0:03

Hello, Hello, Welcome to Smart Talks with

0:05

IBM, a podcast from Pushkin

0:07

Industries, iHeartRadio and

0:09

IBM. I'm Malcolm Glapo. This

0:12

season, we're continuing our conversations with new

0:15

creators visionaries who

0:17

are creatively applying technology in

0:19

business to drive change, but

0:21

with a focus on the transformative power

0:24

of artificial intelligence and

0:26

what it means to leverage AI as

0:29

a game changing multiplier for

0:31

your business. Today's episode

0:33

highlights the power of collaboration. IBM

0:37

has long been a supporter of the better Together

0:39

mindset and embraces partnerships.

0:42

They have been working together with Salesforce

0:44

for more than two decades, but

0:47

have recently launched a new collaborative

0:49

effort surrounding Generative

0:51

AI. Pushkin's very own Jacob

0:53

Goldstein sat down with Matt Candy

0:56

and Susan Emerson. Matt

0:58

is the global managing Partner of Generative

1:01

AI at IBM Consulting, helping

1:03

clients and partners around the world

1:05

embrace this new era of technology,

1:09

and Susan is a senior vice president

1:11

for Salesforce dedicated to AI,

1:14

analytics and data. They

1:17

discussed the historic collaboration between

1:19

the two tech giants, explored

1:21

the opportunity AI presents for customer

1:24

service and walk through how

1:26

businesses can use generative

1:28

AI to interface with

1:30

clients. Okay, let's

1:32

get to the conversation.

1:36

Thank you guys for coming this morning. So

1:39

I'm interested in how you both

1:41

came to generative AI, or maybe it sort

1:43

of came to you in the way it sort of came to all of us,

1:46

But how did you arrive at working

1:48

on generative AI.

1:49

As part of my REMITTEDT Salesforce. Over

1:51

the years, I've brought a lot of analytics

1:54

and data and machine learning products

1:56

to life under the

1:58

Einstein brand at Salesforce. So

2:01

as we pivoted Salesforce

2:04

into taking advantage of the generative

2:06

AI moment, it was natural

2:08

that I became part of the advanced team leveraging

2:11

generative AI and

2:13

it's become interesting.

2:16

But what I see as I speak

2:18

with customers the moment

2:20

that everyone is facing in terms of how they

2:22

incorporate genitive AI into

2:25

their businesses, their workforces,

2:28

and their technical stacks. It's actually

2:30

opening up a lot of doors to other

2:33

utility of analytics, data and AI.

2:36

So it's been this big pull through in

2:39

terms of incorporating not

2:41

just generative AI, but a

2:43

larger conversation around how we become

2:47

all better using data in our day

2:49

jobs.

2:51

So that's a great frame for sort

2:53

of what's going on at Salesforce with generative

2:55

AI. Matt tell us a little bit

2:57

about how that fits with the way

3:00

IBM is approaching the space.

3:02

Yeah, so I guess through three sides

3:04

to that question. And so there's

3:06

the technology side of it. So IBM

3:09

has a technology organization, and so you

3:12

know, we are building and have

3:14

been over many years, decades. In fact, IBM

3:16

has been working in this space a

3:18

generative AI stack that

3:21

allows organizations to adopt

3:24

generative AI technology aimed

3:27

at enterprise and business use within

3:29

their organizations. So

3:31

then within the consulting business, you know,

3:33

we have one hundred and sixty thousand people who

3:36

work every day with clients

3:38

across every industry, regulated industries,

3:41

government organizations, and

3:43

so this, you know, is a really important

3:45

technology that those companies

3:47

are going to be using to drive the next level

3:49

of transformation in their enterprises processes

3:53

and the types of experiences they build for their

3:55

customers. And so you know, we work

3:58

extensively with partnersechnology

4:00

such as Salesforce, AWS, Microsoft,

4:03

as well as our own technology and then

4:05

finally, I guess the third angle is

4:08

the work that we've got to do to reinvent the

4:10

business of consulting. And so if

4:12

I think about you know, consulting in systems

4:15

integration, you know, ultimately we

4:17

are knowledge workers, right, and

4:19

so from an industry perspective, I think,

4:21

you know, our industry is same as many

4:23

others, is is going to

4:25

go undergo a level of disruption caused

4:28

by this technology. But therefore

4:30

that will also create a huge opportunity for

4:32

us as well.

4:33

So those three aspects, Jacob, great.

4:36

So, so that's the point of view sort of from

4:38

your companies in your work.

4:41

I'm curious to talk for a moment

4:43

about AI from the point of view of

4:46

consumers and employees

4:49

kind of out in the world today. So just to start

4:51

with consumers, when I'm just out

4:53

as a person as a consumer in the world, how

4:56

am I experiencing AI today?

4:59

I give you a great li use case.

