0:07

Welcome everyone to the AI and

0:09

Business Podcast. I'm Matthew D'Amelio, Senior

0:11

Editor here at Emerge Technology Research.

0:14

Today's guest on the program is

0:16

Steven Zhang, Managing Director at Deloitte

0:19

Consulting. Steven joins me

0:21

on today's podcast to discuss AI's

0:23

impact on life sciences IT workflows.

0:26

He introduces this episode by

0:29

offering a four pillar framework

0:31

for rapidly adopting generative AI

0:33

for life sciences enterprises. He

0:35

then explains the sector specific

0:37

talent needs of adoption teams,

0:39

underscoring the responsible AI practices

0:42

and executive buy-in tactics that

0:44

lead to effective digital transformations

0:46

in almost any industry. Today's

0:49

episode is sponsored by Deloitte and

0:51

without further ado, here's our conversation.

0:59

Steven, thank you so much for being with

1:01

us on the program today. Glad to be

1:03

here, man. Absolutely. There's so much to dive

1:06

into. It's a very broad topic in pulling

1:08

apart the AI transformations that we're seeing

1:11

take place, especially at this point of

1:13

AI adoption, generative AI making

1:15

a big splash within the last

1:17

year. Obviously, this is a big

1:20

initiative, especially for IT teams,

1:22

and we'll get into that shortly. And we've,

1:25

especially on this program, have covered so

1:27

much in drug development, in

1:30

commercial and clinical trials, but I

1:33

feel like it doesn't get that much attention. Maybe we could

1:35

start you work in this space up

1:37

close and personal. AI is really

1:39

becoming a part of the conversation

1:41

throughout. What are the

1:43

challenges, the trends in these areas,

1:46

the overall situation that's making AI

1:48

relevant for folks in IT in

1:50

the lifecycle? Yeah, that's a great

1:52

question that I think specifically knowing

1:55

this show and the theory for

1:58

The life science Industry. Drink.

2:00

And. Maybe just cleaning out

2:02

the broader industry trends would be

2:05

important as a weapon or asylum,

2:07

or obviously drug pricing. Economy

2:10

and inflation. In Westman

2:12

Innovation and a Workforce and Talent or

2:14

the broader industry trends are the specifically

2:17

for technology. I think they're assuage for

2:19

hims. has emerging. The

2:21

first one being increasing implementation

2:23

and integration of a I

2:25

had a Jet I into.

2:28

The. Business Process He is

2:30

a second method friends is

2:32

really around strengthening the technology

2:34

foundation such as infrastructure. Beta.

2:37

Application Cyber Security and also

2:39

work for some Palins. Or

2:42

because in the another day without

2:44

those technology foundations you can relate

2:46

harness the power of genetic or

2:48

ai. Yeah, an estimated metered friends

2:50

which is a long state Fans

2:52

is. Gonna. Be a worked for

2:54

some talent Now we have seen a rise

2:57

of the When A Man or Engineering Spare

2:59

Funds. And. And also in

3:01

in this day and age learning and

3:03

development of continuously upskilling you or your

3:06

Nixon's. Obviously it's gonna be are

3:08

important for you to leveraging Ginia So

3:10

Stephen I used especially in light of

3:12

all of these trends I think something

3:14

is that time he gets lost in

3:17

the discussion is we tend to ask

3:19

the question you know what do you

3:21

need on the infrastructure side to really

3:23

prepare for generative ai and that has

3:25

a lot of assumptions built into it

3:27

right? Like Enad see it assumes usually

3:30

organizations already started an ai you kind

3:32

of need to build generally have a

3:34

I've since the first generation capabilities and

3:36

machine. Learning and predictive analytics you

3:38

need that grumpy six. But I'm

3:41

wondering if you can provide a

3:43

a framework that might work for

3:45

a broader range of organizations more

3:47

no matter where they are in

3:50

their Ai adoption process of better

3:52

road map to understand where to

3:54

go from zero to really leveraging

3:57

these generative a i use cases

3:59

effectively. Yeah. The

4:01

Leprosy. The question that I think there are

4:03

four. Pillars. And part of the

4:05

free market term. so. Rapidly adopting

4:07

Jenny Iron and A for

4:10

enterprise. Atlas Us one

4:12

is really around establishing a tad Cpt

4:14

a like to read it with the

4:17

enterprise knowledge and the security safety and

4:19

so that you can put in there

4:21

in the hands of older employees. At

4:24

Deloitte we have Para D and

4:26

the Side Kick had making the

4:28

news where we have Hundreds of

4:30

southern employees have started using Jenny

4:32

I technology of within Be Like

4:34

Already. A second fillers

4:36

really around understanding that Technology

4:39

Foundation a spoke about earlier.

4:41

Or lot of enterprise apps will

4:43

be rolling on Weise. Putin.

