Eoghan Casey 38 min

Future IT Considerations for Federal and State IT


0:00

And Gordon Bitco with the Information Technology Industry Council, I'm going to

0:08

ask each of

0:09

our panelists to just introduce themselves really briefly.

0:12

And then to, by way of introduction, just to talk for a few minutes about what

0:16

are some

0:17

of the key challenges that you've learned over the last few years?

0:19

You've all been doing government IT in one way or another.

0:22

Some of you still are.

0:23

What are those big challenges?

0:24

So, Cashant, why don't we start with you since you're your closest?

0:27

Sure.

0:28

Thank you, Gordon.

0:29

Hi, everyone.

0:30

Cuz she's Pandia.

0:31

I am the Chief Technology Officer for Internal Revenue Service.

0:34

Hi, good afternoon.

0:37

Karen Howard.

0:38

I'm the Director of the Office of Online Services with the Internal Revenue

0:42

Service.

0:43

Hello, I'm Owen Casey.

0:45

I, formerly with government and DOD, but I came out a year and a half ago to be

0:49

the VP

0:50

of Cybersecurity Strategy and Product Development at OWN.

0:53

All right.

0:54

Thank you.

0:55

And I'm back with you again.

0:56

Andrew Nielsen.

0:57

I'm the Director of the Government and Wide IT Accessibility Program that sits

1:02

within the

1:03

Office of Government and Wide Policy at GSA.

1:05

And I'm also representing kind of a sub office, the Office of Technology Policy

1:09

So, my colleagues who also work and provide guidance in identity and credential

1:14

ing and

1:15

access management, cloud center optimization and emerging technologies.

1:20

Great.

1:21

Thanks, Andrew.

1:22

And so, why don't we actually start that first question with you.

1:25

What are the couple of key lessons that you've really learned over the last few

1:29

years?

1:30

Yeah, I think, you know, on my, you know, more narrow perspective on

1:35

accessibility, you

1:38

know, I kind of, we talked a little bit about it this morning as part of the

1:42

customer experience

1:44

panel.

1:45

But more and more, of course, more and more of our services are moving to the

1:49

digital

1:50

sphere.

1:51

And so, I think, you know, certainly a trend that we're seeing with the advent

1:57

of, and

1:57

development of low code, no code environments, that obviously the advent of

2:02

artificial intelligence,

2:04

generative AI, just more and more customization, more technologies, specialized

2:10

technologies

2:11

for specific line items of business.

2:15

And so, I think, you know, we're seeing that.

2:17

We'll see only more of that as we make better and more intelligent use of gener

2:23

ative AI's

2:25

and as they get better.

2:27

And so, you know, along with that, you know, of course, the challenges.

2:31

I won't speak to all of them.

2:33

But the challenges that come with that in making sure that we are making

2:37

intelligent

2:38

use of those and that we're safeguarding privacy, we're safeguarding the

2:42

security of our applications

2:44

and our digital services.

2:47

And then again, maybe Naramine, again, on my accessibility focus, ensuring that

2:52

as we

2:53

-- and I think there's probably a question later.

2:55

I'm skipping ahead to the question on innovation.

2:58

But as we innovate, that we don't leave people behind, and that, again, on my

3:03

accessibility

3:04

focus, that we need to ensure the legitimacy of our government, the services

3:12

that we provide

3:13

by ensuring inclusion and ensuring that we meet the needs of all of the public,

3:21

the citizens

3:21

of the U.S.

3:23

So, yeah, sorry.

3:24

>> No, no, great, Andrew.

3:25

I think all super helpful.

3:26

I'll note, you know, not just the eclipse but an earthquake going on here at

3:29

the same

3:30

time, too.

3:31

So, we're full-futures, full services.

3:33

Oh, and same question to you to start with.

3:36

You know, what are the key lessons over the last few years that you've really

3:39

taken away?

3:39

>> So, I have to balance innovation and security.

3:44

So I came from a background of dealing with security breaches and investigation

3:47

, but moved

3:47

into more of a proactive state.

3:50

But trying to translate, I think, some of the lessons we've learned from the

3:54

past -- from

3:55

the past, I think there's been a lot of good discussion today about data being

4:01

so critical

4:02

now in everything that we do, and making effective use of that, and doing it in

4:07

a more

4:08

speedy manner, we have to keep pace with the changes in technology and the

4:12

advances that

4:13

are coming with stronger analytics and more advanced AI, but balancing that

4:19

with proactive

4:20

security from the outset.