5:00

Actually, I was on holiday three

5:02

weeks ago in Tenerif in

5:04

Spain, and I was trying

5:07

to find somewhere to park the car with the family

5:09

for dinner that evening, and

5:11

I found this area

5:14

next to this kind of shopping

5:16

center and there was this sign there and I

5:19

couldn't quite work out if it was saying

5:21

I could park there or not, And

5:23

so I took a photo of the sign and

5:26

I uploaded it to an AI tool,

5:28

and I said, what does this mean? And it basically

5:30

explained to me what the sign was saying and basically

5:32

told me that I shouldn't be parking there.

5:34

And so I drove on and I.

5:36

Found some somewhere else to park. But you

5:38

know, that allowed me,

5:41

in under sixty seconds to probably

5:43

avoid one hundred euro fine

5:46

by parking the car there. So just

5:48

a simple example, but I think the ability

5:51

that these tools have to take friction

5:53

out of our daily lives, you know,

5:55

and to be able to make just

5:58

things that we do in our everyday life simple

6:00

and more friction less. Yeah,

6:02

that's how I look at how

6:04

mat the consumer is going to benefit from some of this type

6:07

of technology.

6:08

And from my perspective, it's also a

6:10

travel story. I spend a lot

6:12

of time on the road for work,

6:15

but recently had to send my

6:18

sister and her family to a destination

6:20

they had never been to for a wedding,

6:23

and it was really quick

6:25

and easy to use some generitive

6:27

tools to come up with a whole

6:29

plan for them because they love to hike and to be outdoors

6:32

and to hike in areas that aren't overly crowded

6:35

with people. And so

6:38

jen Ai very quickly gave me an itinerary

6:40

of a bunch of terrific hikes for them for

6:43

a destination.

6:44

So things like that great.

6:46

And then what about the effect

6:49

of AI and of automation more

6:51

generally on employees on the

6:53

workforce.

6:55

Well, there's so many dimensions to take that from

6:57

generitive AI really can

6:59

up level or workforce in all sorts

7:01

of ways by providing these consistent

7:04

ways to engage with technology,

7:06

with these natural language experiences.

7:08

So I think it changes everything from it

7:11

finds us content, it generates us

7:13

content, It makes it easier to work

7:15

with our systems of engagement

7:17

and operation, and for many

7:20

organizations it can

7:22

be a lifting factor in

7:24

terms of bringing a more consistent workforce

7:27

experience because these tools

7:29

can just be ever present in our

7:31

systems of work.

7:33

I mean, I'll give you a little example here in IBM,

7:35

we have something called our Skajar and

7:38

so that's our conversational AI

7:41

interface that we use to interact

7:43

with HR services, and ninety

7:46

four percent of every employee interaction

7:48

now happens without human intervention

7:51

through that interface, but you would never

7:53

know that. And so if I think about,

7:56

you know, our HR processes,

7:58

you know, we have this amazing conversational based

8:01

AI that we use for all of our HR

8:03

interactions and we surface that

8:06

through SLACK, And so SLACK

8:08

becomes the front door for how we access

8:10

a lot of these different enterprise processes

8:13

and capabilities and how we surface

8:15

AI. In fact, I'm taking a flight shortly back

8:17

to the UK and our our skar boss

8:20

is reminding me that it's raining in the UK and I

8:22

should take an umbrella.

8:24

Isn't it always like raining in England.

8:29

I don't think there's any AI needed for that. I think that's

8:31

just a hard coded If

8:33

England, then take umbrella.

8:35

That's right, that's just a rule.

8:36

That's just a rule, right, and you're able

8:38

to converse. And yeah, I need to book

8:40

holiday, I need to move somebody between managers.

8:43

I need to figure out the policy on this. And

8:46

the AI basically navigates

8:48

across the different systems to be able to

8:50

help get that information, to summarize it

8:52

back, to be able to carry out the transactions

8:55

that I need carried out, and it just removes

8:58

all of that complexity and make it easy

9:00

to get things done.

9:03

When you are working with

9:05

companies to implement generative

9:08

AI, now what

9:11

do you find tends to be their primary

9:14

focus?

9:15

I mean I speak with a lot of customers each week,

9:17

and for the last several months,

9:20

most organizations have just been reorienting

9:23

themselves in terms of where

9:25

are we in this moment, what is this technology

9:28

capable of? What are the risks

9:30

and governance and frameworks

9:32

that I need to establish in order

9:34

to engage and talk to everyone.

9:37

Talk to my vendors, talk to my cloud

9:39

providers, talk to my consultants, talk

9:42

to academics, and generally

9:44

get your sea legs under them. And

9:47

the sort of the unstructured

9:49

hand on keyboards fiddling with technology

9:52

seems to be moving towards let's

9:54

get some points on the board, let's turn

9:57

this stuff on and go. So that's what

9:59

I've been seeing in terms of, you

10:01

know, the work within the salesforce ecosystem.

10:03

Matt, you've got a larger aperture

10:06

as well. What are you seeing?

10:08

Yeah, so I definitely

10:10

agree.

