing that have strengths that you know
Z runs you know most of the world's economic backbone,
and if you use a credit card. Ninety percent of
(10:47):
credit card transactions for the globe go through these Z
mainframes there in every bank there. You know, it's a
big business. It's well known in the industry. Same with power,
very tuned and optimized for smaller operations than our giant
Z mainframes, but really mission critical workloads for retail, for insurance,
(11:08):
for banking, for all of that. Our storage business not
so well known. In fact, when I came I thought,
did they have storage? Well, I have storage when I
come in too. I So I got online and I thought,
it's still hard for me to tell did they have
storage or not? Now I own a storage business. So
one of the things was not just to get the
market perception up, but to invest in that business. Because
(11:29):
if you look at infrastructure overall around the globe, it's
growing at five percent a year. The infrastructure business had
been kind of flat to declining, and so a challenge
was how do we grab onto the growth. Well, one
of the biggest growth areas due to the explosion of
data in the world is storage, So what do you
do to kind of get on that growth rate. So
we did a lot of reinvigoration of the innovation in
(11:51):
that a lot of software value, add a lot of
doubling down on the things that are working. Portfolio rationalization
where you segment the market and you say, okay, we're
going to do less of this and really go big
in these areas. And that's been probably the most dramatic
turnaround inside the group. Is our storage thing. When you
say it's a hard problem, it's not just oh, you know,
how do we do the math? No, it's cultural. It's strategy,
(12:15):
and how do you get the strategy set. It's segmentation,
it's product strategy at a granular level across a bunch
of dimensions, and then putting the investment behind it. It's
a big challenge. It takes a long time, but it's working,
so we're happy with Yeah.
Speaker 3 (12:28):
Tell me give me a little bit of perspective on
you've been there four years. Imagine we're having this conversation
four years ago.
Speaker 4 (12:36):
Yeah, what sorts.
Speaker 3 (12:38):
Of things have happened over the last four years that
have surprised you that you didn't see come? At least
we had exactly the same conversation four years ago.
Speaker 4 (12:48):
No, because I didn't know what was in I'll tell
you some of the biggest surprises I thought from the outside,
and you know, you hear from a lot of customers,
especially ten years ago, we're all going to we're all
so I thought, well, I wonder if the mainframe business
is struggling. When I get inside of there, I found
the opposite to be true. The mainframe business is actually
(13:08):
flourishing because transaction demand across the globe has done nothing
but grow. And even more surprising was the level of
innovation that the team was already doing in mainframes before
I got here was astounding. For example, we have AI.
They were building AI technology into the mainframe processors three
(13:29):
years before chat GPT made everybody talk about it in
the industry, So that was really pleasantly surprising. So that
was wonderful. Other surprises I knew about the kind of
the IP of IBM and the mystique in that, and
I used to joke with people, especially on the outside,
I said, I can't wait to get in there and
see what's in the big blue toolbox? Right, what are
(13:51):
all the things they have going on? I way underestimated
the size of the big blue toolbox and what was
in their meaning amount of really hardcore research that we're
still doing into how to build chips and how to
get to things beyond two nanimeter and that kind of
capability packaging industry leading packaging technologies, and that's in my
(14:15):
hardware kind of patch quantum. The next thing that will
come after we're done talking about AI. You know, all
of those things were surprising, But it wasn't just that.
It was then the software innovations that are going on
heavy investment in AI technologies before it was really popular
to be talking about that. But as I saw that,
(14:35):
I thought this is going to get really fun. Because
I had a good feel for where the industry was going.
I just didn't and I knew, man, I know that
talent is really good and there's brilliant people there, but
I didn't know the level of IP frankly that IBM
had at its disposal. And now you're seeing that in
things like Watson X and things like AI in mainframes,
(14:57):
et cetera.
Speaker 3 (14:58):
Building on that. Since you put AI, can you walk
me through what has to happen from your perspective, from
the infrastructure perspective to make the AI explosion work? Yeah,
so everyone wants to do more of this stuff. Yes,
clearly there has to be some underpinning of it.
Speaker 4 (15:16):
Yeah, I would tell you, you know, I think that
people feel like where we're at right now in the
AI journey had to do with one specific piece of software.
I think the inflection point for that whole thing really
at its root was around hardware, meaning the algorithms needed
to do larger language models. And all of that had
(15:37):
been around, they'd been talked about in the industry, but
at some point you hit a tipping point of hardware
capability where it's like, oh, now we can do this
in a broof force way, massive amounts of matrix math
to get weights correct so that you can do you know,
the right level of predictions that enable large language models.
And once we got to that horsepower, and that's why
(15:58):
you hear about giant g pus that are driving this
and the sales of those, et cetera. It's because we
just barely got over the hump where you can do
these big hard things in terms of hardware capability to
do it.
Speaker 3 (16:11):
Give me a layman, give me a sense of when
you say there was a kind of threshold where suddenly
these things became possible.
Speaker 4 (16:17):
Yeah, I don't know if there's an exact number, but
and more basic question that I get from a lot
of people, you know, my friends and family outside. Is
why GPUs. What does a GPU, a graphics processor have
to do with AI. It's not, Well, graphics processors are
really good at this thing matrix math, because they're figuring
(16:39):
out how do I map a pixel? And as I
move an object across the screen, it's essentially matrix math
to figure out, Okay, what does what does this pixel
on a screen look like? And what it's doing? And
as you you know, we've gotten more high resolution graphics,
more high resolution monitors, et cetera. It's a lot more
pixels and a lot more math and a lot more
(17:01):
matrix math about how you compute that. The first big
thing that kind of started to look like that, it
turns out, was crypto and crypto mining, and so you
saw graphics companies starting to sell to crypto. The technology
got to a certain point and there were use cases
like bitcoin in that that kind of said, hey, we
need to do a lot of this matrix math to
be able to do that. So graphic chips were a
(17:23):
natural fit and that kind of sustain But meanwhile, behind
the scenes, a lot of this AI AI is about
numeric calculations having to do with weights and matrices that say,
you know, giant consolidated things that predict what's going to
kind of happen based on what other things have happened,
just like predicting where pixel goes. But it's really about
(17:45):
being able to do enough data in jest to be
able to do and then the calculations to be able
to simplify things like entire sets of language or giant
chunks of the Internet, to get enough weightings in there
to be able to say, okay, we can predict what
you would say in this language based on all of
the volumes of stuff that we've seen that when you
(18:07):
start talking like this, the next word is likely, oh
it's this. Yeah.
Speaker 3 (18:11):
So, But my point is to get to that point,
that's threshold. We got there because was it because we
simply threw a lot more resources at the problem or
is it because the underlying technology got suddenly or gradually
so much more efficient.
Speaker 4 (18:25):
It's always yes and yes. But you know, the industry
for a lot of years would talk about Moore's law.
Speaker 3 (18:31):
Well, quick, will you define for us More's law for
those of those who's forgotten it.
Speaker 4 (18:37):
Yeah, So Gordon Moore at Intel coined this thing. It
was basically that the horsepower I'm going to translate it
roughly of technology will double every couple of years. We're
still on Moore's law. Moore's law changed a little bit.
For a while, it was always about frequency. Things would
go faster, faster, faster. That kind of petered out. But
(19:00):
what happened is, rather than faster, faster, faster, we did
more and more and more. So rather than one operating
unit going a lot faster on its throughput, you put
ten operating units on a chip, now you put one
hundred operating units on a chip, now a thousand. Some
of these problems, the matrix math problems scale parallel extremely well.
(19:21):
You don't have to do something really fast, you just
have to do a lot of the similar things in
parallel at the same time. So again that kind of
that extension of Moore's law, more and more hardware on
a chip to be able to do more and more
of those calculations in parallel and come up with it.
Speaker 3 (19:36):
And we said, yeah, was that threshold predictable? In other words,
see people in the industry, like you sit down X
number of years ago and say, when we get here,
AI is going to become much more of a It's funny.
Speaker 4 (19:48):
The horsepower that very predictable, the use cases not always
so easy to kind of figure out. That's where the
human spirit kind of gets involved. I think for some
people that say, oh, I saw that coming, but people
have been predicting kind of the rise of AI for
twenty five years. Oh well, then when we get to
(20:09):
this next gener oh when we get here, it kind
of hadn't happened. There's always a magic point where you
kind of get to where the technology and the use
case and somebody does something to kind of make it
catch on. And I think we're at one of those
moments in AI for sure right now. And I don't
think it's you know, people that have said, oh, this
is just the latest wave of you know, I hear
(20:30):
I've heard this about a lot of technologies, but AI
is the technology the future, and it always will be.
I used to hear that. You're not hearing that now, right,
It's like, no, it's primetime. It will change everything, just
like some of these other things changed everything.
Speaker 3 (20:45):
I noticed it if personally when I speak somewhere or
I'm listening an audience somewhere. Over the last let's say
twelve months, there's always a whole bunch of AI questions. Yes,
And if I go back to years ago, there were
no AI questions.
Speaker 4 (21:01):
Yes.
Speaker 3 (21:02):
Now my question is, so there's been this explosion on
the in popular fascination with what's going on AI. It
seems like the last year. I agree with you in
your world, when did the explosion of conversation around this start?
Speaker 4 (21:19):
It's I love this question because IBM had a fairly
big effort and business called Watson before Watson X. And
this is going back kind of ten years. I'll give
you another kind of example. I knew about a lot
(21:40):
of tablet technology before there was an iPad, a lot.
For ten years, there were a lot, but it kind
of takes a magic combination of the technology, the user experienced,
the software, and the need and the market ready for
it to kind of go. Now it's the thing. Now
we all have either an iPad or we have the
Google equivalent Tom and so I think this is a
(22:00):
little like that, meaning IBM was on the right track
with Watson. Some of the hardware wasn't there, the use
cases weren't exactly figured out. Some of the early use
cases didn't pan out. Perfectly. But the good news about
that is it's back to that culture of risk taking.
You don't look back on that and say, oh, we
shouldn't have done that, that was a bad idea. I
know you look back on that and say, what did
(22:21):
we learn? How should we try something new? How would
we pivot this time? That's what we've done with Watson
X and now that's a growing, healthy piece of our
business and very important our strategic sure, so we're all in.
Speaker 3 (22:33):
I've always investigated by the gap between insider sense of
what is happening in an outsider sense, like.