4:22

And so I come in with that integration of the innovation and the protection.

4:28

We can't just let the data go ahead of us and try and catch up.

4:34

We will not catch up.

4:36

I can guarantee it.

4:37

>> Yeah, for sure.

4:38

I think that's a great point.

4:40

And Karen, I'd like your thoughts as well.

4:42

But one of the themes that we just got from both folks is the importance of

4:46

people in

4:47

all of this.

4:48

And given your role, I'm sure you've got some thoughts there as well.

4:51

>> Sure.

4:52

I think one of the things that I think we're really starting to understand is

4:57

the importance

4:58

of user experience design in the design and putting that first, letting that

5:05

drive the

5:06

design of products, the design and the needs of the business so that then the

5:11

technology

5:12

follows behind what's learned in the UX research and the data that helps us

5:17

understand that.

5:19

I think that's one of the things that you're seeing a really good focus on now

5:23

that you

5:23

didn't before.

5:24

Typically, UX was kind of an afterthought.

5:28

It's like, oh, now that we've designed this, now let's get designers to look at

5:32

it.

5:32

And it was about colors of call to action button.

5:36

And now it's really about partnering with our IT partners and making sure UX

5:41

design is

5:42

in the forefront of development.

5:44

And not an afterthought.

5:45

I think the other thing that these two gentlemen have already really spoken to

5:50

is data and

5:51

data-driven decisions and data that supports business cases.

5:56

Sometimes we think we know what our users want and what our customers want.

6:02

But using data to kind of validate or help us understand anomalies that we can

6:06

address

6:07

in design.

6:08

And I would end with saying that letting business needs and business problems

6:13

and business use

6:14

case drive technology solutions and not going for technology and then looking

6:20

to retrofit

6:21

it into the business.

6:22

Yeah, I think that's a great lesson because I'd certainly like to get your key

6:27

lessons

6:28

from the last few years as well.

6:29

So I'm going to add on a spin to that around that point.

6:32

You've got obviously the IRS tons of data.

6:36

And how do you think about those exact issues?

6:38

We've got all this data.

6:39

We've got these really challenging use cases for Americans who expect certain

6:44

behaviors

6:45

on the IRS.

6:46

And so how do you balance those?

6:48

Yeah.

6:49

No, data certainly is and we have more than our fair share of it.

6:56

Over the last couple of years what we have really observed and the pandemic hit

7:01

this

7:01

home is that we need to find better, more intuitive ways to engage with our

7:07

taxpayers.

7:09

And our taxpayers in turn want to engage with us in a similar manner.

7:13

Perhaps you've seen these unfortunate pictures of us having volumes of paper

7:17

stacked in our

7:18

cafeterias and all over our IRS campuses.

7:23

We have really focused on digital transformation.

7:26

So digital transformation, we cannot stop paper from coming in.

7:30

However, we can convert paper and digitize it and digital transformation has

7:35

been a pioneer

7:37

of a program for us.

7:39

However, what we observed is while we were converting our paper to digital

7:45

images internally,

7:46

our people were printing them out.

7:49

Of course they were.

7:51

Because in order for them to get access to the data and if it wasn't

7:55

necessarily the

7:56

highest degree of accuracy, so they're printing out all the paper, we were

7:59

digitally transforming

8:00

at our doorsteps.

8:02

We then realized that the reason they were printing this out is because they

8:05

needed to

8:06

key in the data manually into the systems.

8:09

Our focus then shifted to not just converting them to digital format but really

8:14

honing in

8:14

on extracting the data and feeding it downstream to our systems so that our

8:19

people are focused

8:20

on running the necessary analytics, doing all the work that is required to

8:25

assist taxpayers

8:26

versus printing out and punching in that same data that we were collecting.

8:30

Yeah, I think that's a great example that highlights Karen's point about

8:33

understanding

8:34

the business case and really knowing what the technology is for.

8:37

But at the same time, it also highlights one of the challenges I always had in

8:41

these

8:41

transformations is the people and the cultural aspect of they're used to doing

8:45

something

8:45

the same way for so long or you as a federal agency have compliance

8:49

requirements where

8:50

you have to do things a certain way or the agency's made a risk management

8:53

decision to

8:54

do things a certain way.

8:56

And so Karen, if I could go back to you, how do you deal with changing that

9:00

environment?

9:01

Well, I think when it comes to changing an environment where you've got a lot

9:06

of legacy

9:07

practices and comfort in doing things the same way, you've got to really find

9:13

the win-win

9:14

for the employees or the people doing the work.