10:11

I think you know, there's been lots of getting

10:14

sea legs experimentation just

10:16

trying to build knowledge, being able to try and build

10:20

almost you know, internal organizational

10:23

point of view and reference framework. I've

10:25

seen lots of what I would have referred to as random

10:27

acts of AI.

10:30

In terms of in.

10:32

Terms of experimentation, but I think I think people

10:34

now looking into twenty twenty four and this is

10:36

all about now adoption and scaling,

10:39

what's become really clear is organizations

10:41

have started to realize this is going to be a very

10:44

multi model world that they're going to live in. There

10:46

is no one AI that is

10:48

the answer for their organization, and

10:50

so they're going to have lots of different

10:52

generative AI models and

10:55

technologies that they're going to sit in the organization,

10:57

servicing different use cases, different

11:00

domain areas, different products and services,

11:03

and so therefore having to figure

11:06

out how they're going to navigate and manage

11:08

this kind of open world that they're going to

11:10

be sitting in and the decisions that they're

11:12

going to have to make around that. I

11:14

think the second thing that I've

11:16

seen that people are now becoming very

11:18

clear that this needs to be what I would refer

11:21

to as use case lead and outcome focus,

11:24

and so really needing to start with

11:27

thinking about the business outcome and the problem

11:29

that you know, we're trying to solve, and

11:32

therefore, how do I use generative

11:34

AI as part of the

11:37

mechanism to solve that problem? And I

11:39

think, you know, what Susan and the Salesforce

11:41

team do is an amazing example of that. You know, they've

11:43

got this incredible platform and engine that

11:46

allows companies to transform

11:48

their sales and service processes and

11:50

to be able to put data in the hands of users,

11:52

to be able to make better decisions, et cetera.

11:55

And so now by weaving generative AI into

11:57

that platform, we're going to be able to make those

12:00

processing workflows even more efficient. Right,

12:02

So it's generative AI plus all

12:04

of these other amazing things that are there, but it will

12:06

be led through business outcome, and

12:08

it will be led through use case and

12:10

the business problem or workflow that we're

12:12

trying to improve. And then I think

12:14

the third thing is shifting from this experimentation

12:17

to scale. You know, I think everybody's really early

12:19

in this journey, but

12:21

what's become clear is that you know, everybody

12:24

now need realizes and is starting

12:27

to lay down these these ground

12:30

rules, the guardrails, the frameworks

12:32

to allow them to scale

12:34

this across the organization. So,

12:37

you know, I think we're in for an exciting, exciting

12:39

time in twenty twenty four.

12:41

So now that we're getting to this moment,

12:44

what are the key things companies

12:46

have to figure out about scaling generative

12:48

AI.

12:51

I would put that in kind of two

12:53

categories and following on what

12:55

Matt was saying in terms of use

12:58

case defined and outcome LAD and one

13:00

hundred percent on that in terms of starting with

13:02

a hypothesis of value. While at the same

13:04

time people are getting closer

13:08

to the technology to know what their bounds are.

13:10

But the biggest, you know, set of conversations

13:12

is in the enterprise area

13:16

in terms of embarking and using with generative

13:18

AI, how to do it in ways that

13:21

is safe for use of data

13:24

that is safe around not just the

13:27

larger topic of generative

13:29

AI and hallucinations, which

13:31

which are fun to talk about in the media.

13:33

But it's a fun word, right

13:35

If it was called something other than hallucinations,

13:37

people wouldn't talk about it as much.

13:39

It was just mistakes, Yeah,

13:41

that's right, just things that aren't factually true.

13:44

We've been doing a lot of work at Salesforce around

13:46

using you know, dynamic

13:48

and structured grounding the data so we can

13:51

give very strong and non

13:53

naive prompt instructions to lllms

13:55

to get return on that. So, just

13:57

to summarize, top of mind for organizations

14:00

using you know, large language models

14:03

is using their data in ways that are safe,

14:05

trusted, not exposed

14:08

and reducing the opportunity for hallucinations

14:11

and maximizing relevant content.

14:13

Great.

14:14

So, so Matt Susan was talking about, you know, both

14:16

what organizations are concerned

14:18

with as they scale generative AI and

14:20

how Salesforce is working to sort

14:23

of address those concerns.

14:25

What are you seeing at IBM.

14:27

Here, So I think certainly from a scaling

14:30

of generative AI perspective,

14:32

you know, this topic of governance,

14:36

you know, and how organizations are going to have to govern

14:38

all of these models that sit

14:40

withinside, how they manage kind of bias

14:43

fairness, model drift. You

14:46

know, if you think about the data that's gone into

14:48

a model and the output

14:50

it gives to start with, not because the

14:52

model changes, but because the context of the

14:54

world moves on. And so being able to kind

14:56

of manage this model drift is going to be a really important

14:59

thing. I think data really matters,

15:01

and so quality access security

15:04

around data within the enterprise is going to be critical

15:07

to scaling generative AI. And the other

15:09

one, I think that's going to be really important, and I think

15:11

many organizations haven't even got there yet

15:13

in their thinking is around the ESG implications.

