0:00

Throughout the program all right so to get started  this morning I want to kick it off with the  

0:05

opening remarks it will be given by my supervisor  the director of the office of civil rights and by  

0:10

the way I didn't introduce myself I'm Monique  Dismuke I'm the disability program manager for the  

0:15

Department of Commerce and I'm happy to work with  you work with the team this after this   

0:22

week this month to put on this program so again  I'm going to turn this over now to Mr Jerry beat  

0:29

the director of The Office of Civil Rights take  it away Jerry thank you Mo'Nique um good morning  

0:37

everybody and welcome thank you for joining us for  today's uh fascinating program in recognition of  

0:45

national disability employment month 2023 so as  Monique mentioned my name is Jerry Beat and I'm  

0:52

the director of the office of civil rights here  at the US Department of Commerce and it's an   honor to be here today to introduce our national  National Disability Employment Awareness Month event  

1:02

when invited to provide these opening remarks  I was asked to say a few words about NADEM and   what it means to me um I have several personal  stories um both professionally and personally  

1:14

about people with disabilities in working with  people with disabilities but I think I I would   prefer at this moment to take a moment to link  the meaning of today's events to the mission of  

1:25

our agency now you all know the mission of DOC is  to create the condition for economic growth and  

1:31

opportunity for communities for all communities  let's look at the name of this commemorative  

1:36

month National Disability Employment Awareness  month I want to place emphasis on employment  

1:44

the word employment is linked to our event today  and I suspect you already know this but a large  

1:49

portion of individuals with disabilities are not  employed and in the labor market or they're not  

1:55

in the labor market there are several reasons for  this I suspect one of the biggest reasons is that  

2:01

employers may think they they can't employ uh  cannot support employees with disabilities in the  

2:06

workplace because of the kinds of accommodations  they might need or that they don't have work I.E  

2:12

jobs that someone with a disability could do  these perceptions turn into gaps we see these  

2:19

gaps when we compare the employment rate and the  individuals with disabilities to the employment   rate of individuals in the labor market without  disabilities or benchmark I bet our colleagues at  

2:29

the Census Bureau have a great data that confirms  these gaps I know they have great educational  

2:34

fact sheets on several different underserved  communities including people with disabilities  

2:39

in fact I bet it's available this month on their  homepage so I encourage you to visit it and learn   something new I know I do whenever I read their  wonderful fact sheets when I think about some  

2:48

of the reasons why individuals with disabilities  have such low employment rates I know a portion   of these reasons are due to misperception some  of which I've mentioned such as accommodations  

2:57

are expensive or a person with a disability can't  perform the work in our office well let me tell   you I know firsthand from my work as a reasonable  accommodation specialist which is what we call  

3:08

them at the Food and Drug Administration uh that  accommodations are not expensive and individuals  

3:14

with disabilities can do almost any kind of job  from being an employment lawyer to an economist to  

3:20

a statistician to a fishery biologist physicist or  patent attorney there are people with disabilities  

3:28

in all of these occupations yet this community  remains underemployed you can help change this  

3:36

here at the Department of Commerce and we know  at DOC we have a special obligation to help   create economic opportunities for all underserved  communities including people with disabilities  

3:46

this responsibility aligns with our mission and  it should be something we all think about every  

3:51

day but you probably ask how can I change this  at the Department of Commerce well here's how  

3:58

you can first if you are hiring consider using  the special hiring authority of schedule a to  

4:03

bring someone on board with a disability if you're  looking for interns consider using the workforce  

4:08

recruitment program if you are recruiting  make sure your vacancy announcements go to   Gallaudet University the Federation for the Blind and  other great nonprofits and rehabilitation centers  

4:18

that serve people with disabilities don't forget  to review and update your disability status and  

4:24

NFC's EPP you can do this and OHRM lets employees  DOC employees know about this every year in fact  

4:34

I think we just received a broadcast about this  remember any of us can become somebody with a  

4:39

disability at any points in our lives if you  have a disability disclose your disability and  

4:45

help educate your colleagues about what that  means if you are a supervisor and an employee  

4:51

discloses that they have a disability and they  want something to help them do their job AKA  

4:57

a reasonable accommodation just do it and trust  your reasonal accommodation coordinator to guide   you through the reasonable accommodation process  remember changes in the workplace requested by  

5:07

people with disabilities more times than not  end up creating more efficiencies and better  

5:13

approaches to doing work and they benefit  all employees I want to share two facts  

5:18

about unemployment of people with disabilities  here at the DOC. The EOC's Benchmark for federal   agencies is to have 12% of their total workforce  be made up of people with disabilities here at  