9:18

I think part of it is going to involve upskilling and how it adds value to

9:23

their role sometimes

9:24

to their bank account.

9:25

Wouldn't that be nice?

9:27

Just giving them more skills and showing them the benefit.

9:31

Also making them a part of the solution.

9:33

I think when you talk about change management, first of all, you've got to make

9:37

sure the

9:37

business is ready so there's business readiness or organizational readiness and

9:43

then it's

9:44

really making them part of the decisions around what you're planning to do and

9:49

getting their

9:50

feedback along the way, giving them ownership in the change and also showing

9:55

them the benefit

9:57

to them.

9:58

Not everyone will grasp that and be excited about it, but most people will and

10:03

then the

10:04

outliers, you know, you handle those in the best way you know how and help them

10:08

find their

10:09

happy place.

10:10

But one of the things I can say, change management, it's easy to say and it's

10:14

hard to do.

10:16

And I think organizations need to understand where they are from a cultural

10:21

standpoint

10:22

in the maturity for change and begin to implement change based on that.

10:28

Some organizations are ready for it and some organizations need to move at a

10:33

slower pace.

10:34

But when it comes to change management, if you build it, they will come, does

10:39

not apply,

10:40

especially when you have a lot of legacy employees.

10:43

You've got to figure out how to make it affect them in a positive way.

10:48

Yeah, thanks, just to follow up on that and, Andrew, I'm going to go to you

10:53

with this

10:53

question.

10:55

How do you balance that change management challenge with the potential of all

10:58

these emerging technologies,

11:00

which I would imagine when it comes to accessibility, have the potential for

11:04

huge benefits, right?

11:04

AI can really transform a lot of the challenges that we have in access and

11:09

accessibility.

11:10

Yeah, yeah, I mean, particularly with regard to artificial intelligence, gener

11:16

ative AI,

11:18

maybe narrowing in on that world of accessibility and then broadening.

11:23

There has been some argument in our little world of accessibility that AI will

11:28

just solve

11:29

all of our problems.

11:30

AI as an assistive technology can just help anyone with a disability regardless

11:37

of the

11:38

type of disability.

11:39

When it comes to interacting in the digital sphere, we don't need to worry

11:43

about making

11:44

our stuff accessible anymore because you could just use AI as an assistive

11:48

technology and

11:49

then that fixes everything.

11:52

And then, by the way, it's also, that's to our benefit because it costs too

11:57

much and

11:57

it's just too much effort and there's too much cognitive load.

12:02

I have to train people to know how to make their stuff accessible.

12:05

And that's just kind of wrong-headed in the first place.

12:11

It's subscribed to the model of looking at an individual and their differences

12:15

as differences

12:16

when not recognizing that differences are really dependent on environment and

12:24

context.

12:25

And especially in our broad human experience, everyone has differences and when

12:32

it comes

12:33

to disabilities, rather than relying on somebody to fix their difference, we

12:39

should adapt our

12:40

environment.

12:41

We should adapt our context to meet the needs of individuals and that can go

12:46

for any type

12:46

of difference.

12:48

And so there is still fantastic promise.

12:51

I'm more excited about the use of AI on the intent, on that side, on that

12:57

really side of

13:00

development, whether it's development of a process or development of technology

13:03

, and

13:04

using AI to help as a copilot, if you will, or a tutor or a partner, I want to

13:11

do a thing

13:12

I might not know how and so help me do that thing.

13:15

And so, again, with what we're talking about, low code, no code environments or

13:21

a developer

13:22

who knows all of the bells and whistles but just doesn't know accessibility,

13:28

that is a

13:29

very positive, I think, a positive development we can look forward to.

13:37

And then expanding that out to other spheres outside of just accessibility,

13:44

there's so

13:45

much more where it's applicable.

13:47

Not only when I'm looking to meet the needs of individuals with disabilities,

13:50

but when

13:51

I'm looking to meet broad customer needs, again, in the digital sphere, to

13:56

better understand

13:57

that I don't know everyone's circumstances or all of their needs, but I can use

14:03

AI to

14:04

inform that and to broaden my context and to rely on that rather than only on

14:09

my knowledge.

14:10

Great.

14:11

That's, I think, a really helpful set of insights into the ways that AI can

14:15

help.

14:16

But when there's a flip side, AI introduces threats, changes the threat

14:20

landscape.