15:16

So carbon, you know, the use

15:18

of this technology does not come without

15:20

a cost of carbon.

15:22

Carbon meaning it's very energy intensive.

15:25

Correct.

15:25

Yeah, the training of the models

15:28

and so thinking about carbon disclosures

15:30

and thinking about where I'm infusing it

15:33

into my business and how much I'm using

15:35

it and what the carbon cost of that is. As

15:37

I think about the you know, you know, my

15:41

own organizational responsibilities

15:43

to reduce carbon I think, you know, there's all of

15:45

these things that I think are going to become important

15:48

factors as people are thinking about the scaling

15:50

implications of this technology.

15:52

AI is already making new experiences

15:55

possible, but we must be mindful

15:57

in how we integrate this new technology

15:59

as we can continue scaling generative

16:02

AI. Matt touched on some crucial

16:04

aspects from an IBM perspective. Governance,

16:07

bias, fairness, and security are

16:09

all key considerations when organizations

16:11

aim to expand their use of generative AI.

16:15

The environmental aspect is especially

16:17

important, and it's refreshing

16:19

to hear leading thinkers like Matt and

16:21

Susan highlight these issues.

16:24

As this technology continues to evolve,

16:27

these factors are becoming increasingly

16:29

important for organizations to

16:31

address. The Historic

16:34

collaboration between IBM and Salesforce

16:36

is helping to remedy issues companies

16:38

face when scaling AI.

16:42

So, IBM and Salesforce recently

16:44

announced a new collaborative project

16:47

around generative AI.

16:49

Tell me more about that.

16:51

We've been partners for over two

16:53

decades now IBM and Salesforce,

16:56

and so within our consulting business, we

16:58

work with Salesforce technology to help

17:01

our clients implement that technology to transform

17:03

their businesses. We've got a huge

17:05

practice, over twelve thousand

17:07

people with certifications around Salesforce

17:09

platforms, and so you know, as Susan

17:12

and her team and the broader team in Salesforce are

17:14

infusing more capability into the

17:16

platform around generative AI, then

17:18

our mission is really simple. It's to help

17:20

clients who are using the Salesforce

17:23

platform adopt those capabilities

17:25

to help them get more benefit within their organization.

17:28

You know, we're also a significant user of

17:30

Salesforce technology within IBM.

17:32

We're one of.

17:33

Salesforce's largest customers globally, and

17:35

so you know, as we continue to transform

17:37

our own sales and service processes

17:40

within IBM, then you know, our

17:42

use of the generative AI capabilities

17:44

that they're infusing into sales, cloud service, cloud

17:46

slack, et cetera will be something

17:49

that will become really important to us driving

17:51

productivity within the company.

17:53

And then the other thing that I would say is, you know, as

17:55

I think about the work that we do with clients,

17:58

you know, as they're implementing on their generative

18:00

AI journeys, you know they're going to utilize

18:03

and leverage the salesforce capabilities

18:05

within the platform and their generative AI technologies.

18:08

But then you start thinking about processes and

18:11

workflows that run beyond the walls of CRM

18:13

right that run into supply chain and

18:16

into the finance area of the organization.

18:18

And so there is.

18:19

Work that we're doing with clients where we're using

18:21

IBMS. What's the next platform to

18:23

be able to help get

18:25

access to to generate insights

18:28

from data sources that sit in all

18:30

of these kind of back office areas of the enterprise,

18:33

and to be able to get that data across

18:35

the salesforce into these customer interaction

18:37

points and into the employees

18:39

who are servicing those customers using

18:42

Salesforces AI and generative AI technologies.

18:44

So there's a kind of one plus one

18:47

equals three kind of you know, better together,

18:50

you know, and being able to bring our technologies

18:52

together in service of these clients problems

18:56

as you think about these processes that run

18:58

across their enterprise. So yeah,

19:00

so it's so huge hut unity and what we're

19:02

doing together in the market to help clients.

19:05

Yeah, and just building it on that. It is

19:07

a huge moment for organizations

19:10

and for technology companies like Salesforce,

19:12

and we couldn't be happier to have partnerships

19:15

like we have with IBM. Like the range

19:18

of thought leadership that

19:21

is appropriate at the moment is everything

19:23

from what is that hypothesis

19:25

of value and what are those use cases

19:27

and what is the order of operation in

19:29

terms of approaching it just in terms of focus,

19:32

but then things that would help organizations

19:35

assess their AI readiness and

19:38

then their approach, like you know, we talked earlier

19:40

about frameworks and guardrails.