5:28

DOC according to the Fiscal Year 22 MD715 report  prepared by OCR we are at 10.97% this means that  

5:37

we fall below that Benchmark we also fall short  of this Benchmark at the higher grade levels  

5:43

through the SES the same Benchmark of 12% so we  need everyone here across the department to take  

5:50

take all the steps I mentioned a few minutes ago  to help us meet these EOC benchmarks and to narrow  

5:56

the employment gap secondly we can do it here  at the DOC I am confident the EOC has a second  

6:03

Benchmark for agencies they want all federal  agencies to have 2% of their workforce to be made   up of people with targeted disabilities which are  specific groups of disabilities such as blindness  

6:13

mobility impairments death and hard of hearing etc  when we look at this figure here at DOC and compare  

6:19

it to The Benchmark of 2% we exceed that number  at 2.49% and believe me not many agencies can  

6:28

say that let me clarify there are some bureaus  within DOC that do not beat this Benchmark but  

6:34

I'm confident we can change that so thank you let  me turn to today's exciting program which I'm sure  

6:41

will reate several of the ideas and points I just  shared with you we're talking about something very   cutting edge this year's theme is advancing access  and equity the theme coincides with the Department  

6:51

of Commerce's DEIA goals for building accessible  equitable inclusive workplaces that welcome the  

6:58

talents of all all people including people  with disabilities today's special observance  

7:04

event is designed to highlight the value of AI  yes AI artificial intelligence sure we've all  

7:11

heard a lot about that lately in increasing  opportunities for employment participation  

7:16

economic self-sufficiency and independent living  for people with disabilities we will also have an   opportunity to learn more about uh difference in  the term neurodiversity when used in the context  

7:27

of disability and when there is no disability  our three panelists they all have some degree  

7:33

of expertise in working with AI the panelists  will share how AI technology is contributing  

7:39

to the mission of the Department of Commerce they  will also describe how AI is being used to advance  

7:45

opportunities for neurodivergence Divergent  workers that's exciting so our guest panelists  

7:52

are Mr Andre Mendes the Chief Information Officer  for the office of the secretary Miss Victoria Houed

7:58

director of AI policy and strategy Office of the  under secretary for economic Administration Ms. Elham   

8:05

Tabassi the associate director for emerging  Technologies information at the information  

8:11

technology laboratory National Institutes for  Standards and Technology and I'll stop there  

8:16

and turn it back over to OCR's own disability  Employment Program manager Monique Dismuke who  

8:22

will introduce to you the moderators who will be  leading the discussion thank you all and I hope   you enjoy our exciting panel this morning Jerry  thank you so much for that wonderful opening we  

8:34

are you've excited us I think you've got our  our uh pallet wet for this information today  

8:40

so we are really excited about this thank you so  much so without further ado I'm going to actually   turn the program over to our two wonderful  moderators Ms Ann Ngo is the president of  

8:50

the Department of Commerce Asian-American  employee Resource Group and assisting her  

8:55

today is Miss Sharon Williams and Miss Sharon  Williams is the co-chair of the differing  

9:02

abilities um employee Resource Group here at  the in HCHB so without further ado I'll turn  

9:10

this over to our wonderful moderators who will  take the program forward from here thank you so much [cross talk] Ann did you  

9:28

want to go first uh please go first Sharon oh hi  my name is Sharon Williams I'm the co-moderator  

9:36

for this event and I work as an administrative  assistant um in the office of uh Chief Financial  

9:43

and Administrative officer at the FirstNet  Authority um I'm also a neurodivergent   employee and and I'm the co-chair for the HCHB um  differing abilities um ERG to um our meetings are  

10:02

are monthly and uh on the fourth Thursday of the  month on October 26th of anyone if people would  

10:10

like to come to and join our ERG so let me give it  back to an hi my name is Ann Ngo I am the President  

10:20

of the Department of Commerce Asian-American  native Hawaiian and Pacific Islander ERG employee   Resource Group uh I added the native Hawaiian  there uh as a nod to my uh colleagues uh who  

10:30

currently don't have an ERG of their own uh but  our our name actually is Asian-American Pacific   Islander ERG and uh you some of you may ask why  would the AAPIERG be a co- uh sponsor uh  

10:45

of this event well uh among our uh population in  the United States we all have what I would call  

10:52

intersectionality we are more than one thing we  are more than uh uh we are more than than just our  

11:00

heritage we are more than just our jobs many of us  have um challenges and abilities that are that go  

11:09

unrecognized because they come in a package that's  un um that that people do not see as fitting the  

11:16

norm so many of of those who are brilliant and  talented and working in different scientific  