14:21

So I'm sure you've got some thoughts about how we should be trying to

14:24

incorporate that

14:25

into our understanding of the threat environment.

14:28

Well, I think there's the traditional threats just applied to the new

14:32

technology which we

14:33

have the disruption by adversaries.

14:36

But I think one other factor that we have to consider is bias.

14:40

And this is something that came up in roundtable discussions, what we also see

14:44

in across sectors

14:45

is the potential for some of the training that's been done or even just the

14:50

data that's available,

14:53

introducing certain biases.

14:54

So if you think of, in a hospital setting, doctors are being approached by

14:59

pharmaceutical

15:00

companies to promote a particular prescription, that's then setting a trend, a

15:06

historical

15:07

trend that could ultimately bias any automation that we have to help with

15:12

prescription management,

15:13

prescription efficiencies.

15:15

So we have to, I think, be careful of the risks that are unintended of the use

15:20

of the

15:21

technology also.

15:22

And one of the things that I think pulls us all together is if we have humans

15:27

in the loop

15:27

for sure who are able to more quickly, if you think about the speed of

15:31

iteration that

15:32

we now can achieve with this technology of getting, it's the Uda loop from, I

15:40

guess,

15:41

the military context but applied here, we're talking about observe, orient,

15:44

decide, and

15:45

act.

15:46

We can do that more quickly now and make adjustments and see problems more

15:49

quickly and adapt.

15:51

I think that that's some of the potential that I see where we can avoid some of

15:56

the problems,

15:57

we can be more proactive and avoid some of the problems from a security, a bias

16:01

perspective

16:02

and perhaps come up with better solutions in the long run.

16:04

So we're not getting rid of people is what you're saying.

16:07

I think that's given.

16:09

So because I think that aligns really well with one of the use cases we were

16:13

talking

16:13

about outside around how you're looking at helping the IRS workforce with AI

16:20

tools,

16:20

not replacing them.

16:22

Correct.

16:23

Yes.

16:24

We are being very cautious, naturally so, about how we leverage the

16:29

technologies.

16:30

AI is quite prevalent and what people equate AI to is chat GPT.

16:36

I mean, that is what is typically known.

16:39

However, we want to be much more deliberate about how we apply it.

16:44

And we're being cautious on how we apply it for external facing.

16:47

We want to be very, very careful of how our taxpayers engage with us and

16:52

leveraging AI

16:53

and we don't want to create those hallucinations or have any misleading

16:57

information.

16:59

If we were focused on or populate our data models with scour the internet for

17:05

any and

17:05

all tax related questions, somebody on Reddit might say you don't need to pay

17:09

taxes and

17:10

all of a sudden when a taxpayer asks if I should or not, that's the answer that

17:15

pops

17:15

up.

17:16

We're looking more internally and there are a plethora of opportunities

17:20

internal to IRS

17:22

on how we can apply AI.

17:24

For example, we have a legacy or I would say an aging workforce that is leaving

17:32

and they

17:32

have been very accustomed to reviewing manuals that are physical, that are

17:37

paper and they're

17:38

looking across these lengthy and thick workbooks on how to apply or how to

17:44

evaluate or how to

17:46

assess.

17:47

However, as the newer generation comes in, they are much more open to and

17:52

wanting the

17:53

digital equivalent of it.

17:55

But the volume of data contained, however, hasn't changed.

17:58

In fact, it's done nothing but grow.

18:01

How can we assist our internal employees to not have to figure out which page

18:06

to look

18:06

for and we are applying AI against that?

18:11

Convert our internal information and make it more conversational so that our

18:14

internal

18:15

employees are able to simply ask a question, to go and find that information

18:19

rather than

18:20

have to figure out which book contains what part of the tax law.

18:24

I'm disappointed to hear that Reddit is not in the favor too of a source.

18:29

I might have to sell my stock.

18:32

That's a great example though of how do you use technology to actually drive

18:36

change in

18:37

the organization.

18:38

So I'm going to go back around and ask each of you that question, Karen,

18:41

starting with

18:42

what's the thing where you think there's the best opportunity for technology

18:46

and your

18:47

workforce together to be transformative in that way?

18:50

I think if we can leverage technology to address business problems, customer

19:00

problems,

19:01

user problems, and to bring efficiency in addressing those either from a

19:06

customer facing

19:07

perspective or behind the scenes from an operational perspective.