Speaker 4 (22:42):
It absolutely is that in this case, we've all been
talking about and thinking about AI and is it time
for that and what does this mean? Et cetera. And
yet none of us really predicted that actual moment, which
is kind of you know, early twenty twenty two where
it was like, oh, now you have a simple human
(23:02):
interface of software innovation combined with large language models. There's
a moment there where you're like, oh, Unlike, you know,
I think all of us are frustrated if we ask
our phone, hey, tell me about this, and it says
I found this on the web page. That does you
no good? But you know, all of a sudden, with chat,
GPT and some of these other things, you could ask
(23:24):
a question, it would give you a clear answer. Sometimes
is wrong, but at least it was like I'm getting
an answer rather than hey, I don't know if there's
some references. Good luck to you. And that's really changing.
Speaker 3 (23:34):
Talk about the kind of macro trends that are going
to shape your infrastructure battle.
Speaker 4 (23:42):
Yeah, we've talked about if you already, but I'm actually
going to go a little different direction. So macro trans first,
and this one has been before even even this AI conversation,
that we've had explosion of data. As humans, we don't
think exponentially very well. We really struggle with exponential thinking.
(24:04):
We think linearly, Oh, there'll be more, there'll be more,
they'll be more, but we don't think well when it's
like no, they'll be more, and they'll be ten times more,
and then there'll be ten times that more. That's what's
going on with data right now in our industry. It's
one of the reasons that that storage business is doing
so well is they're just more and more and more data.
You know, you'd say, well, how can there be more data?
It's just life and that thing. The things that we
(24:27):
care about, video capture, video images, you know, the the
you I don't know from my parents, you needed a
drawer with all your family photos. Now we need gigabytes
and gigabytes.
Speaker 3 (24:38):
You knew how many pictures my wife has taken off
our children, you would exactly exactly, So that's your case.
Speaker 4 (24:44):
Now. Think of companies who used to just think about
their transaction data. What's the ledger say that now have
video assets of all of their campaigns and their marketing.
They're trying to figure out, you know, what campaigns are
working the best. So it's just an explosion of data
and that's not going to stop. Dealing with that, and
more importantly, getting value from that data is a massive
(25:08):
trend in the industry. Second trend AI, and this is
the AI. Not like we were just talking about about
how it changes how I search for things or how
I learn about things. But I would argue, dealing with
that data, how do I figure out what's in all
those video streams? How do I figure out Okay, I
want all of the chunks of my corporate video that
(25:30):
have to do with client buying some specific product or something.
That's a different problem. It's not just okay, we'll look
it up in a spreadsheet and here's the math associated
with that. That is a huge trend in the industry.
You're seeing it play out in this regard, it's a
little different bent on AI. Fraud detection is the one
that we cite in our mainframes. It's a similar problem
(25:53):
where it was kind of a traditional AI problem. Look
up a rule. You know, if somebody does two small
trend actions than a massive one, it might be fraud,
right because they were seeing whether it were now to
detect fraud, you might be saying, okay to transactions then
a huge one. Plus does this entity have a real address? Second,
(26:14):
is there any web traffic on you know, better Business
Bureau kind of things that says this is a bad
business that can help you with fraud. So it's a
lot more of a it's an EXPERTI problem. It's a
holistic problem that it takes a lot more than just
you know, little chunks of rules, et cetera. And then
the third one you know after AI, is the nature
of hybrid it or hybrid computing. For a while ten
(26:38):
years ago, when cloud was on the rise, I think
the notion of hybrid computing basically having to do with
things in the cloud versus things that people still have
on the premises inside of business. It was almost a
religious argument. Now it's no, it's the reality. And the
reason is because that data that I talked about is
(26:59):
the life blood of these companies, particularly IBM's companies are
clients that usually that data has to be secure, they
have to be able to get value from it. It
is the lifeblood of the company. If you go to
an ATM and you can't get your money out to
our financial transactions, if that lasts a day, you're probably
going to change banks immediately. So it's like life or
(27:21):
death for these companies. So having that hybrid infrastructure so
that they can still hold their data, you still interact
with clouds and still get value from it from AI.
That's kind of the magic where we play, and it's
a huge business opportunity. It is a true inflection point
(27:41):
for the industry. I'm going to go back.
Speaker 3 (27:45):
I interrupted you when you were in the middle of
a rellion. We were talking about what has to happen
for AI to scale from the infrastructure standpoint. You gave
one example that I got you off on a tangent.
Can you go back and talk very so practically, like, so,
I'm you know, I'm a big company. I have all
these dreams of AI, of how I'm going to use
(28:08):
this dratically, So give me a very granular sense of
the works you have to do, yeah, to make that
dream possible.
Speaker 4 (28:15):
So let me first say what the company has to do,
and then maybe I'll say, then how do I help them?
If that makes sense? So if I'm a company and
I want to do that, So it turns out I
am a company meaning I want to use AI in
my processes. I mentioned that I have a global network
of thirteen thousand employees that support our infrastructure around the world.
(28:37):
That challenge is a great challenge for AI. That means
I have data for every customer situation for thirteen thousand
employees globally around the world on what was their problem,
how did we fix it, what next steps did they
have to do, how did they remediate that? That data
is extremely valuable to me because if I can get
(28:59):
better at doing that than anybody else in the world,
that brings my cost down. I sell more products, I
sell more service, I sell more anything. So what I
have to do to get there is I have to
figure out. Okay, what's my objective? I have a couple objectives. One,
I want customers to be able to support themselves without
even calling me, first off, and I don't want when
(29:19):
they call for the first answer to come back to
be did you try rebooting? Because I think that irritates
every single one of us. Did you try? Of course
I tried rebooting. I've had a laptop, of course I well, okay,
well then tell me, okay, what firmware version, all that
other stuff. Okay, we know this interaction. So that's kind
of the problem set. Do I want that to be
(29:40):
customers solving their own problems? Well, even for my support agents,
I want something in their pocket on their phone where
they say I'm seeing these symptoms and says, oh, this
happening around the globe. Here's kind of specific me. So
there's my problems. What does it mean for infrastructure on
the back end? So first I got to get all
that data together, right, all of those customer law, all
(30:02):
that customer support around the globe, et cetera. That needs
to be stored. That's a big set of data. And
some of it's not just fix and that kind of thing.
Some of it is Okay, you know what was the
firmware version? Who was the tech because it can matter.
Is this their first time fixing this problem? Is it
there one hundred and fiftieth time? What's their level? It's
a very complicated problem. Ingesting all that data takes an architecture.
(30:27):
We have a product called Scale, which is one of
our storage projects that actually makes it easy to ingest
all that data, get it organized, et cetera, and then
have a model. It's a whole different process to kind
of say did we train our model? We can train
our own models inside of IBM. We have a granite
set of models. Those models we fine tune, and then
(30:48):
we inference based on those models. So we can do
that inferencing in our cloud I have a cloud set
of infrastructure, or in my power servers. We can do
inferencing with our capabilities and say, okay, based on what
I'm saying, here's what the remediation that you should do
for that customer. We already are doing that today. We've
seen over a third of our support calls have had
(31:13):
significant reduction in the amount of time that it takes
to resolve that support call just by what I said
right there.
Speaker 3 (31:20):
That I've really been curious about this. If I had
reduced something like AI into that equation as you just did. Yeah,
and you said we've already seen a thirty percent Say
did you say thirty percent reduction?
Speaker 4 (31:33):
Thirty percent of our interactions have seen significant reduction in
those time?
Speaker 3 (31:39):
Was that your primary goal to reduce the time of
the interaction? But it was you if everything else was
the same all, but what you were doing was shrinking
the amount of time?
Speaker 4 (31:48):
That would you want one of the primary goals, So
to us in that business net promoter score kind of
the satisfaction of a client is the supreme goal. What
makes them sad fine, doesn't cost me a fortune, happens
really quickly, and if I can do it myself, I'd
be thrilled. It affects all of those right. It kind
(32:08):
of says it got resolved faster, it didn't cost me
an arm and the leg because the deck was barely here,
because it's a common problem, or I solved it myself
without even calling, So all of those objectives would kind
of hit across all so that now you see it.
So that's a little microcosm. That's just me and my
customer support business. Now think of how many problems for
businesses around the world there are like that. It's not
(32:31):
a it's not like a new AI application that changes
the entire user experience. That's those will come, But right
now it's kind of practical, which is, I just want
to do what I'm doing better and faster, and I
can get immediate economic return from those things.
Speaker 3 (32:47):
How long How long did it take you to just
stick with that example of the customer reaction reducing thirty
percent of the time? How long from the very beginning
of that project, Yeah to that thirty percent reduction it was.
Speaker 4 (33:00):
Hell long, less than a year. And yeah, So one
of the challenges, and this is interesting with a very
large organization, as you can imagine, just like you're seeing
in the industry, we don't have a problem of generating
ideas for how AI could help us. We actually have
a problem filtering the thousands of ideas from our employees
(33:24):
and from everywhere. It's like, hey, we could use AI
to and filtering down and saying, okay, which of these
will have a return on investment quickly and at a
level that sustains that's worth kind of going and investing
in the infrastructure and the software and kind of making
that happen. Is that unusual.
Speaker 3 (33:41):
If I talked to you twenty five years ago and said,
do you have a problem of too many good ideas
or too few?
Speaker 4 (33:47):
What was you said in this specific area, Probably too few,
because at some point you reach diminishing returns. So, for example,
let's use this same example. Can those thirteen thousand technicians
go faster? Can they spend less time driving to the side.
I mean, there's only so much you can kind of
(34:07):
do on those things. But if you can get them
an answer to the problem and maybe even avoid them
having to visit at all because the client helped themselves,
that's a step function. So that's why people are kind
of talking about there's a business revolution coming with AI
where there are some step function changes that can be there.
And notice I didn't say I'm going to have less
(34:29):
of those agents. That's not my objective. My objective and
I think that's the fear in the industry about AI
is going to eliminate all the jobs. No, I just
created thirteen thousand superpowered agents that can do more right.
And so I'm not just going to support IBM products.
I'm going to go out and support other people's products
because I know how to do that really well. And
once I have the data on how to fix their problems,
(34:51):
I may just have a customer support business that's independent
of my boxes. So you know, I think that's where
people sometimes get it wrong. And the AI thing is,
it's like, you know, did word processing eliminate the need
for writers? No? It enabled writing instead of mucking around
with mimeographic machines and click and click typewriters. It may
(35:13):
have enabled too much writing? Yeah, maybe maybe can I
give you a hypothetical?
Speaker 2 (35:18):
Uh?
Speaker 3 (35:18):
And I asked this because I ran I was at
some conmis and I ran into some guy from the
I R S who was really, really, really really excited
about AI. So let's suppose they call you up and
they say you're going to talk to the I R SKY.
I call you up and I say, Rick, Uh, clearly
(35:40):
there's something that we could do for the I R
S if we work together.
Speaker 4 (35:45):
Yeah, what would your answer? Of course?