19:42

You know, what are use cases that we're

19:44

comfortable with given the state of the technology

19:47

that face employees or face customers,

19:49

So creating these much larger roadmaps

19:51

in terms of how to approach this over

19:54

a series of initiatives, the

19:56

way it can fundamentally change

19:59

the way we engage with technology

20:01

and what that means for the you

20:04

know, training and change management

20:06

and use cases that fundamentally

20:09

shift how you engage

20:11

with systems like salesforces. There's just a

20:13

massive opportunity for us together.

20:16

So you're talking in sort of general terms,

20:18

I'm interested in, you know, thinking

20:21

in particular about the way generitive

20:24

AI can essentially lead

20:27

to better business outcomes, right, Like, what does that

20:29

look like?

20:29

How do you measure it?

20:31

You know, there's a certain bottom line question

20:33

there, right like how does AI make businesses work

20:35

better?

20:36

And in what ways?

20:37

You know, as consumers of products

20:39

and services, we all love and

20:42

respect great service, you know, in terms

20:44

of getting timely, quick answers, resolving issues

20:46

quickly, all those those types of things. And

20:49

from the perspective of using

20:52

generative and predictive capabilities

20:54

for agents who are interacting with customers,

20:57

there is just a whole ton

20:59

of opportunity you need to take friction out of the process

21:01

in terms of finding answers, resolving issues,

21:04

in terms of using these generative capabilities

21:06

that will bring you know, answers and content

21:09

to the fingertips more easily to

21:11

the human agents that are working

21:13

with customers. Now, taking

21:15

that to the next step for organizations

21:18

when they're ready to move into more customer

21:20

facing automation, that's yet

21:22

another channel. As a consumer, we'll all

21:25

enjoy with the brands and the products and the services

21:27

that we want in terms of fast answers

21:29

and resolutions to customers, and as we

21:31

all know, great customer experience

21:34

yields return business. Now

21:37

on the sales side, you know, maybe a

21:39

different example, and these

21:41

are areas where I think the capability

21:43

of predictive and generative go very

21:45

well together in terms of focusing

21:47

on business outcomes. And a classic

21:50

example would be, you know, predictions

21:52

that help us understand customer health.

21:55

You know, is this customer engaged, is this

21:57

customer at risk? Predictions

22:00

that help us understand next best

22:02

product or next best conversation. These

22:04

all help focus sales

22:07

team's time on

22:09

a customer or a territory, and

22:11

so that deep focus puts

22:13

all the wood behind an arrow, so to speak, in

22:15

terms of where we should be engaging,

22:19

and those types of driven sales

22:22

organizations that have these capabilities

22:25

just lead to better performance and outcomes

22:27

and customer experience too. Now

22:30

let's also layer in generitive capabilities

22:33

where we're using the generative capabilities

22:36

to assist and augment a sales team,

22:39

where we're using the power de generitive for everything

22:41

like generating personalized

22:43

and relevant customer interaction content

22:46

for example, leveraging our customer

22:48

data like engagement history, product

22:51

purchases, service history to

22:53

create an email or a campaign, and

22:55

this scale of automation has just never been

22:57

possible before. Maybe

23:00

even taking this one step further re genitive,

23:02

where we take all the administrative friction

23:04

out of the day job and doing things

23:06

for sales teams like summarizing their calls

23:08

or creating a meeting plan for them, and

23:11

you know, very broadly speaking, using

23:13

generative AI to change the interaction mode

23:16

with systems like Salesforce from

23:18

clicks and training where

23:20

people have to focus on the process to more

23:22

conversational user experiences

23:25

which are much more engaging and easier

23:27

to use. So all of this together is

23:30

just incredible and transformational

23:33

and makes all businesses and people

23:35

work better.

23:36

So I just want to spend one more moment

23:39

on the partnership between IBM

23:42

and Salesforce and genitive AI. And

23:44

there's this phrase that's interesting to me. It's

23:47

ecosystem partnership that

23:49

I think is relevant here. So what

23:51

is an ecosystem partnership and why

23:54

is it you know, helpful in creating

23:57

scalable AI solutions.

23:59

Idea of being open, I

24:02

think is probably one of the most important

24:05

premises for US as technology

24:07

companies, for US as consultancies

24:10

and system integrators, and for our clients. To think

24:12

about the sources

24:15

of value that can be created through taking

24:17

an open approach is hugely

24:19

important. So if I think about for

24:21

US, ecosystem means making sure

24:23

that we have all of the

24:26

different partnerships that we need with

24:29

technology providers, with service

24:31

providers, that we can bring

24:34

to our clients the right

24:36

set of capabilities to solve the problem

24:38

that they've got, and not thinking that just

24:41

you know, what we have in house, or

24:43

what we have with just one other partner that we

24:45

work with, you know, is the right thing. And

24:48

so you know, I think every problem that

24:50

our clients have is solved through a range

24:53

of technologies that come together in

24:55

service of creating that business outcome.

24:58

I want to.