4:46

They are building regardless whether or

4:48

not it's or your pieces them

4:50

for Crm system of commercial data

4:52

warehouse or you are a real

4:54

what evidence they don't database all

4:56

these at the price obligation will

4:58

comes in wheaton. I blend the

5:00

in. it's really a pound. Id

5:02

leaders to figuring out. What? Exactly

5:04

is gonna be the adoption and

5:06

it's engine management aspects of. Lending

5:09

those capable of the into the

5:12

existing ecosystem you are the third

5:14

pillar in interns. Oh adoption these

5:16

two Really thinking about establish a

5:19

platform. For. Rapid Jenny I

5:21

Innovation, Within. The Lloyd We

5:23

have a Jeanette studio where. We.

5:25

Have experimented different capabilities using in

5:28

the Eye to the extent where

5:30

we can create a powerpoint slides.

5:33

Which. Are trading that for consultants? Automatically

5:36

using genetic managing like my neck

5:38

thinking of navy. couple years later,

5:40

it'll be our car sitting here

5:43

talking to you on a podcast.

5:46

It's into the hall fact that innovation

5:48

has been going on. And lastly, Basically.

5:51

For the digital leaders I think it's

5:53

really applying to in the eye for

5:55

eye teeth farm. understanding.

5:57

And leveraging the tools for converting

5:59

design into. Code. To. Generating

6:01

code automatically and also thinking about

6:03

using genetic analogy to and

6:05

they both it other steps in

6:08

the software development life cycle at

6:10

at creating task, the data, quitting

6:13

tested cases and also completing

6:15

documentation. So these four pillars I

6:17

think provides. A good framework

6:19

in terms of regardless where your

6:21

journey is in in the last

6:24

night's enterprise. what are the key.

6:26

On Constituents as as a leader you

6:28

can sink out of adopting a I

6:31

get our rapidly with the organizations are

6:33

I want to bring up a couple

6:35

things you eat you brought up in

6:37

your last answer and talk about talent

6:39

and it's making it's way to the

6:41

ground floors it to to everyday people

6:43

they're getting some familiarity with this he

6:45

in terms of the I T talent

6:48

not just the difference between generative A

6:50

I as a technology versus the forums

6:52

where as I like to put it

6:54

the first generation capabilities of A I

6:56

we. Sauce On, you know, win

6:58

them. Over the last decade, get

7:01

the more deterministic technologies as they're

7:03

called machine Learning, Predictive Analytics, or

7:06

Optical, all Optical character recognition. You

7:08

know that be I T talents

7:10

that you needed to really drive

7:13

those capabilities has to be difference

7:15

then the Id talent that you

7:18

need to drives probabilistic technologies like

7:20

generative A I'd technologies that are

7:22

guessing at the answer. Needs that

7:25

human feedback to really verify. What

7:27

they're doing. Some I'm wondering, especially

7:29

for Id leaders. For a larger

7:31

management, how should they think about

7:33

the human part of the infrastructure

7:35

and what's the difference between the

7:37

talent they were getting or they

7:39

should be looking for up until

7:41

this point and what will change

7:43

as generative A I becomes more

7:45

part of the conversation. It.

7:47

As met that's that's a important lesson

7:49

and eight the million dollar question to

7:51

answer. I think that let me just

7:54

take a doubt that like to prospective

7:56

one. Specifically. some a

7:58

talent spend on what we have

8:00

observed within the industry and what we have

8:02

been doing as part of Deloitte. But

8:05

also the second part is how the

8:07

talent also kind of interweave

8:09

the benefits and risks of adopting

8:11

Gen-AI. So first of all, from

8:13

talent perspective, you're 100% right

8:16

that with an event of Gen-AI,

8:18

there are new roles and new

8:21

skill sets have been emerging, right,

8:23

and a prompt engineer may headlines

8:25

in terms of as a new

8:27

profession that will garner a lot

8:29

of interest across the industry. There

8:32

are also existing roles all of a sudden

8:34

finding them in higher demand. For example, a

8:37

lot of the first generation of the

8:39

Gen-AI capability are being built in the

8:41

form of application, where you

8:43

will need actually a large team of

8:46

application developer, as well

8:48

as a DevOps engineer to making

8:50

sure those applications can be scaled

8:52

to the enterprise scale and satisfying

8:54

the needs for not just the

8:56

one department, but employees coming from

8:58

globally, right? And also there are

9:00

existing roles now with

9:02

new responsibilities. For example, you

9:05

touched upon responsible AI and

9:08

we are seeing the responsible AI leaders

9:10

within the Life Science Organization now

9:13

have to wrestle with the

9:15

Gen-AI aspects of the AI

9:17

to making sure the solution

9:19

is unbiased and having like

9:21

the right security privacy baked

9:23

in. And also a lot

9:26

of the, I would say, traditional

9:28

NLP or natural language processing

9:30

data scientists also finding

9:33

themselves now have to really

9:35

upscaling on Gen-AI and being able to

9:38

leverage in Gen-AI for some of the

9:40

natural language processing tasks we have been

9:42

doing, right? But I think a

9:44

lot of the times, easy answer is, okay, I

9:46

would just figure out what

9:48

is the learning and development of the program

9:51

and then just upscaling

9:53

my talent. Lloyd

9:55

is approaching this slightly differently. We're

9:57

thinking about this is a broader

9:59

change. Management: Ah starts having

10:01

fun. In a from a

10:03

top perspective we first degenerated

10:06

a lot of awareness inside

10:08

of the organization. For. Example:

10:10

We have. Tried. To smuggle

10:12

modernization passport program that making sure

10:14

all the senior leaders are where

10:16

are one of the technology as

10:19

bring to bear and a what

10:21

we are doing across of from

10:23

using gently I. And. Second

10:25

lane from from generating

10:27

insists. We. Have been running

10:30

spotlights and pop up demos

10:32

across in a virtual space,

10:34

as was our offices. To.

10:36

Really sides the team in terms of. What

10:39

they can achieve by leveraging Deny.

10:41

An aneurysm assert some point be are

10:43

also quitting training for a to live.

10:46

The. Making sure we keep the broader

10:48

in a staff and and broader

10:50

workforce. We the knowledge I'm Jenny

10:52

I and A were also putting.

10:54

Tubes. Of genie out in employees

10:56

hand like I said the sidekick and

10:59

thirty and and last night we are

11:01

also continuously mean this was thing. The

11:04

message: by using A I've had a

11:06

Me as well as our A I'd

11:08

never censored to making sure we continuously

11:11

educating of the broader team really been

11:13

able to stay on the forefront of

11:15

genetic knowledge. So that's what we're doing

11:17

from a talented prospective saw really quick

11:20

question that just on responsible A I

11:22

because I definitely think in the advent

11:24

of generative A I this is moved

11:27

from buzzword to a real tactical discipline

11:29

said that You see where we've had

11:31

a lot of conversations with folks. Talking.

11:34

About responsible Ai and the more

11:37

that we get into the practices

11:39

that work in really prevent a

11:41

lot of the scenarios that people

11:44

can look up in the news

11:46

is it's really more hiring the

11:48

right data governance people in also

11:51

having a management that's educated enough

11:53

on data governance in the relationship

11:55

to artificial intelligence developments that they

11:58

understand and that if you're cutting

12:00

corners in your data governance he

12:02

can lead to the kind of

12:05

biases that lead to some headlines.

12:07

I'm just wondering in terms of

12:09

integrating. You know where the talents

12:12

that speaks to making risk real,

12:14

responsible A I a priority in

12:16

the organization? Where do you wanna

12:19

look in terms of the skills?

12:21

Is it more data governance or

12:23

is it something a little bit

12:26

more cultural within the organization? I.

12:28

Think it's a little bit of both

12:30

mans. I think I'm even in the

12:33

age of Jimmy I that beta you

12:35

to train those foundational luck somewhere tomatoes

12:37

as well as the data you. And

12:40

they voted Ai to the interest

12:43

and of them that arise and

12:45

eventually generating responses using food and

12:47

design pattern such as generated a

12:50

response Argumentative Center resubmit. It's really

12:52

important without a clean data for

12:54

printing that was with without a

12:57

clean data. To. Relate unable

12:59

to retrieval argument a generation that

13:01

response you're getting from Jenny as

13:03

others there's gonna be as bad

13:05

when. Yeah But in addition to

13:07

that the focused on did a

13:09

quality and that a governess. It

13:11

is also importance of focusing on

13:13

helping the should they I moto

13:15

to be the in. The Ota

13:17

weeds privacy, transparency and making sure

13:20

they're in that way to enable

13:22

They explain ability and sadness and

13:24

in terms of those marks on

13:26

with models I think that is.

13:28

A special type our resources

13:30

and require special type of

13:32

framework to be able to.

13:34

Connecting. The dots with the data

13:36

but also as you're building the model finds

13:39

in the bottle were using. Retrieval.

13:41

Argument and generation pattern to

13:43

generated content on long the

13:45

entire pipeline. There's. A clear

13:48

understanding and and the clarity terms

13:50

of data. Yeah. absolutely

13:52

and and i know in terms of

13:54

life sciences and then on to the

13:56

adjacent sectors like health care that you

13:59

know integrating human expert

14:01

feedback, keeping experts in the loop

14:03

is going to be hugely important.