11:23

and Technical Fields uh could be considered  neurodivergent and so in addition to being  

11:28

a minority um they may also behave and learn and  communicate in ways more differently than others  

11:37

and I would like to take this opportunity uh to  highlight the fact that no matter who we are and  

11:42

what background we are from our talents and our  contributions are much needed especially as um  

11:50

artificial intelligence is going to be changing  the landscape of the workplace tremendously and   you will hear from our panelists who uh all of  whom have been invited because they have uh their  

12:00

expertise definitely but they have a passion for  diversity equity inclusion and accessibility and  

12:07

so I um I hope you will enjoy today's session this  is the first but not the last we plan to continue  

12:14

co-hosting co-organizing more uh intersectional  panels of artificial intelligence and DEIA as it  

12:23

pertains to the Department of Commerce uh goals  in uh you know diversity equity inclusion and  

12:29

accessibility and I will pause there and I will  invite actually we have uh two of our panelists on  

12:36

uh screen uh the third one she is incredibly busy  and she will join us when she can but um you know  

12:43

I send her my best wishes and uh uh Good Vibes  oh there she awesome yeah she's in a session she  

12:51

just had have a camera on so I will ask in turn  if uh please Andre Victoria and Ellen could please  

12:59

take the time to introduce yourselves and I hope  that our moderators uh excuse me our um supporters  

13:05

IT uh friends have dropped their bios in the chat  for all of you I'll pause here uh Andre please  

13:12

kick us off thank you uh very good well thank you  so much uh for uh inviting me to participate uh in  

13:19

this fantastic exercise and first of all I wanted  to mention two things Larry's introduction nothing  

13:25

short of spectacular so much great information so  much energy Larry thank you I appreciate it the  

13:32

all the work that you're doing the second one is  the attendance I mean we passed 99 people a long  

13:37

long time ago and they keep coming I love it I  love it people are interested in this topic uh  

13:43

people are committed and this is fantastic um  so I'm Andre Mendes I'm the Chief Information  

13:49

officer uh for the Department of Commerce and  I have been deeply involved in dealing with  

13:55

individuals with varying capabilities uh for a  long long time for three years I was the senior  

14:01

vice president for strategic planning and CIO for  Special Olympics at a global basis on a global  

14:07

basis so uh it goes It goes far back in terms  of of my interest and desire to help uh in in  

14:15

this particular arena I will give you just a very  brief uh you know thought process in terms of how  

14:21

I think about uh neurodivergence I think that  by and large we are all neurodivergent right  

14:28

we all have very different uh you know spectrums  of uh of outgoing you know uh personalities you  

14:36

know introversion versus extroversion we have  different reaction to external stimuli we have  

14:42

different abilities to process auditory uh you  know uh samples we have different abilities to  

14:49

to read from you know from a dyslexia standpoint  for example so we're all dealing with all kinds  

14:54

of different things on different planes uh and  some of them are are more considered than others  

15:01

but the reality is that humanity is a spectrum  uh and that Spectrum applies to every single one  

15:07

of us so it it is absolutely incumbent upon us  uh to accommodate and to work with and leverage  

15:15

especially leverage uh the incredible capabilities  and and contributions uh to humankind that can  

15:22

come from all kinds of people with all kinds of  of neurodivergence uh because we're all in that  

15:28

spectrum and with that I'm going to pass it to  Victoria for uh her opening remarks that sounds  

15:36

great thanks Andre uh so my name is Victoria  Houed I'm currently director of AI policy and   strategy working under Oliver Wise the chief  data officer uh in the under secretary's office  

15:46

at Commerce uh before this I was working with  Schmidt Futures building out a policy a Science   and Tech policy initiative there and then uh  before that was working for Nancy Pelosi as a  

15:56

policy adviser and before that was actually an  engineer uh this topic in particular is  

16:01

incredibly important to me I've always been kind  of working for underrepresented uh communities  

16:06

slash before moving to DC used to run a nonprofit  for black women in technology and am myself and  

16:12

a divergent person who had to deal with a lot of  accommodations and figuring out how to kind of   make my path uh into the federal government and  into um more traditional neurotypical spaces um so  

16:24

I'll be talking a little bit more about that today  but it's really great to be here and thank you all

16:34

passing it to elham you're on mute

16:40

elham good morning everyone thank you so very much  for including me in this important conversations  

16:49

uh first I want to uh uh again in addition  to thank the uh organizers of the event for  

16:55

including me but also all of the participants  for spending one hour uh uh listening to us and  

17:01

our conversations extremely important topic um  I uh work at NIST I have been with National Institute  