Speaker 2 (35:49):
No.
Speaker 4 (35:49):
I think we sell to a lot of government agencies.
I can imagine in the business that we're in, we
enable a lot of social security transaction and things like
that through our mainframes. And I think, you know, we're
in the business of helping whatever client get the most
out of their data and be able to secure it
(36:10):
and be able to do analytics with this. And IRS
has a heck of a lot of data, so yes,
we would help them.
Speaker 3 (36:17):
Do you know how the amount of data they have
compares to some of the corporate clients you've I.
Speaker 4 (36:21):
Don't know specifically for the IRS how much data they have,
but I would assume it's a whole lot. It's mountains.
But that's our business. I mean, it's interesting sometimes people
of that what's the most you know, what is it
that IBM has that's of great value? Is it a server?
Is it a storage array? Is it you know, software
(36:42):
and all that. What we have is the most important
entities in the world have their data on our stuff.
The most important data in the world. It's not you know,
pictures of your grandkids and things like that. Generally for us,
it's all of the financial transactions that have and globally,
right it's all of the it's the world's economy is
(37:03):
kind of running through our systems, and so we take
that really seriously. You know, you would be distraught if
you lost one photo on your laptop or whatever, but
you know, if we lose a transaction, like somebody moves
a big amount of money and it's like, well, don't
know what happened there. It is a massive deal, right,
so that doesn't happen.
Speaker 3 (37:23):
But I want to go back to my irs example
for US, Yes, so one, is it reasonable to assume
that you could that somebody IBM or somebody else could
in a short period of time put together not just
the AI capability to audit returns, but also this the
infrastructure support for that in a reasonable amount of time
(37:45):
for a reasonable amount of cost. Or is it over?
Is it going to the moon? Or is it it?
Speaker 4 (37:51):
Definitely? I mean so we're already doing that kind of
thing right across a network of banks and others. Yeah,
essentially all credit card transactions for all of the world
to go through our systems, So that in some ways
is more volume than the datch returns of the US people.
(38:11):
And they're W two's and all that stuff, and we
do that stuff too. I try not to describe it
too much in detail, but we definitely do a lot
of that. In fact, I think most of if you think, okay,
what is super critical data, who would be doing the
business transaction processing it is most likely us in almost
(38:33):
all cases, whether it's government things or private or banks
or that kind of thing. That's what we do.
Speaker 3 (38:40):
Rick we're going to end with the where we always
end with a couple of quick fire questions. Okay, here
we go. What single piece of advice would you give
to businesses trying to use AI in an effective way?
Speaker 4 (38:52):
The simple version is get started. By get started, I
mean think of what is something that I want to improve.
The things that we have traction on right now in
the market are around business process, automation, digital labor, those
kind of things. But my other little piece of advice
there is keep it simple to begin with. You're going
(39:13):
to learn a lot, but getting started means you'll start
that learning curve. I even advise you My friends like, hey,
should I be playing around with some of this AI stuff?
And I say yeah, because I think it will help
you start to be more comfortable and you may find
a use case personally for that. I think the same
is true for businesses. The first step in that journey
is always with what data. Notice when I talked about
(39:36):
our customer support people, I thought about, Okay, what's the data.
The data is all of those logs of all of
those service engagements around the world, and what could I
do with that? Well, I could use that to get
to a knowledge base that really helps and hopefully that
I can do it in multiple languages because it's global
and I can you know, all of those things. That
(39:57):
was kind of my data sent That one's not super simple,
but we've had a lot of experience in AI for
other people that might just be how do I automate
filling out travel expense reports for my company? We can
help people that we have consulting, we have wats and
X tools. We can do that like this, and we're
doing it globally for people around the world. Pick that thing.
(40:18):
What's the data you have? In that case, it's data
of expense reports and it's like, okay, we can help
you automate that for people where they could do it
just by you know, a verbal interface. What did you spend,
where did you go? Who you were you with? Okay,
we filled out your travel expense report for you and
you don't have to mess around with it.
Speaker 3 (40:36):
So we were playing with this idea where we would
pick a business and go in there and do it
would be AI makeover. Yeah, I love that what's okay,
what is the ideal business to do? We only have
a couple months. We don't want to spend a kajillion dollars.
We want to be able to show tangibly and quickly
what AI can do. What's an ideal business to do
(40:57):
that in It can be a small business. We're not
talking this grand corporate thing there.
Speaker 4 (41:02):
Ah boy, small business that we could do and hey,
I make over. Customer support is one of my favorites
because it's a it's it's I have it on the
business side where I provide customer support. I have it
on the consumer side, where it drives me nuts when
I have to go through thirty layers of phone menus.
(41:23):
Speak to an agent, speak to an agent, speak to
an agent. That for any business, I think is just
ripe to be able to kind of say why do
I have to click through these manucent messages. I just
need to tell you in human language, here's the issue,
and I'll be really good about telling you details about
You know, I tried to set up this thing for
my bank and I do da da da da da.
(41:44):
They can go through all the menus automate that process.
I think it would change everything because all that frustration
as a consumer would go down dramatically, and it's all,
you know, why are you making me the beep booth
press one? Exactly? Well, don't offload to me, offload to AI.
We can help you with that.
Speaker 3 (42:05):
Here's my version of that drives me crazy. Every morning
I go to the same coffee shop and I get
a cup of tea and a croissant.
Speaker 4 (42:15):
And here's what happens.
Speaker 3 (42:16):
A person has their screen and they go, I go,
cup of tea, croissant, sparkling water, like at least twenty keystrokes,
and then like then the screen is turned around. Like
at this point we're like forty five seconds in, I'm like,
why is this? First of all, it's not for me,
all those keystrokes, it's their internal right, right, So they're
(42:39):
burdening me in order to service.
Speaker 4 (42:40):
To back it, you should be able to walk in,
go up and they go, I'm olc them the same
thing and you just go yes, and then.
Speaker 3 (42:46):
The boom, We're done. Can we do AI makeover of
my coffee shop?
Speaker 4 (42:51):
You notice I quickly jumped more to banks than your
coffee shop because I think I'm a business person, but
I'm not trying to kind of do a deal on
one coffee shop.
Speaker 3 (43:01):
But this is interesting because it takes me back to
something you said that I thought was really important. When
you were talking about when you were using AI and
your customer service thing, it was clear that your goal
you could have any number of goals, yes, going in.
It could be to cut costs, it could be to
dramatically improved profits.
Speaker 4 (43:21):
Your goal, quite.
Speaker 3 (43:21):
Specifically, was to improve the experience of your customer, right,
So you were.
Speaker 4 (43:25):
Using it to that. All the other things come from
that come from. That is actually one of the beautiful
pillars of the IBM culture is delighting clients is actually
where all of the good stuff comes from.
Speaker 3 (43:37):
So my coffee shop thing is the same principle. Right now,
they're making my customer experience worse and they don't want to,
but their eyes are glued to the special a moment
when I walk in and I want to say, Hi,
how are you doing? We could have a conversation. You're
too busy, busy pooping, So like, this is the same thing.
If they had it that, oh, this isn't if they
(43:58):
understood they had an operation need to improve the experience
of their customer experience.
Speaker 4 (44:02):
I would not be surprised if a chain comes along
where that is their value proposition. I would not be
surprised at all. Yeah, yeah, right, So I mean and
and when those things kind of catch hold, it becomes
a revolution.
Speaker 3 (44:18):
You know, when the guy comes to do like to
redo your roof and they put a sign out front,
like you know, Joe's roofing. You guys could do the
same with my coffee shop. But like I'd be i'ure
was here exactly exactly.
Speaker 4 (44:35):
In five years, the main frame will be dot dot
dot going strong, the mainframe going strong and with new capabilities,
continuous new capabilities. I think when we announced the last
version Z sixteen, the latest version, I should say, and
(44:55):
we said, hey, there's AI processing built into it. This
was before everybody was talking about that. I think a
lot of people thought, what's that for? And we did
it specifically for traditional AI fraud detection, et cetera. This
next version, not only do we have the traditional AI
built in, but we have optional cards that you can
plug into it to allow you to do large language
(45:16):
models for the enhanced fraud detection cases that we talked about,
where you know, it's more than just what transactions were happening.
So if you take that and say, okay, the next generations,
we have more transaction volume than we've ever had in mainframes. Today,
the business is growing, it's strong, we keep innovating. In
(45:38):
five years it'll be going strong.
Speaker 3 (45:39):
But we're people. You're saying this in the context of
for years people were predicting, weren't they that the main
brand was going to go away.
Speaker 4 (45:48):
There were pundits in the market that said everything will
go away there, no one will ever have a box,
It'll all be online. I think this is something I've
learned big time in my long career. You know in
the IT industry is don't believe everything you hear. So
I went back for my master's degree at Stanford after
(46:08):
I had worked a while in as a hardware designer,
and everybody told me be sure to do your masters
in software. Hardware is dead. I went on to work
for thirty plus years in hardware and infrastructure. Now software
became important, and I'm glad I had that extra training
in software because it helped me in hardware. But hardware
wasn't dead. Then I heard all infrastructure will go into
(46:32):
the cloud. There won't be that hasn't happened, it's not happening.
Then I heard there will only be one cloud because
one of the players will dominate. There's not one cloud.
So I think it's as humans we like to oversimplify
and go, oh, it's all going to be this, And
kind of what I've learned is fit for purpose matters
in everything. It matters in size of infrastructure, it matters
(46:56):
in the stack that goes along with solving a specific
use case. If you're willing to design something that's the
best at that use case, if you're willing to design
the coffee shop that is the best at greeting me,
there's a spot for you, and there may be a
big business in doing that. So oversimplifying is really.
Speaker 3 (47:13):
When you heard all those predictions, did you believe them
at the time.
Speaker 4 (47:19):
They looked like they were trending in that direction. I'll
tell you some right now which might be useful. There
will only be one GPU company and they're going to
end up taking over the world. It's a pretty obvious answer.
Whose economic values risen dramatically. I don't think that's going
to be the case. In fact, I think that ninety
percent of processing for AI actually happen happens at inferencing,
(47:42):
and inferencing is not as GPU and hardware intensive as
the other things and is a lot more amenable to
fit for purpose. So the model size will matter. The
tuning matters a lot. As we're learning. We have a
product around instruct lab that's really focused on tuning. So
that was one thing is there'll be one GPU. The
other thing is that the biggest model will win. I
(48:04):
think is another thing that's kind of people are saying
right now. Don't believe that I believe they will be
fit for purpose models. It takes a lot of money
to run to create a huge model, and then to
run a huge model, or to even infer off of
a huge model. I don't need a massive training GPU
set thing to solve my thirteen thousand people customer support issues.