14:05

I know that's kind of underneath

14:08

the surface of the responsible

14:10

AI conversation because it's a little

14:12

bit more in the model development

14:14

area. It's kind of being considered

14:16

as, especially as folks and organizations

14:19

adopt generative AI, kind of the givens of

14:21

what you need to do to really

14:24

launch these tools effectively. But I'm

14:26

wondering if you can touch on

14:28

a bit in terms of the

14:30

talent that you'll need to hire in

14:32

IT to really prepare to keep

14:35

those human experts in the loop and who needs to

14:37

be at the table to emphasize

14:39

what expertise needs

14:42

to be in that loop to

14:44

verify these answers, especially for the

14:46

more client-facing solutions and models that

14:49

I know that organizations ultimately want to get

14:51

to in their generative AI goals. Yeah,

14:54

I think at the highest level

14:56

within the organization, there need to

14:58

be a leader that's really minding

15:00

the broader adoption of the AI

15:03

across the different disciplines. So

15:05

an overall responsible AI lead

15:07

is really essential to define

15:09

the discipline as well as

15:11

the framework for all

15:14

the different utilization of AI to

15:16

follow. I think that's the first

15:18

role. The second role is really

15:20

around with so many different applications

15:22

and so many different use cases

15:24

adopting AI to

15:26

the point I made earlier. What

15:28

is the value has been realized?

15:31

How many people are really using

15:33

those GenAI in field applications or

15:35

AI in field applications? So innovative

15:37

companies start thinking about establishing a

15:40

role, really focusing on narrowing

15:42

the utilization as well as the

15:45

value generated out of those GenAI

15:47

solutions. We call it like a

15:49

value tracking and a

15:51

generative officer. And the third

15:53

role is really going to be around GenAI

15:56

and AI such a faster

15:58

moving field. All. The

16:01

company will need or scanning mechanism.

16:03

A farewell to to keep the

16:05

past on what is a legacy

16:08

industry trends, what is the hyper

16:10

scanners and and abroad or ecosystem

16:12

vendors are is awaiting using Jenny

16:15

Iso having a external innovation. Lead

16:17

would be central to kind of keeping the

16:20

pass on the market and the make you

16:22

for your not adopting or customize the building

16:24

certain technology that is already gonna be rolling

16:26

out by another vendor. Out of the bass

16:28

though these are the three rows not. I

16:30

think it's can be crucial for the company

16:32

to consider. Absolutely we. I really appreciate you

16:35

laying out like you know the specific titles

16:37

and skill set that they need to hire

16:39

for. Think that's really going to clear for

16:41

our audience and that's that covers that side

16:43

of the table in terms of you know

16:45

those books Listening Horsey, I was out. There,

16:47

how did they build this within their

16:50

own organizations The other half of the

16:52

conversation you were talking about leadership before

16:54

is how are you selling the rest

16:56

of the see sweet the rest of

16:58

management on this and this has been

17:00

a conversation that is really taken a

17:03

one eighty even from when I first

17:05

started here about a year and a

17:07

half ago. But I think it's definitely

17:09

it's a completely different colored than even

17:11

when it was half a decade ago

17:13

In that's you know, it's it Sees

17:15

Meters Ceos leaders were is skeptical about

17:18

ai. They really hadn't known it in

17:20

their own personal lives so was always

17:22

an uphill battle or to really sell

17:24

the stuff now especially in the advent

17:26

of open A I'd this really having

17:29

a bigger moment in the culture you

17:31

know, Ceos their kids are using it's

17:33

I think there's still a lot of

17:35

education out there, even among business leaders

17:37

that needs to be done. or just

17:39

in terms of those capabilities, not just

17:42

in terms of the symptoms that you

17:44

Cbn hallucinations, misinformation button. Why Why do

17:46

we get there and. i'm wondering you

17:48

know what advice that you that you have now

17:50

that those tables have turned now that it's more

17:53

likely to you know the ceo is going to

17:55

bring up haywire with you know where are we

17:57

on on jan van they're going to be driving

17:59

the conversation. They don't need to

18:01

be convinced about the

18:03

destination, but they do need to

18:06

probably have some conversation about realistically

18:09

how to get there. And I'm wondering,

18:11

you know, how have you seen those

18:13

conversations change and what advice that you

18:15

have for especially IT leaders in terms

18:17

of conducting them and getting them in

18:19

the right direction? Yeah, thanks

18:21

for the question Matt. I think we are all

18:23

known now, now it's 2024. The genie is

18:27

out of the box in terms of

18:29

genie and it's rapidly blasting through the

18:31

hype cycle faster than any technology

18:34

innovation before. Yeah. I think if

18:36

I can offer just two suggestions

18:39

to the leaders, one is around

18:41

the benefits and risks balancing, the

18:44

second is around looking through the hype

18:46

on genie. So first on like

18:48

the benefits and risks, I know there are a

18:50

lot of overblown anticipated

18:52

value as well as over

18:55

cautious stance in terms of

18:57

adopting genie. I think both

18:59

of these events are stemmed

19:01

from lack of experience and understanding.