19:12

If we can look at, you know, to, because she's point using artificial

19:17

intelligence to help

19:19

us with a fine anomalies or to fine patterns that help us identify scams and

19:24

schemes which

19:25

is, you know, really a big thing right now and usually is around tax filing

19:30

times.

19:31

We can use it to find pain points in arduous processes that take a long time

19:37

and help eliminate

19:39

those.

19:40

I think if we can leverage platforms, workflow management to really identify

19:45

bottlenecks and

19:46

workflow and really hold people accountable.

19:49

I think right now with a lot of manual processes in place and not knowing where

19:54

the black

19:55

holes are when we're waiting for a response or looking to understand where a

20:00

document

20:01

that, you know, it's going to drive a business decision or a action is located.

20:07

There's just so much if you just look at what's causing problems in the

20:11

organization and focus

20:13

on those and that is serving the team, making your employees involved in what

20:18

are your biggest

20:19

pain points if they say it's, you know, procurement or hiring or onboarding

20:27

employees and kind

20:29

of mapping that out in a value stream way and then saying, okay, how can we

20:33

leverage

20:34

technology to do some of this stuff?

20:36

That's where technology begins to pay off.

20:39

And again, like I said before, that is using business needs to drive technology

20:44

and not

20:44

the other way around.

20:46

Great, can we maybe afterwards just have a purely intellectual conversation

20:50

about how

20:50

you're using models to find scams and tax filings?

20:53

We absolutely can.

20:55

Just hypothetical, like I said.

20:58

Oh, and same question over to you.

21:00

I'd really like to just continue that.

21:03

I completely agree, Karen.

21:04

And I'm happy to say I had a very positive experience with the IRS and working

21:08

with

21:08

someone who knew a lot of the systems in place.

21:11

And I think what we really have a potential for here is to take some of those,

21:16

we might

21:17

call it a pain point, but I also can think of this as finding the things that

21:22

makes us

21:22

successful and amplifying those across the organization.

21:25

So it's using the historical successes or the factors that contribute to

21:29

success historically

21:31

and amplifying that with the technology to help people across our organizations

21:35

be successful.

21:36

Andrew?

21:37

Yeah, I think, you know, broadening it.

21:41

Remember my comments on specific to accessibility.

21:44

While there again, just some fantastic, I think, things to look forward to

21:50

using advancing

21:51

technologies, AI, to help us as partners, as mentors, as co-pilots, to make our

21:59

content

21:59

and our digital tools and services accessible to people.

22:04

I think another thing to look forward to is that while, in some cases, it may

22:10

also be

22:11

perceived as a threat, but the hyper-customization of my experience as a

22:15

customer, especially

22:17

in the digital sphere.

22:19

I think most of us are accustomed to seeing custom ads and maybe get a little

22:25

bit worried

22:25

that they're overhearing, my computer's overhearing, my conversation, how did

22:29

you know I was talking

22:31

about that earlier?

22:33

And so there's some advanced algorithms, but there also, and so we need to be

22:36

concerned

22:37

about privacy and security, but there's also a great opportunity for hyper-

22:41

customers.

22:42

More relying on cues that you provide as a user or as a customer to customize

22:49

your experience

22:51

and to choose your own experience, if you will.

22:54

And so I think, again, technology will only continue to quicken our pace and

23:03

our ability

23:05

to adjust and to provide new technologies that's only going to increase,

23:11

especially with AI.

23:13

And I think, again, more and more, we'll be able to customize our experience to

23:16

our needs

23:18

and excited about that.

23:20

So Andrew, just to stick with that comment for a second, because I think that

23:24

the hyper-customization

23:25

thing is certainly really fascinating in government.

23:28

It's something that we all see and experience in our own personal uses of

23:32

technology.

23:33

Can government policy, can government compliance, can government regulation

23:35

keep up with the

23:37

ability to do things like that?

23:39

You look at how long it takes an executive order to come out or guidance to

23:42

come out

23:43

from OMB to agencies, and it can be years from the time that it's discussed

23:48

until it's

23:49

in place.

23:50

Yeah, I think that there is still room for that.

23:54

I mean, of course, that can be hard.

23:57

I think a starting point, I think, first of all, let's not start from scratch.

24:01

Let's rely on best practices.

24:04

Let's rely on design systems, for instance.

24:06

The agencies, hopefully, are all using the US Web Design System for your web

24:10

presence.

24:13

Start there, and then customize on top of that.

24:16

So start with what you have.

24:21

Don't necessarily start from scratch.

24:23

Build on best practices.