17:07

of Standards and Technology since uh 1999 uh  my background is uh computer vision machine  

17:14

learning and in the past uh five seven years I've  been leading the NIST trustworthy responsible AI  

17:21

program uh what we are trying to do as part  of that program and as part of the output   and outcomes that we produce is uh maximize The  Beneficial use and maximize the positive impact  

17:33

of the AI system so AI systems can uh uh serve  all people in a responsible fair and Equitable  

17:40

way while minimizing their uh negative uh impacts  uh their risks and uh negative consequences AI  

17:48

systems has enormous potential to improve our  lives uh they also can cause cascading negative  

17:54

consequences they can bring the uh bias back into society and sort of Auto automate them  

18:01

and uh with that um kind of again exasperate  those negative consequences um so I think it's  

18:08

uh becomes a shared responsibility for all of us  uh involved in different uh cycles of the AI uh  

18:16

life cycle uh those that uh design and develop  algorithms those that uh uh uh procure or make  

18:23

decisions about what AI system to use those that  deploy AI systems and those that use AI systems  

18:30

to be aware of the risk of AI systems and uh  have a good risk management that's also right  

18:35

space in plan to maximize benefits of AI while  minimizing the risk of AI we have produced  

18:43

several documents uh from understanding bias in  AI systems to security to explainable AI

18:49

systems and then uh January 2023 directed  by Congressional mandate we released the AI risk  

18:57

management framework work which is a voluntary  framework for managing the risk of AI systems in  

19:02

a flexible structured and measurable way uh since  then we have been working with the community on  

19:09

um um basically informing them of this tool exist  and helping them to operationalize AI RMF in AI  

19:17

RMF it talks about what are the risks of AI uh uh  uh that's the uh guidelines in the map function  

19:25

the measure functions provide guidelines on  how to to measure the different risks of AI  

19:30

uh systems the manage functions provide  guidelines on uh how to prioritize risks  

19:36

what are the risk mitigations uh uh processes  uh and also emphasizes that zero risk doesn't  

19:42

exist so let's make sure that how we can um uh  continually monitor the risks to be able to uh  

19:50

basically stay ahead and uh uh have the right  safeguards for the use of AI systems thank you

20:00

all right and so uh I I forgot to say this  and I guess you're too modest to say it but  

20:06

I believe uh Elham you were nominated or maybe  honored by Time Magazine uh in the AI field so  

20:13

you do us proud thank you so much for representing  Commerce so nice of you I I keep saying it um that  

20:22

was a nod to public service because we all know  that public service is an honorable profession

20:34

would we like to go under our first question  for the panelists um okay um please tell us  

20:44

how tell us about yourself and your experience  with working in and working with AI who would  

20:52

like to start Victoria Elham what why don't  you guys start since I went first in my

20:59

introduction I mean I kind of already gave a  a high level but so I currently right now am  

21:09

kind of trying to turn my career into from  like engineer into um Tech Policy Guru and  

21:18

have really focused my last couple of years into  emerging Technologies like artificial intelligence  

21:25

I think a lot of that interest kind of comes from  um initially my degree was in systems engineering  

21:31

but then kind of has shifted a bit to policymaking  just because coming from a lot of different types  

21:37

of communities it's been really important to me to  kind of be the voice in the room um representing  

21:43

a lot of those different types of communities  where I kind of come from as a black woman who   did engineering and a neurodivergent person etc  um and so I think within my initial career in  

21:53

DC policy making under the speaker for example um  often times it was was a lot of folks who were  

21:59

you know from more affluent families or who only  lived on the east coast and I'm from the Midwest   representing a lot of like Farmland etc um and  so I think coming into the AI conversation has  

22:09

been really interesting because most of the the  conversation within the government is around bias  

22:15

and how do we represent these communities that  are not oftentimes represented um and so this   area has always been kind of interesting to me  just because of the fact that it really needs  

22:24

young Voices it needs voices that are um just  different than the stereotypical community that  

22:30

is uh that is typically seen within the industry  right now especially because it is so nascent and  

22:36

so my work right now with the chief data officer  is related to uh building out guidelines for how  

22:43

the government puts out its data and so we're  trying to figure out how do we get how do we   move our data from being just machine readable  and how do we kind of meet the moment with AI  

22:54

and make it more understandable by AI models  and by these systems that are building these   large language models etc and so we're working  with companies like Open AI and Google to kind  

23:04

of figure out if every American one day will be  kind of working with these large language models  

23:10

and getting information out of them how do we get  them to utilize our very accurate Commerce data  

23:16

so for example if you're looking up like what is  the uh population of New Orleans right now it will   give you some random number from a random website  that just has very good SEO but how do we get them  