(48:26):
So why would I feel like I got to go
farm that out for a big expensive thing. I can
do that on a small box. In some cases I
might even be able to do that on a laptop.
The other thing I'll say in this we are so
early innings in AI, A lot of things are going
to change. So anybody kind of saying it will all
be X, Y or Z, I just think you have
no idea how this is going to play out, and
(48:47):
it's up to us to go figure out how it
plays out.
Speaker 3 (48:50):
Yeah, yeah, all right, in five years AI will be
dot dot dot still new.
Speaker 4 (48:59):
It will have moved a bunch in five years, but
the potential for the disruption in the world will still
will still be very early innings in that process. And
I think that's super important to realize. That's why I
say get started, start thinking about how that could change,
because it'll be some little things first, but it will
(49:19):
continue to snowball.
Speaker 3 (49:20):
This is a common observation that we the invention of
the capability massively predates the understanding.
Speaker 4 (49:30):
Of the capability, right, Like I love that.
Speaker 3 (49:33):
Yeah, Like yes, recorded recording shows on television is invented
in the sixties. Probably we don't really understand what it's
used for until the oughts was what's really good for
is being able to tell a story sequentially, Yes, over time,
(49:54):
because you know that the person will always see in
the episode before, so you get the Sopranos. And yes, yes,
Hollywood wanted to ban the VCR in the beginning, Yeah,
because they thought it was good. They thought the point
of it was thought then understand. No, No, No, it's storytelling.
It's actually your business is getting better. Yes, took them
twenty years to figure that out, which is to your point,
(50:14):
why would we know what AI was four and five years.
Speaker 4 (50:16):
Well, that's why you hear people kind of say, oh
my gosh, AI's that will just eliminate jobs. No, it'll
make jobs better. That's how I view it.
Speaker 3 (50:23):
Yeah, what's the number one thing that people misunderstand about AI?
Speaker 4 (50:27):
Is that it that it'll I think that's that. That
would be the human kind of understanding part of it.
The technology part of it, I think would be what
I was talking about fit for purpose, meaning that it
isn't just going to be a GPU arms race all
of AI. I don't believe that at all. It will
change everything, but it's not just going to be a GPU.
Speaker 3 (50:48):
Armed Next question, what advice would you give yourself ten
years ago to better prepare you for today?
Speaker 4 (50:54):
I'm changing this question, okay.
Speaker 3 (50:58):
I would say, let's imagine that at what was your
what college.
Speaker 4 (51:02):
You to go to? I went to three of them.
My undergrad was Utah State University, my NBA was Santa
Clara University, and my master's in w was Stanford.
Speaker 3 (51:12):
Okay, any one of those three culture up and says
we want you to give the commencement address and imagine
it's it's it's let's just say, for the sake of argument,
it's just to the stem people.
Speaker 4 (51:26):
Those are the relevant parties here. What do you tell them? Boy?
What do I tell them? Let's see. I think I
would start with life is a marathon, not a sprint.
It would be the first one. The second thing I
would say that in that spirit is be sure to
(51:49):
set yourself some big, hairy, audacious goals and don't be
overly disappointed if you don't hit them all. Going after
those big, hairy, audacious goals will get you on a
path where you will learn so much. You will achieve
more than you ever could imagine you would have achieved.
(52:09):
That's what the advice I give to my kids is, ye,
set some big goals, get after it. You may or
may not achieve them, but you'll be better for the
whole process when you're done.
Speaker 3 (52:17):
By the way, as someone whose kids are younger than yours,
is it actually useful to give your give advice to
your kids?
Speaker 4 (52:24):
The pointless exercise TVD. We're still on the journey, and
I think we will be for a long time. I
don't know how are you already using AI in your
day to day life today? Personally, I would say it's
replacing a good chunk of my search. You know, I'm
less likely to go blindly stumbling through a bunch of
(52:45):
web pages looking for stuff. I'm more likely to ask
a question from a few AI engines kind of see
get me in the right direction. Then I'll go bumble
through a few things. At work, I can tell you
code development right now, we are seeing massive improvements in
code development and support. Products we have like Watson Code
(53:06):
Assistant that is really showing immediate return for a code developers,
and I think that will again be a tool that
increases productivity for code developers immediately across the globe. Yeah.
Speaker 3 (53:20):
Last question, what's the one skill that every technology leader
needs that has nothing to do with technology.
Speaker 4 (53:26):
Being able to inspire a set of people toward a
common goal and collaborate to achieve it. That's at the
core of everything everything. That's a lovely way to end.
Thank you so much, Rick, Thank you.
Speaker 3 (53:42):
This conversation left me excited. I'm now imagining the potential
for new use cases for AI in all sorts of
different businesses. Rick didn't seem soldom my idea of a
coffee shop makeover, but it's clear there's lots of opportunities
here to increase speed and efficiency, to achieve your objectives,
and to dream beyond the current applications for this technology.
(54:05):
At the end of the day, the scaling of AI
will rely on the right infrastructure to support it. With
the right tools, you can solve problems that are unique
tier industry and improve the experience for your customers. Smart
(54:30):
Talks with IBM is produced by Matt Romano, Amy Gains McQuaid,
and Jacob Goldstein. Were edited by Lydia gene Kott, mastering
by Jake Koorsky. Theme song by Gramoscope. Special thanks to
the eight Bar and IBM teams, as well as the
Pushkin marketing team. Smart Talks with IBM is a production
(54:50):
of Pushkin Industries and Ruby Studio at iHeartMedia. To find
more Pushkin podcasts, listen on the iHeartRadio app, Apple Podcasts,
or wherever you listen to podcasts. I'm Malcolm Gladwell. This
(55:11):
is a paid advertisement from IBM. The conversations on this
podcast don't necessarily represent IBM's positions, strategies, or opinions,
Hey everyone, it's Robert and Joe here. Today we've got
something a little different to share with you. It's a
new season of the Smart Talks with IBM podcast series.
Speaker 2 (00:09):
This season, on smart Talks, Malcolm Gladwell and team are
diving into the transformative world of artificial intelligence with a
fresh perspective on the concept of open What does open
really mean in the context of AI. It can mean
open source code or open data, but it also encompasses
fostering an ecosystem of ideas, ensuring diverse perspectives are heard,
(00:31):
and enabling new levels of transparency.
Speaker 1 (00:33):
Join hosts from your favorite pushkin podcasts as they explore
how opennes in AI is reshaping industries, driving innovation, and
redefining what's possible. You'll hear from industry experts and leaders
about the implication and possibilities of open AI, and of course,
Malcolm Gladwell will be there to guide you through the
season with his unique insights.
Speaker 2 (00:52):
Look out for new episodes of Smart Talks every other
week on the iHeartRadio app, Apple Podcasts, or wherever you
get your podcasts, and learn more at IBM dot com
slash smart Talks.
Speaker 3 (01:05):
Hello, Hello, Welcome to smart Talks with IBM, a podcast
from Pushkin Industries, iHeartRadio and IBM. I'm Malcolm Godwell. This season,
we're diving back into the world of artificial intelligence, but
with a focus on the powerful concept of open its possibilities, implications,
and misconceptions. On today's episode, our season finale, I'm joined
(01:28):
by Rick Lewis, the senior vice president of Infrastructure at IBM.
Rick has had a remarkable career focused around product innovation.
He was actually a few years into retirement when IBM
came calling with an opportunity he just couldn't turn down. Thankfully,
Rick came out of retirement and today he oversees a
(01:49):
vast portfolio from storage and software to global customer support operations,
and he's engaged in one of the key problems facing
companies today, an explosion of data. In talking with Rick,
I can see that this problem of having so much
data is also an incredible opportunity because if you're able
to leverage that data to get the most value out
(02:12):
of it, then you can use it to help bring
your business into the future. We talked about the serious
computing power needed to scale AI, as well as the
ways that infrastructure storage solutions can be essential to enabling
this new world of possibilities. It's a really great conversation,
so let's get to it. I'm here with Rick Lewis. Rick, Welcome,
(02:38):
Thank Here. We are in the IBM's New York City
headquarters at one Madison Avenue. I'm going to start with
you're a hardware guy.
Speaker 4 (02:47):
I'm a hardware guy. I grew up doing hardware chip engineering.
But like I tell a lot of people, a chip
engineering project is actually a giant software project with a
piece of hardware at the end of the project. I
think if you have that analytical brain, you like to
solve problems, you'd like to get things working. You can
do that in soetwork.
Speaker 3 (03:08):
But as being someone coming from a hardware background mean
that you think about problems in a different way.
Speaker 4 (03:14):
I think one thing that you do from a hardware background,
and especially a chip background, is a chip spin and
costs millions of dollars, So you're a lot more likely
to make sure everything has a great chance of being
perfect from the get go. Or if you start kind
of from a software background, your general mindset is I
don't know, try this, see if it works. I don't know,
(03:35):
try that is if it work and you kind of
it it iterate chip people are a little more uptight
about okay, if this first round of the chip breaks
costs us from building another new round of the chip.
Speaker 3 (03:47):
Yeah, so you're a little more You guys are spend
more time.
Speaker 4 (03:52):
Planning and planning verifying, tons of time verifying.
Speaker 3 (03:57):
So you began your career as you look back, yes, correct,
and you were there for how many years?
Speaker 4 (04:02):
I was there for thirty two years?
Speaker 3 (04:04):
Yes, And your last job there was I was leading.
Speaker 4 (04:07):
The software defining cloud business. I had grown up a
hardware guy. I had done all kinds of hardware projects,
big complicated Unix servers and things like that, and then came,
you know, grew out of R and D and more
into the business realm. And then I'm much an innovator
at heart. I really like innovating new concepts things like that.
(04:32):
And what I learned is I enjoyed innovating business models
and software projects as much as I did hardware products
and projects, and so getting teams inspired towards doing that
was really a deep fascination for me. So I ended
up doing a fantastic variety of experiences and had a
successful run and honestly retired, intending to retire and do
(04:54):
some of my outside activities and things like that.
Speaker 3 (04:56):
And then how long did you stay retired before IBM
can close?