24:25

Yes, absolutely, I think our pace will still be slower in government as we

24:30

adapt regulation

24:32

and law and rely on our legislators to do that as well.

24:37

So I don't know that I have a silver bullet answer.

24:40

There will be some lag.

24:42

But I think that there is still a lot of room based on a lot of really great

24:45

work and tools

24:46

that we have available.

24:47

Yeah, I don't think anybody expects government technology to be the same as

24:52

what it is at

24:52

home.

24:53

We'd like that, right?

24:55

But if we can converge, right, if those things can get closer.

24:59

And Karen, one of the things that you said about the role of some of these

25:02

technologies

25:03

in the procurement process, I think, is that the core of trying to do that

25:07

better and smarter

25:08

and faster.

25:09

So I don't know if that's something that IRS has started to think about yet.

25:12

We have, and probably because she can probably speak to this better than I can.

25:16

I know with the funding, the influx of funding from the Inflation Reduction Act

25:21

, we are addressing

25:23

a lot of the historic, iconic processes that we all know and love.

25:30

But I do want to speak to one thing.

25:32

I always function on the mindset of there are some things you can't control or

25:37

that

25:38

are going to be what they're going to be.

25:40

What are the things that I can control?

25:42

So when we talk about customization, part of that is personalization.

25:47

So really looking at one of the things that I really have worked hard on with

25:52

my IT partners

25:54

are online account.

25:56

And how do we make that so people are getting more of what they want after they

26:01

authenticate?

26:02

So to answer your question, I really think that to this point, there is still

26:07

room to

26:08

grow.

26:09

But we can continue to look at the things we can impact while we wait for the

26:14

wheels

26:14

of government to continue to turn and we continue to educate ourselves,

26:18

leveraging our industry

26:20

partners in the private sector to know what's coming and to be ready.

26:25

Is she, did you have more to talk about procurement?

26:29

Absolutely.

26:30

First of all, we have a lot of witnesses who did hear Owen say he had a

26:33

pleasant experience

26:34

with the IRS.

26:38

So procurement is one of those areas that we have found that we can apply

26:43

intelligence

26:44

against to expedite how we procure and how we deliver, how we award contracts.

26:49

In fact, we have, to Karen's earlier points, we really are targeting business

26:55

cases.

26:56

And when we talk about business cases, we're saying what can we do to help

26:59

improve our

27:00

internal business users to do their jobs better, but also our customers as a

27:05

whole, all of

27:05

the taxpayers, ourselves included.

27:08

It is looking beyond technology and thinking about what problem exists and how

27:12

can we apply

27:13

technology to solve it versus saying here's a technology, now let's go find the

27:18

problem.

27:19

The procurement one is really interesting one.

27:21

We have a procurement office where individuals were spending hours and hours,

27:27

if not longer

27:28

than that, and perhaps days scouring the right websites to determine whether or

27:35

not a vendor

27:36

we are going to engage with is tax compliant.

27:39

And if they're not tax compliant, we naturally can't engage with you.

27:43

And this would be a manual effort.

27:45

People would go on websites and do research.

27:48

We applied intelligence against it, and now we've cut down from multiple days

27:53

to minutes,

27:53

five minutes currently to find whether or not a vendor is tax compliant.

27:58

That's a real case example of how fast we can now perhaps determine whether

28:03

somebody's

28:04

qualified or not.

28:05

That's some of the really quick use case.

28:07

So that's our internal team that's focused on that we've been thinking about

28:10

how to solve

28:11

for.

28:12

The other one is again, all of us, the taxpayers.

28:15

When you file a tax return today, it gets submitted, it gets processed within

28:19

hours.

28:20

However, if you have to amend a tax return, if you had to modify it, it takes

28:25

months.

28:26

And the reason for that is because when it gets modified, somebody internal to

28:32

IRS has

28:32

to evaluate your tax return to your new amended return and make a judgment call

28:38

on whether

28:38

or not this is something that needs further scrutiny or it should pass muster.

28:43

What we are doing is applying intelligence to say you don't have to look at

28:47

everything.

28:48

Here are the biggest discrepancies.

28:49

So you really have a decision to make on these three focus areas.

28:54

Now it's cut down the time that our internal representatives spend on

28:59

evaluating the whole

29:00

form versus the three most typical or the three most questionable areas.

29:07

And now it's just their judgment.

29:08

I'm going to pick amongst the three versus spending days and evaluating one by

29:13

one between

29:14

the two.

