WEBVTT - IBM CEO Arvind Krishna Talks AI

0:00:02.759 --> 0:00:08.640
<v Speaker 1>Bloomberg Audio Studios, Podcasts, radio News. We're going to go

0:00:08.720 --> 0:00:11.080
<v Speaker 1>to what's happening with IBM. I'm pleased to welcome the

0:00:11.080 --> 0:00:14.600
<v Speaker 1>CEO of Vin Krishno, who's coming from the Companies Think Conference,

0:00:14.640 --> 0:00:18.040
<v Speaker 1>an annual event, and Avin, I'm just excited to hear

0:00:18.079 --> 0:00:21.159
<v Speaker 1>really from you the pace of change that's happening with

0:00:21.280 --> 0:00:23.640
<v Speaker 1>Agenda AAI. I think you put out there that in

0:00:23.720 --> 0:00:27.280
<v Speaker 1>three years time there's going to be one billion apps.

0:00:27.600 --> 0:00:30.120
<v Speaker 1>It is so noisy, it is cluttered. How are you

0:00:30.160 --> 0:00:32.760
<v Speaker 1>may managing to streamline it for your customers?

0:00:33.159 --> 0:00:37.000
<v Speaker 2>Look first, I think that number is the really interesting number.

0:00:37.479 --> 0:00:41.360
<v Speaker 2>A billion new applications using AI for a third of

0:00:41.400 --> 0:00:44.680
<v Speaker 2>them using agentic AI. And the way that we are

0:00:44.680 --> 0:00:47.000
<v Speaker 2>approaching it is, how do we make it easy for

0:00:47.040 --> 0:00:50.880
<v Speaker 2>our customers leverage their own data, insert their own data

0:00:50.880 --> 0:00:53.559
<v Speaker 2>into the models, and then leverage it to make the

0:00:53.720 --> 0:00:56.880
<v Speaker 2>enterprise a lot more productive. That's what we're after building,

0:00:57.200 --> 0:01:00.000
<v Speaker 2>and that's what we're just announced with our watsonex hocice

0:01:00.040 --> 0:01:05.520
<v Speaker 2>straight with agentic capabilities, including one hundred and fifty new agents,

0:01:06.360 --> 0:01:08.680
<v Speaker 2>some from us and some from our partners.

0:01:09.480 --> 0:01:13.200
<v Speaker 1>One hundred and fifty agents sounds an extraordinary number of it.

0:01:13.280 --> 0:01:15.399
<v Speaker 1>And look, you've got a big book of business already

0:01:15.400 --> 0:01:18.360
<v Speaker 1>when it comes to generative AI six billion dollars in growing.

0:01:18.840 --> 0:01:21.960
<v Speaker 1>But how much are you seeing the CEOs you serve

0:01:22.520 --> 0:01:25.920
<v Speaker 1>pushing back on overwhelm. How are you managing to help

0:01:25.959 --> 0:01:29.120
<v Speaker 1>them navigate what feels like in some ways a less

0:01:29.160 --> 0:01:31.880
<v Speaker 1>than productive environment for the amount that they're investing in

0:01:31.920 --> 0:01:32.559
<v Speaker 1>generative AI.

0:01:34.000 --> 0:01:37.720
<v Speaker 2>Look, I think that what CEOs are concerned about is

0:01:37.760 --> 0:01:40.880
<v Speaker 2>that now that you have gone, the era of extraventation,

0:01:40.959 --> 0:01:43.160
<v Speaker 2>as I call it, is over. Now people are looking

0:01:43.200 --> 0:01:47.360
<v Speaker 2>for business value on AI. As they're doing that, they've

0:01:47.400 --> 0:01:50.280
<v Speaker 2>discovered that about twenty five percent of the time these

0:01:50.320 --> 0:01:53.320
<v Speaker 2>projects are paying off very well for them. So learn

0:01:53.360 --> 0:01:55.960
<v Speaker 2>from that twenty five and then figure out what are

0:01:55.960 --> 0:01:58.480
<v Speaker 2>they enhabitor as in the other seventy five and take

0:01:58.520 --> 0:02:02.040
<v Speaker 2>those out, and those are in variably there's fragmentation in

0:02:02.080 --> 0:02:05.400
<v Speaker 2>the environment. So how do you have a more integrated environment.

0:02:05.720 --> 0:02:08.080
<v Speaker 2>You've got to make sure that there is value in

0:02:08.120 --> 0:02:11.280
<v Speaker 2>the AI projects. Don't just do an experiment and leave

0:02:11.320 --> 0:02:13.760
<v Speaker 2>it on the shelf. You've got to scale it and

0:02:13.840 --> 0:02:16.799
<v Speaker 2>deploy it across the enterprise. So how do you integrate?