19:04

I think from a value

19:06

standpoint, leaders need to set aggressive goals

19:09

to create the motivational poll for

19:12

ease or hard team to start

19:14

really pushing the envelope of adopting

19:17

the technology, right? And then

19:19

the value can be realized through speed

19:21

experience and the cost that can be

19:23

taken out. And from a risk standpoint,

19:26

it's really important to have a

19:28

framework to start

19:30

safely experimenting and building

19:32

institutional understanding of genie

19:35

because if you're not

19:37

doing it, it's not genie going to

19:39

be disrupting your business. It is your

19:41

peers and competitors harnessing the power of

19:43

genie will. So it's really

19:46

important to establish a value realization

19:48

framework and playbook to

19:50

really start experimenting with genie, balancing

19:52

the risks and value all the

19:55

time along the way

19:57

that would be my first suggestion and second

19:59

suggestion. I think it is

20:01

time to look beyond the hype of Gen

20:03

AI. As we have now rolling into 2024,

20:06

we have seen pretty much every

20:09

other day there's another large language model

20:11

have made new splash, right? And

20:13

with that trend, we can all imagine a

20:16

future where large language models

20:18

themselves almost will become

20:20

a utility where

20:23

the wrapped around the services of

20:26

the application, as well as you

20:28

mentioned, Matt, the data will be

20:30

really essential for the company to

20:32

leveraging Gen AI. We can imagine

20:34

a world where large language model

20:36

become electricity and there will

20:39

be commonplace and

20:41

we can easily get it, but in

20:43

another day, it's a laptop, the refrigerator,

20:46

the cell phone, and all of these

20:48

other applications building on top of this

20:50

Gen AI utility, we'll really be

20:52

powering the enterprise. I think it's really

20:54

important for the leaders to think

20:57

about what is the future with

20:59

Gen AI become readily

21:01

available, what is that this

21:03

landscape could shape with the

21:05

power of Gen AI and

21:07

how to prepare the

21:09

transition between current state technology core foundation

21:11

to the future state gonna be and

21:14

start laying out some strategy and the

21:16

roadmaps to prepare to get there. Absolutely,

21:19

I think that's a really effective metaphor in

21:21

terms of utility and in terms of if

21:23

I can even not to get

21:25

too mixed up in metaphors, but even that the

21:28

LLMs will be like a lifeblood to

21:30

the organization in terms of its function

21:32

for all of these modules and capabilities

21:34

on top. I think that's a really,

21:36

really great way of looking at it.

21:38

We had Scott Zoldy, a FICO on

21:40

the program, not too

21:43

long ago talking about responsible AI

21:46

and the more that we get into

21:48

the practices that work and

21:50

really prevent a lot

21:52

of the nightmare scenarios that people can look

21:54

up in the news. It tends not to

21:56

be so much about hiring

21:59

folks. that speak to

22:01

the PR side of what ends

22:03

up happening in those worst case

22:06

scenarios. It's really more hiring the

22:08

right data governance people and also

22:10

having a management that's educated enough

22:13

on data governance and the relationship

22:15

to artificial intelligence development that they

22:17

understand that if you're cutting corners

22:20

in your data governance, it can

22:22

lead to the kind of biases

22:24

that lead to some really ugly

22:26

headlines. So I'm just wondering in

22:28

terms of integrating, you know, talent

22:32

that speaks to making

22:34

real responsible AI a

22:36

priority in the organization, where

22:39

do you wanna look in terms of

22:41

the skills? Is it more data governance

22:43

or is it something a little bit

22:46

more cultural within the organization? I

22:48

think it's a little bit of both,

22:51

Matt. I think even in the age

22:53

of Gen AI, the data you use

22:55

to train those foundational or language models

22:57

as well as the data you enable

23:00

Gen AI to ingest and

23:03

then vectorize and eventually generating

23:05

responses using proven design patterns

23:07

such as generated response

23:10

argument and generation. It's

23:12

really important without clean data

23:14

for training as well as

23:16

without clean data to

23:18

really enable the retrieval argument and generation,

23:21

the response you're getting from Gen AI

23:23

solution is gonna be as bad, right?

23:25

But in addition to the

23:27

focus on data quality and data governance, it

23:30

is also important to focusing

23:33

on helping the Gen AI

23:35

model to be in build

23:37

with privacy, transparency, and

23:39

making sure there is no way

23:42

to enable the explainability and fairness

23:45

in terms of those large language models. I think

23:47

that is a special type

23:50

of resources and requires special type of

23:53

framework to be able to Connecting

23:55

the dots with the data, but also as you're

23:57

building the model, finding the model Where you're building

23:59

the model. You again. Retrieval.

24:01

Argument A generation pattern to

24:03

generated content and long the

24:05

entire pipeline. There's. A clear

24:07

understanding and and the clarity in

24:10

terms of data. Yeah.

24:12

Absolutely In and I know in

24:14

terms of life sciences and then

24:16

on to the adjacent sectors like

24:18

health care that you know it,

24:20

really integrating. it's human experts feedback.