Speaker 4 (05:00):
Almost two years? And when I first got at ALL,
I thought, no, I'm having too much fun. But I
would say three things really got me thinking hard about it. One,
the industry that we're in, the IT industry. I think
it's the golden age. And what I mean by that
is for twenty years of that career, it is kind
(05:22):
of in the back office. Hey, make sure that stuff
doesn't crash, and can you please reduce the cost as
much as possible, because it's not that important to the
main business. It's just a back office function. You can
see it right now. It is at the forefront of
all business revolution. It happened with the Internet. It happened
again with cloud and how that changed every ounce of business,
(05:44):
not just IT business, but all business. And I think
it's happening again with AI. So to be in that
career that long and to miss the kind of this
age where it's like this is front and center. This
changes everything about all businesses, not just technology businesses. I
was kind of feeling like, gosh, you you trained to
be in these really awesome environments. Why wouldn't you do
(06:08):
that for a little while longer while you still can
do it. That combined with IBM and IBM seeing the
talent pool, the brilliant people at IBM, I worked with
a ton of brilliant people before I saw a chance
to work with even a larger staff of brilliant people.
And then the assets that IBM had, which is, you know,
they'd already been doing a lot of experimentation in AI,
(06:30):
they're working in quantum, the deep, rich heritage of successful projects.
I thought, who wouldn't want to kind of see if
they could be part of that next great wave of IBM.
And so I kind of decided, all right, I'm going
to put the outside interest on hold for a while
and get back in the game.
Speaker 3 (06:46):
Along between the phone call, the first phone call and
you say, yes.
Speaker 4 (06:50):
It was a while, It was probably six months. Arvin's
our CEO, teases me about that a lot. Yeah, he
was like, I.
Speaker 3 (06:58):
Don't think six months is that long? It took a while.
You're a retirement I know. Yeah.
Speaker 4 (07:02):
It's one thing to compare I'm working here and doing
this stuff versus working there. It's really hard to compare.
I'm doing exactly as I want to do every single
day when I wake up, and now I'm not going
to get to do that again. It took a while
for me to get over and I thought, I can't
miss this wave, and I'm really really happy that I did,
(07:22):
because we're doing some amazing, fun things and I'm getting
challenged in ways that I never did, so it's really fun.
Speaker 3 (07:29):
Talk a little bit about your job here at IBM.
You oversee a kind of massive portfolio.
Speaker 4 (07:36):
It's a big group, so I run the Infrastructure organization.
There's three main groups of products at IBM. There's the
Infrastructure group, which I run, the Software group, and the
Consulting group. And infrastructure is built up of mainframes, which
is called our Z portfolio, our servers which is our
power portfolio storage. By the way, those business include the
(08:00):
supply chain to build all of that stuff, so that's
in the group. Then I have the worldwide Customer Support
Organization it's called TLS Technology life Cycle Services, which is
a network of about thirteen thousand people around the globe
that make sure that everything runs and works when you
buy IBM products. And then also our IBM Cloud, which
is how we host applications and deliver as a service
(08:24):
products for our client base. So there's a lot. I
think it's about forty five thousand people total. Do those.
Speaker 3 (08:32):
Components of the Infrastructure group are they aligned in their
trajectory or are they on different paths?
Speaker 4 (08:38):
And I'm just curious what son little of both. It's
interesting you would ask that because I think of all
of the challenges coming to the new company, there were
things I expected, things that they didn't expect. But getting
that culture right in that group has been a big challenge.
IBM has a great culture toward quality products, toward emphasizing
(09:04):
passion for the client and making sure that the client
is happy, and for delivering innovation on a scale that
you know, for more than one hundred years has been
extremely powerful. But with success comes some challenges. And with
that success you can tend to get a little bit insular,
like you don't keep an eye on the competition. As well,
(09:25):
you can get more siloed, where you know, this is
my business unit, this is my business unit, I compete
with the other business unit. That's not a good thing
when you're a company, and you can get really risk averse,
meaning you feel like Hey, this is a successful business.
I don't want to do anything to mess it up,
so I don't need to try new things. Well, that's
exactly the recipe to kind of be shrinking, and infrastructure
(09:48):
had been shrinking for a little while, and so a
lot of what the challenge was for me was to
invigorate that risk taking and get to a growth mindset
where you're trying new things and seeing what works and
what doesn't work, and changing some of the models, like
investing a little bit less in hardware for some software
differentiation that goes into the hardware. So it's been very
(10:10):
successful so far, and it's been a good journey. It's
almost four years now.
Speaker 3 (10:14):
Give me an example of what was a really hard
problem that you've dealt with in those four years.
Speaker 4 (10:19):
So, boy, a really hard problem.
Speaker 3 (10:22):
An interesting and are you interesting is a better word
than art.
Speaker 4 (10:25):
One of the first things that I kind of chewed
on a little bit is I talked about how we
have Z power and storage. The Z and power product
lines are well known in the industry. Is is really
fit for purpose computing that have strengths that you know
Z runs you know most of the world's economic backbone,
and if you use a credit card. Ninety percent of
(10:47):
credit card transactions for the globe go through these Z
mainframes there in every bank there. You know, it's a
big business. It's well known in the industry. Same with power,
very tuned and optimized for smaller operations than our giant
Z mainframes, but really mission critical workloads for retail, for insurance,
(11:08):
for banking, for all of that. Our storage business not
so well known. In fact, when I came I thought,
did they have storage? Well, I have storage when I
come in too. I So I got online and I thought,
it's still hard for me to tell did they have
storage or not? Now I own a storage business. So
one of the things was not just to get the
market perception up, but to invest in that business. Because
(11:29):
if you look at infrastructure overall around the globe, it's
growing at five percent a year. The infrastructure business had
been kind of flat to declining, and so a challenge
was how do we grab onto the growth. Well, one
of the biggest growth areas due to the explosion of
data in the world is storage, So what do you
do to kind of get on that growth rate. So
we did a lot of reinvigoration of the innovation in
(11:51):
that a lot of software value, add a lot of
doubling down on the things that are working. Portfolio rationalization
where you segment the market and you say, okay, we're
going to do less of this and really go big
in these areas. And that's been probably the most dramatic
turnaround inside the group. Is our storage thing. When you
say it's a hard problem, it's not just oh, you know,
how do we do the math? No, it's cultural. It's strategy,
(12:15):
and how do you get the strategy set. It's segmentation,
it's product strategy at a granular level across a bunch
of dimensions, and then putting the investment behind it. It's
a big challenge. It takes a long time, but it's working,
so we're happy with Yeah.
Speaker 3 (12:28):
Tell me give me a little bit of perspective on
you've been there four years. Imagine we're having this conversation
four years ago.
Speaker 4 (12:36):
Yeah, what sorts.
Speaker 3 (12:38):
Of things have happened over the last four years that
have surprised you that you didn't see come? At least
we had exactly the same conversation four years ago.
Speaker 4 (12:48):
No, because I didn't know what was in I'll tell
you some of the biggest surprises I thought from the outside,
and you know, you hear from a lot of customers,
especially ten years ago, we're all going to we're all
so I thought, well, I wonder if the mainframe business
is struggling. When I get inside of there, I found
the opposite to be true. The mainframe business is actually
(13:08):
flourishing because transaction demand across the globe has done nothing
but grow. And even more surprising was the level of
innovation that the team was already doing in mainframes before
I got here was astounding. For example, we have AI.
They were building AI technology into the mainframe processors three
(13:29):
years before chat GPT made everybody talk about it in
the industry, So that was really pleasantly surprising. So that
was wonderful. Other surprises I knew about the kind of
the IP of IBM and the mystique in that, and
I used to joke with people, especially on the outside,
I said, I can't wait to get in there and
see what's in the big blue toolbox? Right, what are
(13:51):
all the things they have going on? I way underestimated
the size of the big blue toolbox and what was
in their meaning amount of really hardcore research that we're
still doing into how to build chips and how to
get to things beyond two nanimeter and that kind of
capability packaging industry leading packaging technologies, and that's in my
(14:15):
hardware kind of patch quantum. The next thing that will
come after we're done talking about AI. You know, all
of those things were surprising, But it wasn't just that.
It was then the software innovations that are going on
heavy investment in AI technologies before it was really popular
to be talking about that. But as I saw that,
(14:35):
I thought this is going to get really fun. Because
I had a good feel for where the industry was going.
I just didn't and I knew, man, I know that
talent is really good and there's brilliant people there, but
I didn't know the level of IP frankly that IBM
had at its disposal. And now you're seeing that in
things like Watson X and things like AI in mainframes,
(14:57):
et cetera.
Speaker 3 (14:58):
Building on that. Since you put AI, can you walk
me through what has to happen from your perspective, from
the infrastructure perspective to make the AI explosion work? Yeah,
so everyone wants to do more of this stuff. Yes,
clearly there has to be some underpinning of it.
Speaker 4 (15:16):
Yeah, I would tell you, you know, I think that
people feel like where we're at right now in the
AI journey had to do with one specific piece of software.
I think the inflection point for that whole thing really
at its root was around hardware, meaning the algorithms needed
to do larger language models. And all of that had
(15:37):
been around, they'd been talked about in the industry, but
at some point you hit a tipping point of hardware
capability where it's like, oh, now we can do this
in a broof force way, massive amounts of matrix math
to get weights correct so that you can do you know,
the right level of predictions that enable large language models.
And once we got to that horsepower, and that's why
(15:58):
you hear about giant g pus that are driving this
and the sales of those, et cetera. It's because we
just barely got over the hump where you can do
these big hard things in terms of hardware capability to
do it.
Speaker 3 (16:11):
Give me a layman, give me a sense of when
you say there was a kind of threshold where suddenly
these things became possible.
Speaker 4 (16:17):
Yeah, I don't know if there's an exact number, but
and more basic question that I get from a lot
of people, you know, my friends and family outside. Is
why GPUs. What does a GPU, a graphics processor have
to do with AI. It's not, Well, graphics processors are
really good at this thing matrix math, because they're figuring
(16:39):
out how do I map a pixel? And as I
move an object across the screen, it's essentially matrix math
to figure out, Okay, what does what does this pixel
on a screen look like? And what it's doing? And
as you you know, we've gotten more high resolution graphics,
more high resolution monitors, et cetera. It's a lot more
pixels and a lot more math and a lot more
(17:01):
matrix math about how you compute that. The first big
thing that kind of started to look like that, it
turns out, was crypto and crypto mining, and so you
saw graphics companies starting to sell to crypto. The technology
got to a certain point and there were use cases
like bitcoin in that that kind of said, hey, we
need to do a lot of this matrix math to
be able to do that. So graphic chips were a
(17:23):
natural fit and that kind of sustain But meanwhile, behind
the scenes, a lot of this AI AI is about
numeric calculations having to do with weights and matrices that say,
you know, giant consolidated things that predict what's going to
kind of happen based on what other things have happened,
just like predicting where pixel goes. But it's really about
(17:45):
being able to do enough data in jest to be
able to do and then the calculations to be able
to simplify things like entire sets of language or giant
chunks of the Internet, to get enough weightings in there
to be able to say, okay, we can predict what
you would say in this language based on all of
the volumes of stuff that we've seen that when you
(18:07):
start talking like this, the next word is likely, oh
it's this. Yeah.