23:26

to prioritize Census Data and a lot of that can  kind of feed into this larger problem of how do  

23:33

we get access to data to every single individual  within the country how do we get um people kind of  

23:39

thinking about how like basically being able  to just do simple data analysis on data that  

23:46

they're not usually able to have access to using  AI and so yeah that's kind of at a high level of  

23:52

kind of where I am now and in my journey into  coming into working on artificial intelligence  

23:57

great thank you well I think I also alluded to  if the question is that how I get to the AI uh  

24:07

uh to the field of working in AI so that's that's  has been my background uh um my educations and my  

24:15

first uh project at NIST when I joined NIST as I said  June 1999 was development of a open vocabulary  

24:23

speaker recognition my task was to develop a  algorithm then when you walk in the room and say  

24:30

computer bring up my calendar and that was exactly  what I did that was the demo that I did uh the

24:37

the computer can recognize from the voice of uh  the speaker that this is Elham's calendar versus  

24:43

Andre's calendar or Victoria's calendar and bring  up the right calendar up speaker recognition uh  

24:49

uh is is uh you know at that time I I trained  a model to do this and then uh I get into the  

24:57

to work in the the biometrics after the uh 911  uh formation of Department of Homeland Security  

25:03

NIST was tasked to certify biometric uh systems  for entry exit program and and I work in that  

25:10

area and uh developed again based on machine  learning algorithms to uh look for the quality  

25:16

of fingerprint so if you have your PIV cards you  give fingerprint there uh there was a system that  

25:21

looked at the quality of fingerprint that was the  system that I developed and face recognition Iris   recognition but why why responsible AI so I was  coming from all of you know these years working  

25:31

on machine learning and AI but why responsible  AI the way the way we are working working in  

25:36

this field I guess um again going back to 1999  when as young as a younger scientist I was  

25:45

developing those things uh I would develop models  and build systems because I could because um I do  

25:53

the research develop the systems the question of  of is it fair what's the impact who is going to be  

26:01

us it and how the users is going to be impacted by  that was not really in my mind I think it was by  

26:08

growing up having children seeing the uh impact of  the technology on people uh getting more educated  

26:16

and more informed we all learned that how uh uh  the more technology is being ingrade in our um  

26:23

in our lives every day uh we expect more from  the technology Alondra Nelson at one of the NIST

26:32

Workshop very eloquently said uh we don't demand  justice and fairness from each other but also  

26:41

demand justice and fairness from our machines  and technology and just think about it for a  

26:46

little that's that's huge right um technology  is not just uh uh a a u about data on Computing  

26:56

systems is about data Computing systems and their  interactions with environment with human um uh so  

27:04

so the tremendous uh kind of responsibility that  uh uh we all as anybody who is involved in any uh  

27:14

uh stages of the life cycle of the technology  we have a responsibility to make sure that   the systems uh are fair and just but also demand  fairness and justice of our systems and algorithm  

27:29

uh very good those are fantastic uh so uh my  involvement with AI uh goes back as as much as I  

27:37

hate to say it about 40 years uh I've been around  that that long in in the professional environment  

27:44

uh first start working with AI actually you know  machine learning which is really where where the  

27:49

uh where the Crux of the the questions are uh  back in the late 80s uh you know correlating uh  

27:55

health promotion data and individual data with  then outcomes from a medical standpoint to try to  

28:02

figure out what were the patterns of behaviors  and genetic uh you know antecedents that would lead  

28:07

to higher risks of uh of uh of for example  cardiovascular um incidents uh neurovascular  

28:15

incidents uh all all kinds of of bad outcomes so  so it goes back a long way and since then there's  

28:21

been a variety of of organizations that I've  worked with that are always using machine learning  

28:27

in order to perfect their ability uh to analyze  data to look at data and come out with reasonable  

28:33

propositions either for further research or  even for the creation of market marketable  

28:39

products uh and uh I had the opportunity to work  at the beginning of the century in 2001 with a  

28:46

biotechnology company that was heavily involved in  the uh in the decoding of the genome and a lot of  

28:53

that was driven by Machine learning and heavy  duty machine analysis of patterns of um of uh  

29:00

bases uh in the DNA sequence that would allow you  to to start doing some prediction in terms of not  

29:07

only the next set of bases that would be coming up  but eventually the decoding of the entire genome  

29:12

more recently of course uh you know as that that  industry and that science uh continue to evolve  

29:20

uh very much involved into uh you know what it  means to productize uh artificial intelligence to  

29:27

leverage artificial intelligence in our day-to-day  living and of course at Commerce we have amazing  