24:22

Keeping experts in the loop is

24:24

gonna be hugely important. I know

24:26

that's kind of underneath the surface

24:28

of the responsible Ai conversation because

24:30

it's a little bit more in

24:32

the model development area. It's kind

24:35

of being considered as especially as

24:37

folk folks and organizations or adopt

24:39

generative a I kind of the

24:41

given of. What you need to

24:43

do to really launch these tools

24:45

effectively. But I'm wondering if you

24:47

can touch on a bits in

24:49

terms of the talent that you'll

24:51

need to hire in in Id

24:53

to really prepare to keep those

24:55

human experts in the loop? And

24:57

who needs to be at the

24:59

table to emphasize what expertise needs

25:01

to be in that loop to

25:03

verify these answers especially for the

25:05

more klein facing solutions and models

25:07

or that I know that organizations

25:09

ultimately wanna get to in their.

25:11

Generative Ai goals. Yeah.

25:14

I think of events that the highest

25:16

level within other mason there needs to

25:18

be a leader that's real a mining,

25:20

the broader adoption of the ai. Across

25:23

the different disciplines so I have

25:25

a rod responsible A I lead

25:27

his release and showed to define

25:30

the discipline as well as a

25:32

framework for all that different utilization

25:34

of A to follow. I think

25:36

that's a first, the road, the

25:38

second are always really around with

25:40

so many different applications and the

25:42

so many different use cases. adopting

25:44

A I. Thought. To the

25:46

point I made Gloria what is of

25:49

value has been realized. How many people

25:51

are really using those Jenny I you

25:53

feel applications or a I have she

25:55

was applications. So the an innovative company

25:58

or start thinking about established. Really

26:00

focusing on. Matter ring

26:02

the utilization and will at the

26:05

value generated out with those to

26:07

the exhibitions. We. Caught it like

26:09

a valued. Tracking. At a

26:11

generative officer right and it's third

26:14

are always related to be around

26:16

much any. I'm a ice success

26:18

Such a fast moving field all

26:20

the company will need A scanning

26:22

mechanism is a well to to

26:24

keep the past on what is

26:27

allegedly industry trends, what is the

26:29

hyper skimmers and and abroad or

26:31

ecosystem vendors are is awaiting using

26:33

Jenny. I saw having a external

26:35

didn't have a son lead would

26:37

be central to canals Keeping the

26:40

policy. On. The market and the make you

26:42

for your not adopting or customize the building.

26:44

certain technology that is already gonna be rolling

26:46

out by another vendor out of the box

26:48

that these are the three rows not. I

26:50

think it's can be crucial for the company

26:53

to consider. Absolutely we. I really appreciate you

26:55

laying out like you know the specific titles

26:57

and skill set that they need to hire

26:59

for. Think that's really going to be clear

27:01

for our audience and that's not publish.sight of

27:03

the table in terms of you know those

27:05

books. Listening Horsey I was out there. How

27:08

did they build this within their own organizations?

27:10

The. Other half of the conversation you

27:12

were talking about leadership before is how

27:14

are you selling the rest of the

27:16

see sweet the rest of management on

27:19

this and this has been a conversation

27:21

that is really taken a one eight

27:23

even from when I first started here

27:25

about a year and a half ago.

27:28

But I think it's definitely a completely

27:30

different colored than even when it was

27:32

half a decade ago. In that's you

27:34

know, it's it Sees Meters Ceos leaders

27:36

were is skeptical about Ai they really

27:39

hadn't known in their own. personal lives

27:41

so was always an uphill battle or

27:43

to really sell the stuff now especially

27:45

in the advent of open a i'd

27:48

sad cbt this really having a bigger

27:50

moment in the concerts you know ceos

27:52

their kids are using it in in

27:54

in school is there a lot more

27:57

excited and they also might be caught

27:59

up in a media hype cycle that

28:01

isn't really depicting more accurately

28:03

the in capabilities of these

28:05

technologies accurately for even for

28:07

as much as there's been

28:10

coverage over the hallucinations and the misinformation and

28:12

things like that. I think there's still a

28:14

lot of education out there even among business

28:16

leaders that needs to be done just

28:19

in terms of those capabilities, not just

28:21

in terms of the symptoms that you

28:23

see at the end, hallucinations, misinformation, but

28:25

why, why do we get there? And

28:28

I'm wondering what advice

28:30

that you have now that

28:32

those tables have turned, now that it's more

28:34

likely the CEO is gonna bring

28:37

up, hey, why aren't we, where are we

28:39

on general day? They're gonna be driving the

28:41

conversation. They don't need to be convinced about

28:44

the destination, but they do need

28:46

to probably have some conversation about

28:49

realistically how to get there. And I'm

28:51

wondering how have you seen

28:53

those conversations change and what advice that

28:56

you have for especially IT leaders in

28:58

terms of conducting them and getting them

29:00

in the right direction? Yeah, thanks for

29:02

the question, Matt. I think we are all known

29:04

now, now it's 2024. The

29:07

genie is out of the box in terms

29:09

of gen AI and it's rapidly blasting through

29:11

the hype cycle faster than

29:14

any technology innovation before.