Speaker 3 (18:11):
So, But my point is to get to that point,
that's threshold. We got there because was it because we
simply threw a lot more resources at the problem or
is it because the underlying technology got suddenly or gradually
so much more efficient.
Speaker 4 (18:25):
It's always yes and yes. But you know, the industry
for a lot of years would talk about Moore's law.
Speaker 3 (18:31):
Well, quick, will you define for us More's law for
those of those who's forgotten it.
Speaker 4 (18:37):
Yeah, So Gordon Moore at Intel coined this thing. It
was basically that the horsepower I'm going to translate it
roughly of technology will double every couple of years. We're
still on Moore's law. Moore's law changed a little bit.
For a while, it was always about frequency. Things would
go faster, faster, faster. That kind of petered out. But
(19:00):
what happened is, rather than faster, faster, faster, we did
more and more and more. So rather than one operating
unit going a lot faster on its throughput, you put
ten operating units on a chip, now you put one
hundred operating units on a chip, now a thousand. Some
of these problems, the matrix math problems scale parallel extremely well.
(19:21):
You don't have to do something really fast, you just
have to do a lot of the similar things in
parallel at the same time. So again that kind of
that extension of Moore's law, more and more hardware on
a chip to be able to do more and more
of those calculations in parallel and come up with it.
Speaker 3 (19:36):
And we said, yeah, was that threshold predictable? In other words,
see people in the industry, like you sit down X
number of years ago and say, when we get here,
AI is going to become much more of a It's funny.
Speaker 4 (19:48):
The horsepower that very predictable, the use cases not always
so easy to kind of figure out. That's where the
human spirit kind of gets involved. I think for some
people that say, oh, I saw that coming, but people
have been predicting kind of the rise of AI for
twenty five years. Oh well, then when we get to
(20:09):
this next gener oh when we get here, it kind
of hadn't happened. There's always a magic point where you
kind of get to where the technology and the use
case and somebody does something to kind of make it
catch on. And I think we're at one of those
moments in AI for sure right now. And I don't
think it's you know, people that have said, oh, this
is just the latest wave of you know, I hear
(20:30):
I've heard this about a lot of technologies, but AI
is the technology the future, and it always will be.
I used to hear that. You're not hearing that now, right,
It's like, no, it's primetime. It will change everything, just
like some of these other things changed everything.
Speaker 3 (20:45):
I noticed it if personally when I speak somewhere or
I'm listening an audience somewhere. Over the last let's say
twelve months, there's always a whole bunch of AI questions. Yes,
And if I go back to years ago, there were
no AI questions.
Speaker 4 (21:01):
Yes.
Speaker 3 (21:02):
Now my question is, so there's been this explosion on
the in popular fascination with what's going on AI. It
seems like the last year. I agree with you in
your world, when did the explosion of conversation around this start?
Speaker 4 (21:19):
It's I love this question because IBM had a fairly
big effort and business called Watson before Watson X. And
this is going back kind of ten years. I'll give
you another kind of example. I knew about a lot
(21:40):
of tablet technology before there was an iPad, a lot.
For ten years, there were a lot, but it kind
of takes a magic combination of the technology, the user experienced,
the software, and the need and the market ready for
it to kind of go. Now it's the thing. Now
we all have either an iPad or we have the
Google equivalent Tom and so I think this is a
(22:00):
little like that, meaning IBM was on the right track
with Watson. Some of the hardware wasn't there, the use
cases weren't exactly figured out. Some of the early use
cases didn't pan out. Perfectly. But the good news about
that is it's back to that culture of risk taking.
You don't look back on that and say, oh, we
shouldn't have done that, that was a bad idea. I
know you look back on that and say, what did
(22:21):
we learn? How should we try something new? How would
we pivot this time? That's what we've done with Watson
X and now that's a growing, healthy piece of our
business and very important our strategic sure, so we're all in.
Speaker 3 (22:33):
I've always investigated by the gap between insider sense of
what is happening in an outsider sense, like.
Speaker 4 (22:42):
It absolutely is that in this case, we've all been
talking about and thinking about AI and is it time
for that and what does this mean? Et cetera. And
yet none of us really predicted that actual moment, which
is kind of you know, early twenty twenty two where
it was like, oh, now you have a simple human
(23:02):
interface of software innovation combined with large language models. There's
a moment there where you're like, oh, Unlike, you know,
I think all of us are frustrated if we ask
our phone, hey, tell me about this, and it says
I found this on the web page. That does you
no good? But you know, all of a sudden, with chat,
GPT and some of these other things, you could ask
(23:24):
a question, it would give you a clear answer. Sometimes
is wrong, but at least it was like I'm getting
an answer rather than hey, I don't know if there's
some references. Good luck to you. And that's really changing.
Speaker 3 (23:34):
Talk about the kind of macro trends that are going
to shape your infrastructure battle.
Speaker 4 (23:42):
Yeah, we've talked about if you already, but I'm actually
going to go a little different direction. So macro trans first,
and this one has been before even even this AI conversation,
that we've had explosion of data. As humans, we don't
think exponentially very well. We really struggle with exponential thinking.
(24:04):
We think linearly, Oh, there'll be more, there'll be more,
they'll be more, but we don't think well when it's
like no, they'll be more, and they'll be ten times more,
and then there'll be ten times that more. That's what's
going on with data right now in our industry. It's
one of the reasons that that storage business is doing
so well is they're just more and more and more data.
You know, you'd say, well, how can there be more data?
It's just life and that thing. The things that we
(24:27):
care about, video capture, video images, you know, the the
you I don't know from my parents, you needed a
drawer with all your family photos. Now we need gigabytes
and gigabytes.
Speaker 3 (24:38):
You knew how many pictures my wife has taken off
our children, you would exactly exactly, So that's your case.
Speaker 4 (24:44):
Now. Think of companies who used to just think about
their transaction data. What's the ledger say that now have
video assets of all of their campaigns and their marketing.
They're trying to figure out, you know, what campaigns are
working the best. So it's just an explosion of data
and that's not going to stop. Dealing with that, and
more importantly, getting value from that data is a massive
(25:08):
trend in the industry. Second trend AI, and this is
the AI. Not like we were just talking about about
how it changes how I search for things or how
I learn about things. But I would argue, dealing with
that data, how do I figure out what's in all
those video streams? How do I figure out Okay, I
want all of the chunks of my corporate video that
(25:30):
have to do with client buying some specific product or something.
That's a different problem. It's not just okay, we'll look
it up in a spreadsheet and here's the math associated
with that. That is a huge trend in the industry.
You're seeing it play out in this regard, it's a
little different bent on AI. Fraud detection is the one
that we cite in our mainframes. It's a similar problem
(25:53):
where it was kind of a traditional AI problem. Look
up a rule. You know, if somebody does two small
trend actions than a massive one, it might be fraud,
right because they were seeing whether it were now to
detect fraud, you might be saying, okay to transactions then
a huge one. Plus does this entity have a real address? Second,
(26:14):
is there any web traffic on you know, better Business
Bureau kind of things that says this is a bad
business that can help you with fraud. So it's a
lot more of a it's an EXPERTI problem. It's a
holistic problem that it takes a lot more than just
you know, little chunks of rules, et cetera. And then
the third one you know after AI, is the nature
of hybrid it or hybrid computing. For a while ten
(26:38):
years ago, when cloud was on the rise, I think
the notion of hybrid computing basically having to do with
things in the cloud versus things that people still have
on the premises inside of business. It was almost a
religious argument. Now it's no, it's the reality. And the
reason is because that data that I talked about is
(26:59):
the life blood of these companies, particularly IBM's companies are
clients that usually that data has to be secure, they
have to be able to get value from it. It
is the lifeblood of the company. If you go to
an ATM and you can't get your money out to
our financial transactions, if that lasts a day, you're probably
going to change banks immediately. So it's like life or
(27:21):
death for these companies. So having that hybrid infrastructure so
that they can still hold their data, you still interact
with clouds and still get value from it from AI.
That's kind of the magic where we play, and it's
a huge business opportunity. It is a true inflection point
(27:41):
for the industry. I'm going to go back.
Speaker 3 (27:45):
I interrupted you when you were in the middle of
a rellion. We were talking about what has to happen
for AI to scale from the infrastructure standpoint. You gave
one example that I got you off on a tangent.
Can you go back and talk very so practically, like, so,
I'm you know, I'm a big company. I have all
these dreams of AI, of how I'm going to use
(28:08):
this dratically, So give me a very granular sense of
the works you have to do, yeah, to make that
dream possible.
Speaker 4 (28:15):
So let me first say what the company has to do,
and then maybe I'll say, then how do I help them?
If that makes sense? So if I'm a company and
I want to do that, So it turns out I
am a company meaning I want to use AI in
my processes. I mentioned that I have a global network
of thirteen thousand employees that support our infrastructure around the world.
(28:37):
That challenge is a great challenge for AI. That means
I have data for every customer situation for thirteen thousand
employees globally around the world on what was their problem,
how did we fix it, what next steps did they
have to do, how did they remediate that? That data
is extremely valuable to me because if I can get
(28:59):
better at doing that than anybody else in the world,
that brings my cost down. I sell more products, I
sell more service, I sell more anything. So what I
have to do to get there is I have to
figure out. Okay, what's my objective? I have a couple objectives. One,
I want customers to be able to support themselves without
even calling me, first off, and I don't want when
(29:19):
they call for the first answer to come back to
be did you try rebooting? Because I think that irritates
every single one of us. Did you try? Of course
I tried rebooting. I've had a laptop, of course I well, okay,
well then tell me, okay, what firmware version, all that
other stuff. Okay, we know this interaction. So that's kind
of the problem set. Do I want that to be
(29:40):
customers solving their own problems? Well, even for my support agents,
I want something in their pocket on their phone where
they say I'm seeing these symptoms and says, oh, this
happening around the globe. Here's kind of specific me. So
there's my problems. What does it mean for infrastructure on
the back end? So first I got to get all
that data together, right, all of those customer law, all
(30:02):
that customer support around the globe, et cetera. That needs
to be stored. That's a big set of data. And
some of it's not just fix and that kind of thing.
Some of it is Okay, you know what was the
firmware version? Who was the tech because it can matter.
Is this their first time fixing this problem? Is it
there one hundred and fiftieth time? What's their level? It's
a very complicated problem. Ingesting all that data takes an architecture.