29:33

examples of how our bureaus are leveraging uh  in machine learning in order to better faster  

29:40

and cheaper uh execution of their missions and  Charters I mean a prime case is the United  

29:46

States patent and trade office where they have  been using machine learning for quite a long   time in order to optimize uh the the the work of  the uh of the patent examiners uh and to and to  

29:58

make it uh easier and more efficient for them to  perform their duties so uh go back goes back a  

30:05

long time but there is definitely a pattern where  you know you look at pattern recognition in order  

30:10

to enhance algorithms and then move forward in  terms of effectiveness uh and and capability in  

30:17

recognizing certain circumstances and adapt the  uh the environment uh around them uh to produce  

30:23

better outcomes uh you know from a variety of  standpoints as I mentioned Health Care efficiency  

30:29

in the workplace safety and of course nowadays  we're seeing a lot of influence into the cyber  

30:35

security arena uh where we are trying to leverage  artificial intelligence in order to do heavy duty  

30:41

pattern recognition uh and and matching uh so that  we can detect uh you know anomalies uh that will  

30:49

potentially be driven by by nefarious actors  trying to do us harm of course unfortunately  

30:55

on the other side of the equation those nefarious  actors are leveraging AI in order to figure   out how to penetrate our systems but that's part  of the tradeoff and the negotiations that Ms  

31:06

Elham so uh eloquently described in terms of  the positives and the negatives that come from  

31:12

any uh you know uh scientific uh or you know  evolutionary uh environment back to you Ann and  

31:21

and and and Sharon well you know that this is  fascinating and uh I am so happy the three of  

31:30

you agreed to have a conversation with us because  um you anticipate that the question I was going   to ask and I'm sure uh many people in the audience  would be interested to know is how Commerce the  

31:40

Commerce Department is using AI but before we get  to that I just want to uh pick up the point about   that we must demand um Justice and Equity from  our machines as well as from our systems I think  

31:52

uh in order to do that we need transparency we  need access and we need a way to understand how  

31:58

codes are written and who's writing them and what  data set goes into them that goes back to what 

32:05

Victoria was saying you mentioned PTO using machine learning and AI already and there's also a

32:11

controversy right now internet about machines AI  using author's book and their work and stealing it  

32:21

or or students who might put up their uh papers  to do spell check for example might have their  

32:27

work be subsumed by the AI uh data set or database  and then spit back out that you just plagiarized  

32:33

somebody even though that somebody might be  yourself so there are lots of safeguards that   need to be put in place but I'll go back to uh  the question that may be very pertinent for all  

32:43

of us working inside of Commerce is how is the  Commerce Department using AI where is being used  

32:49

and uh and how do we get insight transparency  and accountability into that process thank you  

32:58

okay I've already talked a little bit about it  Victoria do you do you want to tackle this one   first I mean I'll be I feel like my my response  will be very different than Elham's because I  

33:07

feel like I'm a little bit more on the innovation  side versus trying to mitigate risks so I'll kind   of cover that portion um so I think I had already  mentioned like our work right now in exploring how  

33:18

we utilize all of our Commerce data for the sake of  mitigating misinformation in a lot of these large  

33:23

language models that already exist kind of what  like addressing a lot of the problem that you're   kind of identifying right now which is trying to  get more fact-based information versus the types  

33:33

of information that potentially might find on  Wikipedia or I mean not that that's necessarily   inaccurate but other websites that potentially  just have higher hits and or are higher up on  

33:42

Google or what have you um so that's number  one number two I think we're also exploring  

33:48

uh and this is all I'm speaking on the work that  I'm kind of been a part of is thinking about how   like what are the tools that we can use to kind  of make our work a lot more efficient so um we're  

33:57

thinking about like should we be creating our  own internal secure environments for like large  

34:02

language models that we can actually pull together  the types of all of our private information and be   able to do our work a lot better and faster um so  I think there's like a lot of topics right now on  

34:12

like how can we improve our efficiency in a way  that's also protecting all of our data slash um  

34:17

you know kind of working on like a role-based  um like dependent on the user what data does  

34:22

that person have access to within Commerce um but  obviously I think Elham used to maybe speak a bit   on um maybe like the AI labs and the research that  they're doing related to the risk-based efforts

34:37

happy to do that Andre do you want to go first or  you want no please by all thank you um so we are  

34:45

as Victoria said we are not on the operational  operational side of doing it but I just want  

34:51

to say that at least at NIST we have started  working with our uh office of information  

34:57

security management or OISM on implementing AI  RMF in trying to basically if they decide to  

35:04

use deploy any AI based tools for their operation  what are the considerations and uh they need to  