24:59

Touch briefly on ethics

25:01

and governance. Something

25:04

like eighty percent of CEOs see

25:06

explainability, ethics, bias,

25:08

trust as major concerns

25:11

on the road to AI adoption, and

25:13

so I'm curious how

25:16

business leaders navigate

25:18

these things, and in particular, how Salesforce

25:20

and IBM are building these

25:22

concerns into how

25:24

they work with customers.

25:26

You know, we've been incorporating

25:28

predictive machine learning into our

25:31

products since mid last decade, and

25:33

at that time we started

25:35

with all of our ethics and governance

25:38

work at that time in terms of frameworks

25:41

for engaging with AI in ethical

25:43

and safe ways and have

25:45

a lot of guidance for customers in terms of those

25:48

programs. The machine learning

25:50

focus that we've had at Salesforce has always

25:52

been deeply focused on explainability.

25:55

So if we're making you know, predictive

25:58

recommendations to explain how

26:00

we got to that, you know, whether that's

26:02

something that a user sees, is

26:04

they're engaging with it so they have full trust

26:07

in terms of interacting with it, but

26:09

also for the practitioners who

26:11

are building it. So we have this like

26:13

long standing vibe and capability

26:16

with our predictive side of the

26:18

house and on the generative side

26:20

of the house. You know, the state of the marketplace

26:23

right now is llms for most

26:25

people are largely black

26:27

boxes in terms of not

26:29

fully interpretable in terms of how they come up

26:32

with their content. Now that said,

26:34

there is a lot that you can do in

26:36

terms of audit in terms

26:39

of you know, transparency in terms

26:41

of what are the prompts that are being

26:43

submitted to these llms, what

26:46

do these llms provide back

26:48

in terms of return, and then what

26:50

did the human do to change it, use it, or

26:53

adjust it. So we've been updating all

26:55

of our ethics and governance frameworks.

26:57

Now I guess I would call it with safety

26:59

compuls it as well in terms of how

27:01

to work with data in safe

27:04

ways and with these turned parents governance

27:06

models.

27:07

Yeah, so, I mean this is an area that IBM

27:09

has been kind of working on for many years.

27:11

And so you know, our AI Ethics Board

27:14

that we have internally kind

27:16

of governs and provides frameworks

27:18

and guidance for everything that we do in the company.

27:21

There's a lot of work that we do to help our

27:23

clients and organizations establish

27:25

their strategies for AI governance as

27:28

well as their own internal policies, models,

27:31

approaches, ethics boards, et cetera.

27:34

And so, you know, helping them put in

27:36

place these ground rules and guardrails,

27:39

organizational process changes,

27:42

et cetera. I think is a really important

27:45

part of this scaling discussion that we were having

27:47

earlier, as people are going to be kind of

27:49

rolling out more of this technology internally,

27:52

and then I think there's a

27:54

lot that organizations are going to have to do to

27:56

think about, especially in the generative world, around

27:59

all the different types of models that they're using,

28:02

models that they're training and

28:04

tuning and building, and how they manage

28:06

all of those for explainability and bias

28:09

drift, and actually regulatory requirements,

28:12

Like if you think about what's

28:15

happening around the world, there's different countries,

28:18

the EUAI Act, you know, there's lots

28:20

of different regulatory requirements that are

28:22

going to be coming in, and so for multinational

28:24

companies operating

28:26

across multiple countries, how

28:29

they're going to have to make sure that

28:31

they're complying with all of not

28:33

only their own internal policies, but

28:36

the requirements of the country

28:39

as well as potentially industry

28:41

regulatory requirements as well.

28:44

And so there's a lot that we are.

28:46

Doing and going to be doing in helping

28:48

them manage complexity. But

28:50

IBM has a very firm view that we

28:52

believe that this is all about regulating AI

28:55

risk, not AI algorithms,

28:57

and so focusing on precision regulation

29:00

so you know, use

29:02

the bodies and regulatory bodies that are out

29:04

there to provide the control

29:08

as opposed to trying to regulate the technology.

29:10

So genitive AI is changing

29:13

kind of absurdly quickly. Right, a year

29:15

and a half ago, we wouldn't have been having this conversation.

29:17

We're here today. Everything's happening now.

29:19

I'm curious what.

29:20

You both think about about the near

29:22

term future of genitive A. Right, if we

29:24

came back in a year or let's

29:26

say two years from now, if we came back two years

29:28

from now to talk about the work you're

29:30

doing in genitive AI, what would we be talking

29:33

about.

29:35

I use this example sometimes I

29:37

have three kids, and I

29:39

don't think any of them have

29:42

ever gone into a bank to deposit

29:44

a check. Right, they pull out their mobile

29:47

phone and they scan the check

29:49

with the camera and they're done.

29:50

I'm surprised that they even know what a check is,

29:52

for the record, But.