29:17

I think if I can offer

29:19

just two suggestions to the leaders,

29:21

one is around the benefits and

29:23

risks balancing, the second is

29:25

around looking through the hype on gen

29:27

AI. So first on like

29:29

the benefits and risks, I know there

29:31

are a lot of overblown anticipated

29:34

value as well as over-couser stance

29:36

in terms of adopting gen AI.

29:39

I think both of these

29:41

stance are stemmed from lack

29:43

of experience and understanding. I

29:46

think from a value standpoint, leaders need

29:48

to set aggressive goals to

29:50

create the motivational poll for

29:52

ease or hard team

29:54

to start really pushing the envelope

29:57

of adopting the technology, right? And

29:59

Then the... That you can be relied

30:01

speed, experience and and across that

30:03

can be take out. And from

30:06

a viscous standpoint, it's really important

30:08

to have a framework to. Start.

30:11

Say Slate Spare Amending and

30:13

a building institutional understanding of

30:15

Jenny I Because. If.

30:17

You're not doing it. It's not funny. I

30:19

gonna be disrupting your business. It is your

30:22

peers and a competitor's started seeing the power.

30:24

I'm Jenny I well. So. It's

30:26

really important to establish of value than

30:28

a decent framework and pointed. To

30:31

really start experimenting with Jenny I

30:33

balancing the risks and value. On

30:36

all the time along the way that

30:38

it would be my for suggestion and

30:40

and seven suggestion is I think it

30:42

is time to look beyond the hi

30:44

I'm Jenny I As we have now

30:46

rolling into Twenty Twenty four we have

30:48

seen pretty much every other day. There's

30:50

another large on good model have and

30:52

make new splash lights on a way

30:54

that trend with a all imagine a

30:57

future where. Large. Somebody models

30:59

and cells. Almost. Will

31:01

become a utility. Where.

31:04

The wrapped around the services of the

31:06

application as well as humans in a

31:09

map the say that the will be

31:11

released and show for the company to

31:13

leveraging. Ginny I can imagine a world

31:15

where large language amount of become electricity.

31:18

And and there will be commonplace and

31:20

and we can easily get it, but

31:22

in another day it's a laptop that

31:25

Spades reader, the cell phone and all

31:27

of each other up the case and

31:29

building on top of the student at

31:31

a utility. Will. Really be powering

31:33

the enterprise. I think it's very

31:35

important for the leaders to think

31:37

about what in the future which

31:40

turning I become. Readily

31:42

available whether that is this landscape

31:44

said could save with the power

31:46

of Jenny I and the how

31:48

to prepare. The. Transition between sovereign

31:51

States technology course on they sent to

31:53

us he trusted gonna be and start

31:55

laying out some strategy and roadmaps the

31:57

prepared to death. And.

32:00

I think that's a really effective metaphor in terms

32:02

of utility and in terms of if I can

32:05

even Not to get too mixed

32:07

up in metaphors But even that the LLMs will

32:09

be like the lifeblood To the

32:11

organization in terms of its function for all

32:13

of these, you know modules and capabilities on

32:15

top I think that's a really really great

32:17

way of looking at it I I've admittedly

32:19

you know through the course of this interview

32:21

been kind of dancing around the cultural aspect

32:23

And I really appreciate in your last answers

32:25

you've you've gotten right in there and talked

32:27

about it But I think every

32:29

leader knows out there that real leadership is

32:32

not about when you talk it's when you

32:34

don't talk and when you let somebody else

32:36

talking when you listen and This

32:38

is you know a skill that

32:41

I'm still working on admittedly in

32:43

my leadership capacities But

32:45

especially knowing when not to talk and when to

32:47

let with somebody else in

32:49

the team in the adoption process

32:52

Really take the reins. I think is instrumental

32:54

in terms of the difference it can make

32:57

Really really making sure that this

33:00

can succeed so I'm wondering in

33:02

your experience in terms of you

33:04

know, IT adoption initiatives and Especially

33:07

in this new generative AI

33:09

landscape and infrastructure Needs

33:11

that we need to consider who typically

33:13

serves as the catalyst or or group

33:16

of catalysts in leading AI Work

33:18

streams in driving cultural change within organizations

33:20

Especially where you were talking about in

33:22

your last answer about how education really

33:25

needs to not just start at the

33:27

top But you know really really get

33:29

down to the bottom there to every

33:31

last employee Is it a single leader

33:33

or is this more of a collaboration?