0:02:17.040 --> 0:02:19.639
<v Speaker 2>How do you create the right hybrid environment. That's what

0:02:19.639 --> 0:02:22.120
<v Speaker 2>we're helping them with, and then we believe that twenty

0:02:22.160 --> 0:02:24.600
<v Speaker 2>five we'll go all the way up to fifty or

0:02:24.639 --> 0:02:27.760
<v Speaker 2>seventy five, at which point that's a great return for

0:02:27.840 --> 0:02:28.520
<v Speaker 2>the CEOs.

0:02:29.240 --> 0:02:32.880
<v Speaker 1>That software that you're unveiling, those consulting offerings that you

0:02:32.960 --> 0:02:35.400
<v Speaker 1>continue to provide of in at the moment. How much

0:02:35.480 --> 0:02:38.160
<v Speaker 1>is that bearing fruit in this current environment? How willing

0:02:38.160 --> 0:02:41.560
<v Speaker 1>are these CEOs to experiment and move on from experimentation

0:02:41.639 --> 0:02:44.280
<v Speaker 1>when we've got a really uncertain economy on our hands.

0:02:45.639 --> 0:02:49.440
<v Speaker 2>Look, I'm a more of an optimist on the uses

0:02:49.480 --> 0:02:53.800
<v Speaker 2>of technology in the current environment than a pessimist. And

0:02:53.840 --> 0:02:56.320
<v Speaker 2>why do I say that. So when you have supply

0:02:56.400 --> 0:03:00.840
<v Speaker 2>chain issues, you have cost issues, you have productive issues,

0:03:01.280 --> 0:03:04.840
<v Speaker 2>technology becomes an answer to help you scale the business

0:03:05.200 --> 0:03:07.840
<v Speaker 2>without having to add a lot of physical cost into

0:03:07.880 --> 0:03:10.880
<v Speaker 2>the business. So I find the appetite of the CEOs

0:03:10.919 --> 0:03:14.120
<v Speaker 2>is quite high to spend on technology as long as

0:03:14.160 --> 0:03:16.560
<v Speaker 2>they can see a return that is our college short

0:03:16.639 --> 0:03:20.520
<v Speaker 2>to medium term meaning six months, twelve months, eighteen months.

0:03:20.840 --> 0:03:23.400
<v Speaker 2>I find people leaning in very heavily, and I think

0:03:23.400 --> 0:03:26.840
<v Speaker 2>that's why you see enterprise tech spending has remained robust

0:03:27.200 --> 0:03:30.240
<v Speaker 2>despite all of these issues that could be talked about.

0:03:30.840 --> 0:03:34.480
<v Speaker 1>What about government in the moment Scott beson getting grilled

0:03:34.480 --> 0:03:37.040
<v Speaker 1>on a hill at the moment around DOGE and that

0:03:37.120 --> 0:03:39.400
<v Speaker 1>has had implications for your own business. I think it's

0:03:39.440 --> 0:03:43.720
<v Speaker 1>about fifteen contracts we understand having canceled. Are more to come.

0:03:43.880 --> 0:03:46.400
<v Speaker 1>Are you seeing an end to that cancelation process.

0:03:48.160 --> 0:03:52.520
<v Speaker 2>I actually choose not to get worried about dos. I

0:03:52.560 --> 0:03:55.680
<v Speaker 2>think that the areas where we have very robust contracts,

0:03:55.880 --> 0:03:59.960
<v Speaker 2>when we're helping the VA process claims, we help the GSA,

0:04:00.080 --> 0:04:05.600
<v Speaker 2>say on expense management, we help the DD on payroll,

0:04:05.920 --> 0:04:08.480
<v Speaker 2>those remained very robust, and I don't believe that those

0:04:08.520 --> 0:04:12.400
<v Speaker 2>will come under any significant attack. Areas which do come

0:04:12.440 --> 0:04:15.400
<v Speaker 2>under attack are much more what I call discretionary, or

0:04:15.480 --> 0:04:19.760
<v Speaker 2>are much more time and material or labor augmentation based.

0:04:20.120 --> 0:04:22.960
<v Speaker 2>Those may come under attack. It actually does not worry

0:04:22.960 --> 0:04:25.480
<v Speaker 2>in me. I believe that even if those come under attack,

0:04:25.680 --> 0:04:27.720
<v Speaker 2>there is more than enough offset from the areas that

0:04:27.760 --> 0:04:29.680
<v Speaker 2>are more critical interesting.

0:04:29.760 --> 0:04:34.320
<v Speaker 1>So basically any cancelations will be outweighed by business that

0:04:34.360 --> 0:04:36.800
<v Speaker 1>you potentially can win within the government as well.

0:04:38.040 --> 0:04:42.200
<v Speaker 2>I believe that from having talked to very senior members

0:04:42.240 --> 0:04:45.600
<v Speaker 2>of the administration, there is a huge appetite to use

0:04:45.680 --> 0:04:49.280
<v Speaker 2>technology to modernize government and to help make government itself

0:04:49.320 --> 0:04:54.120
<v Speaker 2>more efficient. You can't achieve efficiency just based on cost cutting.