(30:27):
We have a product called Scale, which is one of
our storage projects that actually makes it easy to ingest
all that data, get it organized, et cetera, and then
have a model. It's a whole different process to kind
of say did we train our model? We can train
our own models inside of IBM. We have a granite
set of models. Those models we fine tune, and then
(30:48):
we inference based on those models. So we can do
that inferencing in our cloud I have a cloud set
of infrastructure, or in my power servers. We can do
inferencing with our capabilities and say, okay, based on what
I'm saying, here's what the remediation that you should do
for that customer. We already are doing that today. We've
seen over a third of our support calls have had
(31:13):
significant reduction in the amount of time that it takes
to resolve that support call just by what I said
right there.
Speaker 3 (31:20):
That I've really been curious about this. If I had
reduced something like AI into that equation as you just did. Yeah,
and you said we've already seen a thirty percent Say
did you say thirty percent reduction?
Speaker 4 (31:33):
Thirty percent of our interactions have seen significant reduction in
those time?
Speaker 3 (31:39):
Was that your primary goal to reduce the time of
the interaction? But it was you if everything else was
the same all, but what you were doing was shrinking
the amount of time?
Speaker 4 (31:48):
That would you want one of the primary goals, So
to us in that business net promoter score kind of
the satisfaction of a client is the supreme goal. What
makes them sad fine, doesn't cost me a fortune, happens
really quickly, and if I can do it myself, I'd
be thrilled. It affects all of those right. It kind
(32:08):
of says it got resolved faster, it didn't cost me
an arm and the leg because the deck was barely here,
because it's a common problem, or I solved it myself
without even calling, So all of those objectives would kind
of hit across all so that now you see it.
So that's a little microcosm. That's just me and my
customer support business. Now think of how many problems for
businesses around the world there are like that. It's not
(32:31):
a it's not like a new AI application that changes
the entire user experience. That's those will come, But right
now it's kind of practical, which is, I just want
to do what I'm doing better and faster, and I
can get immediate economic return from those things.
Speaker 3 (32:47):
How long How long did it take you to just
stick with that example of the customer reaction reducing thirty
percent of the time? How long from the very beginning
of that project, Yeah to that thirty percent reduction it was.
Speaker 4 (33:00):
Hell long, less than a year. And yeah, So one
of the challenges, and this is interesting with a very
large organization, as you can imagine, just like you're seeing
in the industry, we don't have a problem of generating
ideas for how AI could help us. We actually have
a problem filtering the thousands of ideas from our employees
(33:24):
and from everywhere. It's like, hey, we could use AI
to and filtering down and saying, okay, which of these
will have a return on investment quickly and at a
level that sustains that's worth kind of going and investing
in the infrastructure and the software and kind of making
that happen. Is that unusual.
Speaker 3 (33:41):
If I talked to you twenty five years ago and said,
do you have a problem of too many good ideas
or too few?
Speaker 4 (33:47):
What was you said in this specific area, Probably too few,
because at some point you reach diminishing returns. So, for example,
let's use this same example. Can those thirteen thousand technicians
go faster? Can they spend less time driving to the side.
I mean, there's only so much you can kind of
(34:07):
do on those things. But if you can get them
an answer to the problem and maybe even avoid them
having to visit at all because the client helped themselves,
that's a step function. So that's why people are kind
of talking about there's a business revolution coming with AI
where there are some step function changes that can be there.
And notice I didn't say I'm going to have less
(34:29):
of those agents. That's not my objective. My objective and
I think that's the fear in the industry about AI
is going to eliminate all the jobs. No, I just
created thirteen thousand superpowered agents that can do more right.
And so I'm not just going to support IBM products.
I'm going to go out and support other people's products
because I know how to do that really well. And
once I have the data on how to fix their problems,
(34:51):
I may just have a customer support business that's independent
of my boxes. So you know, I think that's where
people sometimes get it wrong. And the AI thing is,
it's like, you know, did word processing eliminate the need
for writers? No? It enabled writing instead of mucking around
with mimeographic machines and click and click typewriters. It may
(35:13):
have enabled too much writing? Yeah, maybe maybe can I
give you a hypothetical?
Speaker 2 (35:18):
Uh?
Speaker 3 (35:18):
And I asked this because I ran I was at
some conmis and I ran into some guy from the
I R S who was really, really, really really excited
about AI. So let's suppose they call you up and
they say you're going to talk to the I R SKY.
I call you up and I say, Rick, Uh, clearly
(35:40):
there's something that we could do for the I R
S if we work together.
Speaker 4 (35:45):
Yeah, what would your answer? Of course?
Speaker 2 (35:49):
No.
Speaker 4 (35:49):
I think we sell to a lot of government agencies.
I can imagine in the business that we're in, we
enable a lot of social security transaction and things like
that through our mainframes. And I think, you know, we're
in the business of helping whatever client get the most
out of their data and be able to secure it
(36:10):
and be able to do analytics with this. And IRS
has a heck of a lot of data, so yes,
we would help them.
Speaker 3 (36:17):
Do you know how the amount of data they have
compares to some of the corporate clients you've I.
Speaker 4 (36:21):
Don't know specifically for the IRS how much data they have,
but I would assume it's a whole lot. It's mountains.
But that's our business. I mean, it's interesting sometimes people
of that what's the most you know, what is it
that IBM has that's of great value? Is it a server?
Is it a storage array? Is it you know, software
(36:42):
and all that. What we have is the most important
entities in the world have their data on our stuff.
The most important data in the world. It's not you know,
pictures of your grandkids and things like that. Generally for us,
it's all of the financial transactions that have and globally,
right it's all of the it's the world's economy is
(37:03):
kind of running through our systems, and so we take
that really seriously. You know, you would be distraught if
you lost one photo on your laptop or whatever, but
you know, if we lose a transaction, like somebody moves
a big amount of money and it's like, well, don't
know what happened there. It is a massive deal, right,
so that doesn't happen.
Speaker 3 (37:23):
But I want to go back to my irs example
for US, Yes, so one, is it reasonable to assume
that you could that somebody IBM or somebody else could
in a short period of time put together not just
the AI capability to audit returns, but also this the
infrastructure support for that in a reasonable amount of time
(37:45):
for a reasonable amount of cost. Or is it over?
Is it going to the moon? Or is it it?
Speaker 4 (37:51):
Definitely? I mean so we're already doing that kind of
thing right across a network of banks and others. Yeah,
essentially all credit card transactions for all of the world
to go through our systems, So that in some ways
is more volume than the datch returns of the US people.
(38:11):
And they're W two's and all that stuff, and we
do that stuff too. I try not to describe it
too much in detail, but we definitely do a lot
of that. In fact, I think most of if you think, okay,
what is super critical data, who would be doing the
business transaction processing it is most likely us in almost
(38:33):
all cases, whether it's government things or private or banks
or that kind of thing. That's what we do.
Speaker 3 (38:40):
Rick we're going to end with the where we always
end with a couple of quick fire questions. Okay, here
we go. What single piece of advice would you give
to businesses trying to use AI in an effective way?
Speaker 4 (38:52):
The simple version is get started. By get started, I
mean think of what is something that I want to improve.
The things that we have traction on right now in
the market are around business process, automation, digital labor, those
kind of things. But my other little piece of advice
there is keep it simple to begin with. You're going
(39:13):
to learn a lot, but getting started means you'll start
that learning curve. I even advise you My friends like, hey,
should I be playing around with some of this AI stuff?
And I say yeah, because I think it will help
you start to be more comfortable and you may find
a use case personally for that. I think the same
is true for businesses. The first step in that journey
is always with what data. Notice when I talked about
(39:36):
our customer support people, I thought about, Okay, what's the data.
The data is all of those logs of all of
those service engagements around the world, and what could I
do with that? Well, I could use that to get
to a knowledge base that really helps and hopefully that
I can do it in multiple languages because it's global
and I can you know, all of those things. That
(39:57):
was kind of my data sent That one's not super simple,
but we've had a lot of experience in AI for
other people that might just be how do I automate
filling out travel expense reports for my company? We can
help people that we have consulting, we have wats and
X tools. We can do that like this, and we're
doing it globally for people around the world. Pick that thing.
(40:18):
What's the data you have? In that case, it's data
of expense reports and it's like, okay, we can help
you automate that for people where they could do it
just by you know, a verbal interface. What did you spend,
where did you go? Who you were you with? Okay,
we filled out your travel expense report for you and
you don't have to mess around with it.
Speaker 3 (40:36):
So we were playing with this idea where we would
pick a business and go in there and do it
would be AI makeover. Yeah, I love that what's okay,
what is the ideal business to do? We only have
a couple months. We don't want to spend a kajillion dollars.
We want to be able to show tangibly and quickly
what AI can do. What's an ideal business to do
(40:57):
that in It can be a small business. We're not
talking this grand corporate thing there.
Speaker 4 (41:02):
Ah boy, small business that we could do and hey,
I make over. Customer support is one of my favorites
because it's a it's it's I have it on the
business side where I provide customer support. I have it
on the consumer side, where it drives me nuts when
I have to go through thirty layers of phone menus.
(41:23):
Speak to an agent, speak to an agent, speak to
an agent. That for any business, I think is just
ripe to be able to kind of say why do
I have to click through these manucent messages. I just
need to tell you in human language, here's the issue,
and I'll be really good about telling you details about
You know, I tried to set up this thing for
my bank and I do da da da da da.
(41:44):
They can go through all the menus automate that process.
I think it would change everything because all that frustration
as a consumer would go down dramatically, and it's all,
you know, why are you making me the beep booth
press one? Exactly? Well, don't offload to me, offload to AI.
We can help you with that.
Speaker 3 (42:05):
Here's my version of that drives me crazy. Every morning
I go to the same coffee shop and I get
a cup of tea and a croissant.
Speaker 4 (42:15):
And here's what happens.
Speaker 3 (42:16):
A person has their screen and they go, I go,
cup of tea, croissant, sparkling water, like at least twenty keystrokes,
and then like then the screen is turned around. Like
at this point we're like forty five seconds in, I'm like,
why is this? First of all, it's not for me,
all those keystrokes, it's their internal right, right, So they're
(42:39):
burdening me in order to service.
Speaker 4 (42:40):
To back it, you should be able to walk in,
go up and they go, I'm olc them the same
thing and you just go yes, and then.
Speaker 3 (42:46):
The boom, We're done. Can we do AI makeover of
my coffee shop?
Speaker 4 (42:51):
You notice I quickly jumped more to banks than your
coffee shop because I think I'm a business person, but
I'm not trying to kind of do a deal on
one coffee shop.
Speaker 3 (43:01):
But this is interesting because it takes me back to
something you said that I thought was really important. When
you were talking about when you were using AI and
your customer service thing, it was clear that your goal
you could have any number of goals, yes, going in.