35:11

take into account what questions they need to ask  what information they need to gather to make sure   that it's a it's uh the risks are understood uh  measured and managed um I uh and and we are uh  

35:27

we will definitely share those experiences and we  look forward to work with um Commerce colleagues  

35:34

I think we're already talking with the uh data  governance board on uh some of the uh possible  

35:43

uh again considerations and questions that may we  want to ask about um uh the data quality and data  

35:49

uh governance um and you you brought up a several  times a really good points about transparency  

35:58

uh and uh and that's really important and AI RMF  talks a lot about uh transparency transparency can  

36:04

enable it's important for enabling accountability  the way we have uh defined transparency uh is uh  

36:12

it's about documentation of all of the different  activities and decisions made across the AI life  

36:18

cycle so if somebody is uh you know the data  that gathered for deploying um I I mean training  

36:25

or um testing for the deployment um making sure  that data is valid and reliable bias issues has  

36:34

been considered uh understood and measured you  brought up another really good point about data  

36:40

uh data set representativeness uh uh making  sure that the um the users uh or the people  

36:47

that are being impacted by systems they have  been represented in the training uh environment  

36:52

and the training data um uh so uh I don't want  to get too technical but some of you may have  

36:59

heard about data shets for data cards or model  cards uh the transparency includes all of those  

37:04

mechanisms and tools but also uh uh bigger  and more kind of in-depth uh uh detailed  

37:12

documentations uh it's also important to know  that what uh you know how much information is  

37:17

needed to be shared right transparency to whom  for what purpose um and uh and and related to  

37:26

topic of transparency Ai RMF also talks about  explainability and interpretability so before  

37:32

we work on AI RMF or maybe still some Community  are using these three concepts as interchangeable  

37:39

but we have distinct definitions for that so  transparency gets about the documentations um  

37:45

uh explainability answers the questions of what  and gets into the explanations of how model works  

37:52

and then interpretability answers the questions  of uh of how and the context on and you context of  

38:00

use and meaning of the informations and the output  of the AI systems the reason I said that is we don't  

38:07

necessarily need to know uh all of the parameters  and weights and architecture of the model to be  

38:15

able to have the right level of transparency or  be able to do the right level of the evaluations  

38:21

to understand uh models or systems functionality  and trustworthiness um but if we have uh enough  

38:29

transparency on on what are the things that has  been done the documentation of what are the things   that has been done and enough interpretability  of the context and meaning of the decisions uh  

38:39

that that could that could go a long way on uh  being able to uh understand and measure systems

38:46

trustworthiness great well I guess I can I can  give you some uh some examples that we are uh  

38:58

starting to work on we've already talked a little  bit about USPTO and the work they're doing uh in  

39:03

helping the examiners with the patents but there  are a myriad of other initiatives that are taking  

39:10

place in in the Department of Commerce that use uh  artificial intelligence you know machine language  

39:17

um and in order to to bring efficiencies to the  table for example you might not know but MBDA is  

39:25

actually been using a chatbot uh that is driven  by Machine learning in order to better address uh  

39:31

the questions from all of their constituents  without having to have a massive increasing  

39:36

staffing that is constantly uh you know answering  these questions so that chatbot has been deployed  

39:42

over the last two years and it has been very very  successful um ITA is right now the International  

39:49

Trade Administration is dealing with the project  to better uh to enhance the matching of business to  

39:56

business relationships uh you know for the  people who are trying to do exports and and  

40:01

the people on the other side doing imports uh  so that you have a faster progress towards a  

40:07

a workable negotiable uh and profitable mutually  rewarding uh business relationship when it comes  

40:13

to to increasing exports of American Products  NOAA is doing fabulous work in a lot of  

40:21

areas including of course weather forecasting for  large scale events such as uh you know hurricanes  

40:27

and tornadoes predicting paths comparing uh  you know previous uh you know um uh routs  

40:34

and speeds um wind speeds uh and uh and wave  elevation uh with with the history uh in order  

40:43

to predict the patterns based on similarity of  uh atmospheric uh you know conditions they're  

40:49

also doing fantastic work in terms of the acoustic  detection of marine mammals so that they can use  

40:56

AI uh you know/ machine Lang learning in order to  recognize the migration patterns for example of  

41:03

marine mammals and where they are and eventually I  believe even certain specific individuals because  

41:09

much like human beings we all have a different  voice pattern right um uh they're also doing  

41:14

fantastic work in terms of doing optical surveys  of large fish uh schools and determining species  

41:22

and count within those schools so that work is  going to be fascinating because it's going to   allow them to see uh the sustainability of the  of the ecological environment when it comes to  