0:04:54.360 --> 0:04:57.440
<v Speaker 2>You actually also will have to invest in technology. And

0:04:57.640 --> 0:05:00.560
<v Speaker 2>I believe, and not just my belief from how spoken

0:05:00.600 --> 0:05:02.400
<v Speaker 2>to them, they have an appetite to do that.

0:05:03.040 --> 0:05:06.360
<v Speaker 1>I'll tell you where there's appetite. Air traffic control and

0:05:06.440 --> 0:05:10.200
<v Speaker 1>Trump himself talking about maybe raythean IBM coming in to

0:05:10.279 --> 0:05:12.479
<v Speaker 1>help solve some of the real anxieties we've got just

0:05:12.480 --> 0:05:16.599
<v Speaker 1>a loss of radar over happening in Newark? Is that progressing? Ovin?

0:05:16.600 --> 0:05:18.440
<v Speaker 1>Are you able to share on how any sort of

0:05:18.440 --> 0:05:19.720
<v Speaker 1>contract might shape up?

0:05:20.960 --> 0:05:24.160
<v Speaker 2>Look, I don't want to talk about any specific contract

0:05:24.240 --> 0:05:27.479
<v Speaker 2>or any specific agency, but I think what he just

0:05:27.520 --> 0:05:30.680
<v Speaker 2>pointed out is a great example where technology will be

0:05:30.720 --> 0:05:35.480
<v Speaker 2>part of the answer. Better integration, better computing capabilities, better

0:05:35.480 --> 0:05:39.359
<v Speaker 2>communication capabilities, all of those will play a role because

0:05:39.440 --> 0:05:43.880
<v Speaker 2>right now we are living on not exactly but essentially,

0:05:44.080 --> 0:05:47.160
<v Speaker 2>I'll let's call it a slightly evolved system that was

0:05:47.200 --> 0:05:48.520
<v Speaker 2>built in the nineteen seventies.

0:05:49.800 --> 0:05:53.719
<v Speaker 1>Ovin, what's crucial about air traffic control is accuracy, and

0:05:53.800 --> 0:05:56.479
<v Speaker 1>I know that's almost where everyone's eyes are when it

0:05:56.520 --> 0:05:58.960
<v Speaker 1>comes to a GENTERKI when it comes to generative AI,

0:05:59.160 --> 0:06:02.080
<v Speaker 1>is the risk of the underlying data being inaccurate and

0:06:02.120 --> 0:06:05.240
<v Speaker 1>how that compounds? How are you managing to tackle that

0:06:05.279 --> 0:06:07.440
<v Speaker 1>sort of area of anxiety and reality.

0:06:09.160 --> 0:06:12.360
<v Speaker 2>I always break the problem down. The core of air

0:06:12.360 --> 0:06:16.200
<v Speaker 2>traffic control has to be built on very precise data,

0:06:16.200 --> 0:06:19.320
<v Speaker 2>completely accurate data. You're not going to go use agentic

0:06:19.360 --> 0:06:23.159
<v Speaker 2>AI to estimate where a plane is. That's going to

0:06:23.160 --> 0:06:29.440
<v Speaker 2>be based on much more physics and hard engineering. Now,

0:06:29.480 --> 0:06:33.479
<v Speaker 2>could agentic AI come in trying to optimize routes in

0:06:33.680 --> 0:06:35.919
<v Speaker 2>worrying about how do you increase the total flow and

0:06:36.000 --> 0:06:39.479
<v Speaker 2>capacity at an airport? Perhaps? But I'm using the word

0:06:39.480 --> 0:06:41.520
<v Speaker 2>perhaps because there's a lot of careful work to be

0:06:41.560 --> 0:06:45.839
<v Speaker 2>done there. But could agentic AI be used to process

0:06:45.880 --> 0:06:50.960
<v Speaker 2>claims faster at Social Security or the VA? Absolutely, and

0:06:51.000 --> 0:06:54.840
<v Speaker 2>then you make citizens happier, you deliver them services faster

0:06:55.440 --> 0:06:58.000
<v Speaker 2>at a lower cost. That's where I would go first

0:06:58.040 --> 0:06:58.839
<v Speaker 2>with agent KI.

0:07:00.440 --> 0:07:03.920
<v Speaker 1>In the private sector, they're looking for accuracy and that's

0:07:03.960 --> 0:07:06.400
<v Speaker 1>something you're offering at the moment. How much is you

0:07:06.560 --> 0:07:09.720
<v Speaker 1>just seeing the adoption and the use and productive use

0:07:09.960 --> 0:07:13.280
<v Speaker 1>of agenda KI being limited by that underlying data right now,

0:07:13.400 --> 0:07:16.320
<v Speaker 1>or is that something that we're streamlining already.