29:15

You would think that there would be a self-selection rate that companies that

29:19

are not tax compliant

29:20

aren't going to look to do business with the government, with the IRS in

29:23

particular.

29:23

But apparently that's a bad assumption.

29:26

So I'm glad to know that you guys check that.

29:28

Oh, and from the outside, what does modernized procurement look like?

29:34

How is technology helping?

29:36

I was hearing some of this and was thinking the modernization and some of these

29:40

efficiencies

29:41

are to be encouraged.

29:44

But the data that we're using as we move into some of these modernized

29:49

capabilities and

29:50

we can talk about sales force here, the lessons that we've learned from the

29:55

past about securing

29:56

that data, having continuity of operations as we move forward with that new

30:01

technology

30:02

and new solution, we have to translate.

30:05

And we're running behind in terms of our ability to do that.

30:09

So I think one of the key factors that I would say is as we move to these

30:14

solutions,

30:15

we don't forget the lessons that we learned from failures in the past.

30:19

And just keep in mind, for example, the lessons that I learned in my dealing

30:24

with insider

30:25

threat and APT threats, the zero trust approach has to be applied in these new

30:31

environments

30:32

as well.

30:33

And that's very data centric.

30:35

It's least privilege access management.

30:37

It's making sure that people who have access to data, they need it.

30:40

It's not broadly accessible.

30:42

And that we're monitoring some of the activities so that we, if we do see a

30:46

problem of somebody

30:47

doing something that they shouldn't in that environment, we're aware of it and

30:50

we can

30:50

be proactive.

30:51

And that applies also to the automation and the AI where auditability, I think,

30:56

is something

30:56

that we've lost our -- we've kind of lost the threat a little bit in these new

31:00

environments.

31:01

Automability of AI applications is critical.

31:03

To be able to figure out, when you do make a decision, if it's comparing two

31:08

tax returns,

31:10

that decision can be tracked back through some of the processing that led to it

31:15

so that

31:15

it's not just a black box and potentially a liability if somebody then made the

31:20

wrong

31:20

decision.

31:21

Do you think the -- something like auditability, though, depends on the use

31:27

case and the particular

31:29

risk.

31:30

Clearly, taxpayers are going to be very upset if their filings are wrong

31:35

because of an

31:36

AI.

31:37

But there are other cases where government agencies and others can use AI where

31:41

the impact is

31:41

different, right?

31:42

I'd say kind of going back to the keynote with a speaker, Nataka, was talking

31:47

about some

31:48

of the innovations.

31:50

We can use technology to help, but I don't think that there are many situations

31:54

where

31:54

we won't, at some point, want to -- or need to explain to a lawyer how that

32:00

decision was

32:01

made.

32:02

And we can't just say that the AI told me to do it.

32:05

So auditability is going to come in once it's medical malpractice.

32:09

It's maybe in the higher risk.

32:10

It will be more necessary and more detailed.

32:15

But if we can't at least explain to some degree how the decision was aided by

32:19

the technology,

32:22

we're opening ourselves to the problems.

32:24

I just had a Megan flashback.

32:26

Did you all see the movie Megan?

32:29

But the AI went -- oh, so I digress.

32:32

Okay.

32:33

I saw the trailer for it.

32:36

And the accountability was they didn't know because the AI did it, and they

32:40

didn't know

32:40

what it was doing.

32:41

And so you just kind of brought that back to my mind.

32:44

That's a scary thought.

32:45

There it is.

32:46

Yeah.

32:47

So I want to be respectful of time, but I did want to get into that point just

32:51

a little

32:51

bit more because one of the issues that -- break when it comes up in settings

32:55

like this is

32:56

how do you balance those new additional government requirements?

33:00

Some people might call them compliance, but whatever we call them, with the

33:03

desire to

33:04

move quickly on the technologies, to innovate, to compete with other nation-

33:08

state adversaries,

33:10

all of those things.

33:11

And it's a balancing act.

33:12

So I don't know -- let me start with you if you've got thoughts on how to do

33:14

that.

33:14

Well, I'll just remind you of mine, the one who asked Taka how he got the ATO

33:18

process down

33:19

to two or three weeks because I was in the DOD and I couldn't figure out how to

33:23

get it

33:23

done faster.

33:24

So it's a problem, but obviously Taka has solved it, so go ask him.

33:30

No, it's something we have to strive for, and I think one of the key factors is

33:35

a forum

33:36

like this where we can share solutions and figure out how to do better.