33:35

I? Think

33:38

some more of a collaboration that

33:40

what we have seen from our clients

33:43

is our top-down Monday

33:45

from the C-suite to really Empower

33:48

the executive leaders from the line of

33:50

the business to partner up with the

33:52

digital leaders The IT leaders

33:55

to work together on realizing the value of

33:57

Gen AI and then I think The

33:59

partner. The ship also proven to

34:01

be really successful for quite a

34:04

few clients we have seen as

34:06

the business leaders understanding the heart

34:08

of possible in terms of were

34:10

in their business. They. Could

34:12

really use the help of this

34:14

new innovation and the digital leaders

34:17

will will be the granting of

34:19

forcing from though quickly helping the

34:21

business leaders understand what's possible with

34:23

not possible with the current constraints

34:25

on technology has about what is

34:28

on a road map. With.

34:30

These two leaders working together and

34:32

also in consulting with and responsible

34:34

a i'll either to making sure

34:36

there's guard rails be put into

34:38

place We've seen the adoption as

34:40

was a broader education or do

34:42

other narratives to the probably why

34:44

you are readily the wow and

34:46

will have far reaching a positive

34:48

impact of for also motivating the

34:50

workforce to start adopting an Ai

34:52

and to that day to day

34:54

work. Yeah,

34:57

I think especially on you know the

34:59

cultural side. I don't think there's one

35:01

person that you can hire for. Yeah

35:03

I think it's really needs to be

35:06

a collective effort and with really robust

35:08

in in clear frameworks like what you

35:10

were talking about at the beginning of

35:12

the episodes that are a lot more

35:14

all purpose in can cover organizations no

35:16

matter where they are in their ai

35:19

adoption. I think we're going to see

35:21

a lot more people get on board.

35:23

it's it's just a very interesting moments

35:25

but Stevens really illuminating episode. Really appreciate

35:27

it. The specificity of your answers and really laying

35:29

out enlists you know what people need to look

35:32

for, what people need a higher forth and how

35:34

to put it together. Thanks so much for being

35:36

a Thank you So much for having me that

35:38

I presented. before

35:47

we close out today's episode seem

35:49

pointless even brought up that i

35:51

think deserves some greater spotlight first

35:53

to deliver faster services with minimal

35:56

frictions throughout the delivery life cycle

35:58

focus on saleable flexible and

36:00

industrialized products and services to

36:03

establish a robust digital foundation

36:05

for agile operations. Steven tells

36:07

our executive audience that it's

36:10

crucial for management and data

36:12

science teams to work together

36:15

intentionally. Intentionally being the key

36:17

word there. Steven outlines a

36:20

four-pillar framework for rapid generative

36:22

AI adoption that includes first

36:24

establishing a large language model,

36:27

second understanding foundational technologies, thirdly

36:29

envisioning a platform for genetic

36:31

innovation, and fourth and

36:34

finally applying generative AI for

36:36

IT including understanding and leveraging

36:39

tools for code conversion. Prompt

36:42

engineers and other existing roles are

36:44

in high demand to meet these

36:47

talent needs. Addressing challenges in upskilling

36:49

and rescaling existing talent, especially considering

36:51

biases and security concerns, is a

36:54

priority for Deloitte. Steven tells our

36:56

audience they approach this

36:58

through a comprehensive change management

37:01

strategy, including creating awareness, training

37:03

curriculum development, and equipping employees

37:05

with generative AI knowledge and

37:07

tools. Integrating expert

37:09

human feedback into AI development

37:12

is crucial in life sciences. Steven

37:14

tells our audience for more insights

37:16

check out the August 31st, 2023

37:20

episode of the AI in

37:22

Business podcast featuring Centaur Labs

37:24

CEO Eric Duham discussing expert

37:27

feedback challenges in healthcare model

37:29

development. Steven elaborates

37:31

on different roles within a

37:34

successful digital transformation, including a

37:36

leader driving broad AI adoption,

37:38

role focused on maximizing the

37:41

value of generic solutions, value

37:43

tracking, and a generative officer.

37:45

This most of the time gets called

37:48

and a leader responsible for

37:50

staying updated on

37:52

rapidly evolving generative AI industry

37:54

trends. So that's two different

37:57

leaders, one focused on

37:59

maximizing optimizing the value of generic

38:01

solutions, and another leader

38:04

responsible for staying updated on

38:06

the rapidly evolving generative AI

38:08

industry trends. Steven predicts

38:11

that large language models will become

38:13

accessible and widely used as electricity

38:15

in the future, much like other

38:17

utilities. And in leading AI work

38:19

streams, Steven emphasizes the

38:22

importance of collaboration over a

38:24

single defined leadership. collaborative

38:26

collaboration supported by robust

38:29

frameworks enables organizations to

38:31

adopt AI across different stages

38:33

of adoption. And

38:35

finally, on behalf of Daniel Fajela, our CEO

38:38

and Head of Research, as well as the

38:40

rest of the team here at Emerge Technology

38:42

Research, thanks so much for joining us today

38:44

and we'll catch you next time on the

38:47

AI in Business Podcast. Thank

38:56

you.