0:07:16.440 --> 0:07:20.480
<v Speaker 2>I albit advise our clients start with the use cases

0:07:21.040 --> 0:07:23.760
<v Speaker 2>where you can live with some risk, where you do

0:07:23.920 --> 0:07:26.880
<v Speaker 2>not need it to be one hundred percent accurate. So

0:07:26.960 --> 0:07:30.280
<v Speaker 2>if you think about customer service, you think about end

0:07:30.280 --> 0:07:35.640
<v Speaker 2>client experience. I think about our internal IT help lists.

0:07:35.960 --> 0:07:38.560
<v Speaker 2>Those are all areas where where you were dealing with

0:07:38.680 --> 0:07:41.800
<v Speaker 2>people you actually had a ten or fifteen percent errorrate.

0:07:42.160 --> 0:07:44.320
<v Speaker 2>So now that you deal with the AAI, you can

0:07:44.360 --> 0:07:47.640
<v Speaker 2>actually have a lower error than that, but maybe not perfect,

0:07:47.800 --> 0:07:48.920
<v Speaker 2>but you can still live with it.

0:07:50.200 --> 0:07:53.440
<v Speaker 1>Think is all about innovation. I'm looking at the car

0:07:53.560 --> 0:07:56.120
<v Speaker 1>behind you. I'm thinking about the way in which you're

0:07:56.120 --> 0:07:59.200
<v Speaker 1>also pushing the agenda, not only when it comes to hybrid,

0:07:59.200 --> 0:08:01.040
<v Speaker 1>when it comes to on when you thinking about the

0:08:01.080 --> 0:08:04.440
<v Speaker 1>innovations you're making on mainframes, but quantum A You're getting

0:08:04.440 --> 0:08:06.679
<v Speaker 1>a lot of questions on quantum right now. Scott Beson

0:08:06.720 --> 0:08:09.840
<v Speaker 1>wants the US to lead in AI but also in

0:08:09.920 --> 0:08:12.160
<v Speaker 1>quantum computing. How is that making a difference in your

0:08:12.160 --> 0:08:13.320
<v Speaker 1>business right now?

0:08:14.160 --> 0:08:16.880
<v Speaker 2>So we are very very pleased with the innovation we

0:08:16.960 --> 0:08:19.440
<v Speaker 2>brought out here. We talked about the AGENTAAI. We just

0:08:19.480 --> 0:08:22.920
<v Speaker 2>mentioned the mainframe. We can do four hundred and fifty

0:08:23.520 --> 0:08:27.720
<v Speaker 2>billion AI inferences per day on the latest mainframe we

0:08:27.880 --> 0:08:31.760
<v Speaker 2>just announced this quarter, the Z seventeen. Think about using

0:08:31.760 --> 0:08:35.040
<v Speaker 2>that in financial systems to help reduce fraud. So that's

0:08:35.080 --> 0:08:39.040
<v Speaker 2>an incredible area of innovation on quantum. Part of the

0:08:39.080 --> 0:08:43.959
<v Speaker 2>reason we announced the huge investment last year and into quantum,

0:08:44.040 --> 0:08:46.120
<v Speaker 2>and we just brought out the quantum system to this

0:08:46.240 --> 0:08:49.640
<v Speaker 2>year is because we are very excited where that will go.

0:08:50.240 --> 0:08:52.840
<v Speaker 2>That's an area in which we believe the US must

0:08:53.240 --> 0:08:56.880
<v Speaker 2>and can lead on quantum computing, both for national security,

0:08:57.120 --> 0:09:00.959
<v Speaker 2>but also for competitive advantage and for growing a business.

0:09:01.120 --> 0:09:03.439
<v Speaker 1>And is that a competitive advantage continue to grow here

0:09:03.480 --> 0:09:05.840
<v Speaker 1>in the United States. You've just unveiled a huge amount

0:09:05.840 --> 0:09:07.960
<v Speaker 1>of investment into the US.

0:09:09.280 --> 0:09:13.120
<v Speaker 2>Yes, So we put one hundred and fifty billion dollars

0:09:13.160 --> 0:09:16.240
<v Speaker 2>here into the US because we believe that if I

0:09:16.240 --> 0:09:18.960
<v Speaker 2>look at AI, I look at hybrid cloud, and I

0:09:18.960 --> 0:09:23.360
<v Speaker 2>look at quantum coming, the growth opportunities are here. So

0:09:23.440 --> 0:09:26.439
<v Speaker 2>we want to invest to take advantage of these opportunities.

0:09:27.160 --> 0:09:30.600
<v Speaker 1>IBM Chairman President CEO Arvin Christian and live from think

0:09:30.960 --> 0:09:33.040
<v Speaker 1>We very much appreciate your time today