29:53

Yeah, well yeah, sometimes

29:55

their parents give them one, like

29:58

they get direct deposit. But anyway,

30:00

like this experience of like, what

30:02

do you mean I go into a branch

30:05

in cash a check. I just do this with my mobile

30:07

phone. And I think a little bit of

30:09

it that way. In terms of the systems that we use

30:11

at work, I can imagine explaining

30:14

to my kids like, oh, yeah,

30:16

at Salesforce, you know, back when someone

30:19

had their first day on the job, you know,

30:21

as a service agent or as a salesperson,

30:24

they would have tabs on the screen and

30:26

they would be trained where to click, and they'd

30:28

have documented processes in manuals

30:31

and that showed them where to get

30:33

from point A to point B. And

30:35

as the clock turns forward, they're

30:37

just interacting with a natural language

30:40

prompt. But it just kind

30:42

of fundamentally changes the

30:44

way we'll be able to interact with our

30:46

systems a record at work.

30:48

It'll be just much more conversational. Instead

30:50

of clicking through something, you'll just basically

30:52

have a conversation.

30:54

Much more conversational.

30:55

Yeah, this is the biggest paradigm shift in

30:57

how we interact with technology, I think since

31:00

invention of the graphical user interface, and

31:02

it's going to enable us to almost

31:05

put aside all of that complexity within

31:07

organizations around system silos,

31:10

process silos, flows, because

31:12

you're just going to layer this just simple

31:14

natural language interface over all of

31:16

that complexity.

31:18

Yeah, it's going to amplify.

31:21

I think the potential of every person

31:23

on every team in a way that we've never

31:25

been able to see before. And the

31:27

other thing that I think as you project forward

31:30

in a couple of years and Susan just picking up on the

31:32

point that you talked about about banking, you

31:34

know, I.

31:34

Think there's a wonderful little example.

31:36

Look, if you think back to the seventies and the eighties

31:39

when the ATM kind of cash machines

31:41

were rolling out, and at that time

31:44

it wasn't really a reaction

31:46

that was one of awe or appreciation for convenience,

31:48

but people were concerned that we were automating

31:51

away the bank teller jobs.

31:53

Right.

31:53

But now, when you think about it, what

31:56

actually happened was this technology

31:58

allowed the banks to scale their

32:00

branch networks, more branches never

32:03

before, more bank tellers than ever before.

32:05

Bank teller employment and salaries increased,

32:08

even though we automated them out of work,

32:10

because when they weren't having to spend

32:12

their time counting cash out for

32:14

people, they were able to do more valuable things, right,

32:16

and new types of financial products and services

32:19

and mortgages and so like. If I think

32:21

back to that in the seventies and eighties,

32:23

and then I project to where we are today, we're

32:25

just going to unleash this creativity and potential

32:29

for employees and enterprises by freeing up

32:31

the time that they're spending on things that

32:33

you.

32:33

Know, they can do far more value added tasks.

32:36

And so I think we're going to be amazed I

32:38

think around what happens and what companies

32:40

and people are going to be able to do as we give them

32:43

the time and space to be able to do that. Great.

32:46

So, just to close, I want to

32:48

talk about how both of you use

32:50

creativity in your own work.

32:53

Just to start with you, Matt, I know that you

32:55

love to combine creativity and technology

32:58

through design.

33:00

Do you use generative AI in your own

33:02

creative process?

33:04

Yeah.

33:04

So I'm a firm believer

33:07

that this combination of experience

33:09

in AI is going to be the thing that

33:11

makes a difference. Like these large language models,

33:13

and this technology has been around actually

33:15

for a number of years, and it's only

33:18

at the point late twenty twenty two

33:20

where open ai wrapped a digital

33:22

experience around this and put it in the hands

33:25

of people that suddenly the transformative

33:27

power of this technology

33:29

was realized. And so I think the way that we surface

33:31

these capabilities and put

33:33

them in the hands of people to

33:35

be able to adopt it in a really frictionless

33:38

way is the thing that's going to be hugely

33:41

important to the adoption and scaling of this. So

33:43

I think the most important thing

33:45

for companies to do is to make people,

33:47

not technology central to their strategy.

33:50

Just to go more broadly into

33:52

your work, Susan, I mean, I know that

33:55

you have launched salesforce as AI

33:57

products into the market, and that you

33:59

know a lot of those have been built obviously

34:01

given salesforce business around helping

34:04

people build stronger customer

34:06

relationships, right, and so I'm curious what

34:09

creativity did you bring to that work.

34:11

Some of the products that I've worked with a salesforce

34:14

there, they're deeply visually focused.