It could be to cut costs, it could be to
dramatically improved profits.
Speaker 4 (43:21):
Your goal, quite.
Speaker 3 (43:21):
Specifically, was to improve the experience of your customer, right,
So you were.
Speaker 4 (43:25):
Using it to that. All the other things come from
that come from. That is actually one of the beautiful
pillars of the IBM culture is delighting clients is actually
where all of the good stuff comes from.
Speaker 3 (43:37):
So my coffee shop thing is the same principle. Right now,
they're making my customer experience worse and they don't want to,
but their eyes are glued to the special a moment
when I walk in and I want to say, Hi,
how are you doing? We could have a conversation. You're
too busy, busy pooping, So like, this is the same thing.
If they had it that, oh, this isn't if they
(43:58):
understood they had an operation need to improve the experience
of their customer experience.
Speaker 4 (44:02):
I would not be surprised if a chain comes along
where that is their value proposition. I would not be
surprised at all. Yeah, yeah, right, So I mean and
and when those things kind of catch hold, it becomes
a revolution.
Speaker 3 (44:18):
You know, when the guy comes to do like to
redo your roof and they put a sign out front,
like you know, Joe's roofing. You guys could do the
same with my coffee shop. But like I'd be i'ure
was here exactly exactly.
Speaker 4 (44:35):
In five years, the main frame will be dot dot
dot going strong, the mainframe going strong and with new capabilities,
continuous new capabilities. I think when we announced the last
version Z sixteen, the latest version, I should say, and
(44:55):
we said, hey, there's AI processing built into it. This
was before everybody was talking about that. I think a
lot of people thought, what's that for? And we did
it specifically for traditional AI fraud detection, et cetera. This
next version, not only do we have the traditional AI
built in, but we have optional cards that you can
plug into it to allow you to do large language
(45:16):
models for the enhanced fraud detection cases that we talked about,
where you know, it's more than just what transactions were happening.
So if you take that and say, okay, the next generations,
we have more transaction volume than we've ever had in mainframes. Today,
the business is growing, it's strong, we keep innovating. In
(45:38):
five years it'll be going strong.
Speaker 3 (45:39):
But we're people. You're saying this in the context of
for years people were predicting, weren't they that the main
brand was going to go away.
Speaker 4 (45:48):
There were pundits in the market that said everything will
go away there, no one will ever have a box,
It'll all be online. I think this is something I've
learned big time in my long career. You know in
the IT industry is don't believe everything you hear. So
I went back for my master's degree at Stanford after
(46:08):
I had worked a while in as a hardware designer,
and everybody told me be sure to do your masters
in software. Hardware is dead. I went on to work
for thirty plus years in hardware and infrastructure. Now software
became important, and I'm glad I had that extra training
in software because it helped me in hardware. But hardware
wasn't dead. Then I heard all infrastructure will go into
(46:32):
the cloud. There won't be that hasn't happened, it's not happening.
Then I heard there will only be one cloud because
one of the players will dominate. There's not one cloud.
So I think it's as humans we like to oversimplify
and go, oh, it's all going to be this, And
kind of what I've learned is fit for purpose matters
in everything. It matters in size of infrastructure, it matters
(46:56):
in the stack that goes along with solving a specific
use case. If you're willing to design something that's the
best at that use case, if you're willing to design
the coffee shop that is the best at greeting me,
there's a spot for you, and there may be a
big business in doing that. So oversimplifying is really.
Speaker 3 (47:13):
When you heard all those predictions, did you believe them
at the time.
Speaker 4 (47:19):
They looked like they were trending in that direction. I'll
tell you some right now which might be useful. There
will only be one GPU company and they're going to
end up taking over the world. It's a pretty obvious answer.
Whose economic values risen dramatically. I don't think that's going
to be the case. In fact, I think that ninety
percent of processing for AI actually happen happens at inferencing,
(47:42):
and inferencing is not as GPU and hardware intensive as
the other things and is a lot more amenable to
fit for purpose. So the model size will matter. The
tuning matters a lot. As we're learning. We have a
product around instruct lab that's really focused on tuning. So
that was one thing is there'll be one GPU. The
other thing is that the biggest model will win. I
(48:04):
think is another thing that's kind of people are saying
right now. Don't believe that I believe they will be
fit for purpose models. It takes a lot of money
to run to create a huge model, and then to
run a huge model, or to even infer off of
a huge model. I don't need a massive training GPU
set thing to solve my thirteen thousand people customer support issues.
(48:26):
So why would I feel like I got to go
farm that out for a big expensive thing. I can
do that on a small box. In some cases I
might even be able to do that on a laptop.
The other thing I'll say in this we are so
early innings in AI, A lot of things are going
to change. So anybody kind of saying it will all
be X, Y or Z, I just think you have
no idea how this is going to play out, and
(48:47):
it's up to us to go figure out how it
plays out.
Speaker 3 (48:50):
Yeah, yeah, all right, in five years AI will be
dot dot dot still new.
Speaker 4 (48:59):
It will have moved a bunch in five years, but
the potential for the disruption in the world will still
will still be very early innings in that process. And
I think that's super important to realize. That's why I
say get started, start thinking about how that could change,
because it'll be some little things first, but it will
(49:19):
continue to snowball.
Speaker 3 (49:20):
This is a common observation that we the invention of
the capability massively predates the understanding.
Speaker 4 (49:30):
Of the capability, right, Like I love that.
Speaker 3 (49:33):
Yeah, Like yes, recorded recording shows on television is invented
in the sixties. Probably we don't really understand what it's
used for until the oughts was what's really good for
is being able to tell a story sequentially, Yes, over time,
(49:54):
because you know that the person will always see in
the episode before, so you get the Sopranos. And yes, yes,
Hollywood wanted to ban the VCR in the beginning, Yeah,
because they thought it was good. They thought the point
of it was thought then understand. No, No, No, it's storytelling.
It's actually your business is getting better. Yes, took them
twenty years to figure that out, which is to your point,
(50:14):
why would we know what AI was four and five years.
Speaker 4 (50:16):
Well, that's why you hear people kind of say, oh
my gosh, AI's that will just eliminate jobs. No, it'll
make jobs better. That's how I view it.
Speaker 3 (50:23):
Yeah, what's the number one thing that people misunderstand about AI?
Speaker 4 (50:27):
Is that it that it'll I think that's that. That
would be the human kind of understanding part of it.
The technology part of it, I think would be what
I was talking about fit for purpose, meaning that it
isn't just going to be a GPU arms race all
of AI. I don't believe that at all. It will
change everything, but it's not just going to be a GPU.
Speaker 3 (50:48):
Armed Next question, what advice would you give yourself ten
years ago to better prepare you for today?
Speaker 4 (50:54):
I'm changing this question, okay.
Speaker 3 (50:58):
I would say, let's imagine that at what was your
what college.
Speaker 4 (51:02):
You to go to? I went to three of them.
My undergrad was Utah State University, my NBA was Santa
Clara University, and my master's in w was Stanford.
Speaker 3 (51:12):
Okay, any one of those three culture up and says
we want you to give the commencement address and imagine
it's it's it's let's just say, for the sake of argument,
it's just to the stem people.
Speaker 4 (51:26):
Those are the relevant parties here. What do you tell them? Boy?
What do I tell them? Let's see. I think I
would start with life is a marathon, not a sprint.
It would be the first one. The second thing I
would say that in that spirit is be sure to
(51:49):
set yourself some big, hairy, audacious goals and don't be
overly disappointed if you don't hit them all. Going after
those big, hairy, audacious goals will get you on a
path where you will learn so much. You will achieve
more than you ever could imagine you would have achieved.
(52:09):
That's what the advice I give to my kids is, ye,
set some big goals, get after it. You may or
may not achieve them, but you'll be better for the
whole process when you're done.
Speaker 3 (52:17):
By the way, as someone whose kids are younger than yours,
is it actually useful to give your give advice to
your kids?
Speaker 4 (52:24):
The pointless exercise TVD. We're still on the journey, and
I think we will be for a long time. I
don't know how are you already using AI in your
day to day life today? Personally, I would say it's
replacing a good chunk of my search. You know, I'm
less likely to go blindly stumbling through a bunch of
(52:45):
web pages looking for stuff. I'm more likely to ask
a question from a few AI engines kind of see
get me in the right direction. Then I'll go bumble
through a few things. At work, I can tell you
code development right now, we are seeing massive improvements in
code development and support. Products we have like Watson Code
(53:06):
Assistant that is really showing immediate return for a code developers,
and I think that will again be a tool that
increases productivity for code developers immediately across the globe. Yeah.
Speaker 3 (53:20):
Last question, what's the one skill that every technology leader
needs that has nothing to do with technology.
Speaker 4 (53:26):
Being able to inspire a set of people toward a
common goal and collaborate to achieve it. That's at the
core of everything everything. That's a lovely way to end.
Thank you so much, Rick, Thank you.
Speaker 3 (53:42):
This conversation left me excited. I'm now imagining the potential
for new use cases for AI in all sorts of
different businesses. Rick didn't seem soldom my idea of a
coffee shop makeover, but it's clear there's lots of opportunities
here to increase speed and efficiency, to achieve your objectives,
and to dream beyond the current applications for this technology.
(54:05):
At the end of the day, the scaling of AI
will rely on the right infrastructure to support it. With
the right tools, you can solve problems that are unique
tier industry and improve the experience for your customers. Smart
(54:30):
Talks with IBM is produced by Matt Romano, Amy Gains McQuaid,
and Jacob Goldstein. Were edited by Lydia gene Kott, mastering
by Jake Koorsky. Theme song by Gramoscope. Special thanks to
the eight Bar and IBM teams, as well as the
Pushkin marketing team. Smart Talks with IBM is a production
(54:50):
of Pushkin Industries and Ruby Studio at iHeartMedia. To find
more Pushkin podcasts, listen on the iHeartRadio app, Apple Podcasts,
or wherever you listen to podcasts. I'm Malcolm Gladwell. This
(55:11):
is a paid advertisement from IBM. The conversations on this
podcast don't necessarily represent IBM's positions, strategies, or opinions,
Stuff To Blow Your Mind News
Hosts And Creators
Robert Lamb
Joe McCormick
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.
What Are We Even Doing? with Kyle MacLachlan
Join award-winning actor and social media madman Kyle MacLachlan on “What Are We Even Doing,” where he sits down with Millennial and Gen Z actors, musicians, artists, and content creators to share stories about the entertainment industry past, present, and future. Kyle and his guests will talk shop, compare notes on life, and generally be weird together. In a good way. Their conversations will resonate with listeners of any age whose interests lie in television & film, music, art, or pop culture.