33:39

So I'm glad that we're -- thank you for the opportunity to join you, and I hope

33:43

we can

33:44

learn more after.

33:45

Yeah, Andrew?

33:46

Yeah, you know, I made a comment earlier about safeguarding or protecting the

33:52

legitimacy of

33:53

our government.

33:55

And I think that there's just some acknowledgement for us bureaucrats that we

34:01

rely on, we need

34:03

some of that red tape.

34:05

And that is, again, to safeguard to the legitimacy of our government.

34:10

And so, yes, there is a balance.

34:13

We absolutely want to innovate.

34:15

We want to make things easier.

34:16

We want to make things quicker, but we also need to slow down sometimes and

34:19

make sure that

34:21

we are compliant, that we're complying with the law, conforming to technical

34:25

standards

34:26

for whether it's security or for accessibility, and that is necessary.

34:32

And I think that we can also look forward to technologies that can help us

34:36

navigate that

34:37

more easily.

34:38

And so, yes, there, again, still will be a lag, but a necessary one.

34:44

And we still can look for ways to improve those.

34:46

And our challenge, of course, is, again, as bureaucrats to determine what red

34:52

tape can

34:53

I eliminate and what needs to stay and for what reasons.

34:57

And that's our job.

34:59

Great.

35:00

Thanks, Karen.

35:01

Yeah, I think that no one expects us to move at the pace that industry does,

35:06

private

35:07

sector, because we have more to lose.

35:10

We have more threat that has a more radical implication if those cyber surfaces

35:19

are, you

35:20

know, attacked.

35:22

So I think while we do have the desire to move fast, and I think we can move

35:28

faster by learning

35:29

from industry, seeing, letting them, you know, run with and learning from them

35:34

in events

35:35

like this, learning what not to do, what to do, what lessons learned.

35:39

But also, I like the idea of test and learn, you know, piloting MVPs.

35:46

And let's try a little bit and let's learn, let's dip our toe in the water and

35:51

see what

35:51

happens and then scale up as we get comfort as legislation changes.

35:57

And actually using those tests to drive changes in legislation and increase

36:02

compliance.

36:03

I think the risk for government far exceeds, especially with the IRS, what

36:08

would be tolerated

36:10

for a, you know, private sector organization.

36:13

I mean, how many times have you all received letters where there was a data

36:17

breach and

36:17

you get a subscription to monitor and life's good?

36:23

So if IRS sent you one of those, we would be tired and feathered on the steps

36:28

of the

36:29

White House.

36:30

So we just don't have the tolerance, but we can learn and we can test and grow.

36:35

Well, I did get one of those from OPM once upon a time.

36:38

I did just think we just let them.

36:39

I just let some people being tired and feathered in congressional hearings.

36:42

Yeah.

36:43

No, I, so Karen is absolutely right.

36:49

Our primary responsibility is to protect all of your data, all of our data.

36:55

Really we see about two and a half billion unauthorized attempts against us,

37:00

two and a

37:00

half billion.

37:01

We're talking about multiple millions a day.

37:05

And these are bad actors who are trying to access information that they know we

37:10

possess.

37:10

So we are extremely careful.

37:13

We are very risk averse and naturally so because we have a very big

37:17

responsibility.

37:18

However, there are two things that were said here is the approach that we've

37:22

taken.

37:23

One is exactly what Karen said.

37:25

Let's try it out.

37:26

Let's do it on a smaller scale.

37:28

We don't have to do this for everyone all at once.

37:31

You know, of a population of 400 million, we have over 200 million that are

37:36

paying taxes.

37:37

We don't have to target and find a solution for applied technology that

37:41

addresses the

37:41

needs of 200 million.

37:43

Perhaps we look at 100,000 and testing out those waters.

37:47

The other thing is what Owen said, there is a great opportunity to engage with

37:51

all of

37:51

you, with our other partners, other agencies, and learn how they're doing it.

37:57

One of my areas of responsibilities that I've now started to expand on is not

38:02

just containing

38:02

it within the United States, but frequently speaking with the tax agencies of

38:08

UK, Australia,

38:09

Canada, the ones who also have a similar scale and size of ours, how are they

38:15

addressing

38:15

some of these challenges while keeping that risk at bay and keeping it minimal

38:21

at best,

38:22

because we don't have that luxury to, again, be the one that sends out the

38:26

letter that

38:27

tells you that we're going to help you now monitor your credit for the next 12

38:30

months.

38:30

[BLANK_AUDIO]