34:16

And my personal perspective is

34:18

is that the world can be really

34:20

noisy. We're just inundated

34:24

with all sorts of demands on our time

34:26

through so many channels, right, Like the

34:29

phone is firing off, you're getting instant

34:31

messages, you're getting slack messages, you're

34:33

getting you know, DMS, you're getting

34:35

emails, your phone is ringing, There's

34:37

processes that are bearing down on you. And

34:40

if we can use really good design to

34:43

filter out and essentially weed the

34:46

garden, because you know, we have this

34:48

this phrase at Salesforce is everything. If everything's

34:50

important, nothing's important. So using

34:53

really good design to create

34:55

the user experience in Salesforce,

34:58

that just brings stuff to

35:00

life in the most powerful way. So

35:02

I always think of it from that perspective, like,

35:04

if I'm going to put this on a screen and

35:07

Salesforce, what did I not

35:09

put on? Is this the most important

35:11

thing? And is this the thing that's going to

35:13

align everyone to the larger initiative

35:15

of the firm. So it's that kind

35:17

of design thinking that I use

35:20

probably every moment of the day, whether

35:22

I'm building a demo or talking

35:24

to an executive as a company in terms of

35:26

as I see a vision for how they might deploy our

35:28

products to actual product

35:30

development.

35:32

Just to kind of bring.

35:33

Together these two themes we've been talking about,

35:35

on the one hand, the sort of ecosystem

35:37

partnerships and on the other hand

35:40

creativity. I mean, can you talk a

35:42

little bit about how work

35:45

working with partners can foster a

35:47

different kind of creativity.

35:49

More perspectives are always better than few

35:52

perspectives.

35:53

I completely agree.

35:54

I think the more minds, the more perspectives,

35:57

the more experiences. You

36:00

know, if I think about some of the best sessions,

36:02

best workshops, best work we do

36:04

with clients, it's when you've got people

36:08

not just from one industry, but

36:10

from many industries, because actually the

36:12

adjacencies and the things

36:14

that are happening in these other spaces trigger

36:16

new thoughts and new ideas, and so, you

36:19

know, I think the richness that we get when

36:22

we partner with Salesforce together around

36:24

helping clients transform their front office,

36:26

their sales service marketing processes.

36:29

We all bring these unique experiences, and

36:31

I think that just opens the aperture to

36:33

better outcomes and better

36:36

perspectives for our clients.

36:38

Well, you know, you've been asking these questions about

36:40

like the use of tech and AI

36:42

and creativity are sort of in the same sentence.

36:45

And one of the things that I also think of

36:47

is in terms of remaining

36:49

deeply creative, is the

36:52

actual process of unplugging from

36:54

all that stuff. So taking

36:57

a trail run with note earphones in

36:59

your head for me is

37:01

always a really good way of

37:05

unleashing and unbridening a lot of you know, creative

37:07

spirit. Just that downtime

37:10

and the unstructured time where your brain can just

37:12

run free, actually not assisted by

37:14

any kind of device in my head

37:16

or in my face.

37:17

So I think with.

37:19

That praise of unplugged

37:21

time, we should say goodbye and let's unplug

37:23

it. It's lovely to talk with you, guys. It was really interesting

37:26

to learn about your work and the relationship between the company.

37:28

So thank you for your time.

37:30

Thank you, Jacob, thank you.

37:33

A huge thanks is due to Jacob, Matt and Susan

37:35

for illuminating the possibilities

37:38

of generative AI. This

37:40

technology has great promise for creating

37:43

new experiences in the future, but

37:45

requires the scaling capabilities

37:47

made possible by partnerships like

37:50

IBM and Salesforce. As

37:53

our conversation with Susan and Matt illustrated,

37:56

we're at an exciting phase of adoption.

37:59

Most companies have moved beyond experimentation

38:02

and are now prioritizing scaling.

38:04

The key areas of focus for organizations

38:07

now include managing multiple AI

38:09

models, as well as thinking about

38:11

specific use cases and desired

38:14

outcomes. However, this scale

38:16

is difficult for companies to do on

38:18

their own. To unlock the real potential

38:20

of generative AI in transforming

38:23

experiences, they'll require the

38:25

scaling capabilities made possible

38:28

by partnerships like IBM and

38:30

Salesforce. This conversation

38:32

showed the promise of teamwork. When

38:35

massive companies combine their brain power

38:37

to push forward technology, their

38:39

collaborative efforts have the potential

38:42

to revolutionize industries. One

38:46

quick programming note, we will be taking

38:48

a little time off and will be returning

38:50

in just a few weeks with a new

38:52

episode. Smart Talks

38:54

with IBM is produced by Matt Romano,

38:57

Joey fish Ground, David jaw and

39:00

Jacob Goldstein. We're edited by

39:02

Lydia Jane Kott. Our engineers

39:04

are Jason Gambrel, Sarah Bruguier

39:06

and Ben Tolliday. Theme

39:08

song by Gramoscope. Special

39:11

thanks to Andy Kelly, Kathy Callahan

39:14

and the eight Bar and IBM teams,

39:16

as well as the Pushkin marketing team.

39:19

Smart Talks with IBM is a production of

39:21

Pushkin Industries and Ruby Studio

39:23

at iHeartMedia. To find more

39:25

Pushkin podcasts, listen on the

39:28

iHeartRadio app. Apple Podcasts

39:30

or wherever you listen to podcasts.

39:33

I'm Malcolm Gladwell. This is

39:35

a paid advertisement from IBM.