41:32

Fisheries when it comes to uh to to the uh the  the you know the the fish populations migrating  

41:38

through different uh uh areas of the uh of the  bodies of water so I mean it it is absolutely  

41:44

amazing and fascinating how it can apply to such  a large scale uh you know usability within the  

41:51

Department of Commerce and all of its bureaus  some of it with massive massive uh condition uh   you know um outcomes positive outcomes imagine  for example um that in the midst of an outbreak  

42:02

of tornadoes in the mid West in the famous  Tornado Alley that you're able to identify  

42:08

each and every one of the occurrences and start  having very high quality prediction of their  

42:14

paths right so that you can warn populations you  can get them out of harm's way uh you can conduct  

42:21

evacuations faster than ever has been possible  before and therefore you know uh you know limit  

42:28

the losses to property rather than human lives so  the the opportunities are absolutely outstanding  

42:35

um and and this is just the tip of the iceberg at  Commerce and of course the tip of the iceberg in  

42:40

the industry out there uh you know in the medical  field which is my background I believe that the   ability uh to start uh you know resolving uh you  know for example uh genetic uh diseases uh that  

42:53

are influenced by one or two uh genes uh you  know and by by doing correlation and machine  

42:58

learning processes in terms of you know protein  folding and all of those things that go into the  

43:04

bioinformatics arena is absolutely amazing and  we're going to see a lot of those diseases being  

43:09

eradicated uh at least in the Western World to  start with but eventually I believe across the  

43:15

entire uh globe so very very exciting stuff  very very exciting stuff all the guard rails  

43:20

need to be there without a shadow of a doubt  but but let me pause it that with every single  

43:26

introduction of Technology okay going back to uh  to Australopithecus and the origins of humanity  

43:34

there have always been good uses of technology and  bad uses of technology so with AI it's nothing new  

43:42

is just that the consequences are potentially uh  you know substantially higher so we need to be  

43:47

very very aware uh because the future of humankind  might be at stake so you know the the the the the  

43:54

um um the the potential for for major uh you know  conditions is very very high so there you go um  

44:03

we have lots of questions in the in the chat  um um the first one was about from I think had  

44:13

something to say yeah if I can just quickly jump  in and react to something on the chat about uh  

44:20

inventory of AI system for Commerce and I just  want to give a little bit of a context to that  

44:25

that that inventory is response to executive  order 13960 that was published in uh December  

44:34

I believe of uh 2020 2020 right I think so or  or maybe 2021 um uh reporting on the research  

44:47

activity was exempted from that inventory so when  you look at that inventory you may not see any of  

44:54

the NIST work there because again because research  was exempted and it was more around operational  

44:59

I just want to give that context thank you yeah  absolutely and my just jumping in because we are  

45:05

uh running out of time I want to ask everyone's  preference right now uh how which one of you have  

45:10

a hard stop that you must jump off and which  one of you can stay for a few minutes uh so   that we can uh help answer as many questions as  possible unfortunately I do have a hard stop at  

45:19

11 my apologies but uh you know it's a very very  busy day every day is very busy I need machine  

45:25

learning to help me with my business so so if you  do have to jump off I do want to sneak in this  

45:30

question because it is about neurodivergence  after all the the theme of today's panel um and  

45:36

my apologies to to everyone for jumping in like  this but I I do want to ask um that question and  

45:41

then if we have time we will go uh into the other  order and answer other people's questions but in  

45:46

what way do you see neurodivergent individuals  contributing to the mission of the Department of Commerce okay um well maybe you can tell stories  if you have stories or specific examples um I I  

46:00

I'll be glad to to to go first but but I will  tell you this I I mean when when we look at   humankind right and and the concept that we're  all a little neurodivergent right we have to  

46:10

understand that some of the biggest contributions  in fields like like science uh and technology and  

46:18

uh and and then the Arts and the Performing Arts  have been given to us by individuals with neuro-  

46:24

divergent characteristics right Albert Einstein  I mean how much higher can you get right Albert  

46:30

Einstein had a lot of characteristics uh of  of his behavior that were you know coincident  

46:36

potentially with being in the autism spectrum  right Charles Darwin you know the the uh the  

46:41

the individual that gave us sort of the the  theory of evolution to start with uh was an  

46:47

individual that had an enormous amount of societal  anxiety and other traits absolutely consistent  

46:52

with neurodivergence Nicola Tesla you know  one of the brightest minds ever to inhabit the  

46:59

Earth uh was an individual that displayed traits  associated with the obsessive compulsive disorder  

47:04

right and we could go on and on and on Vincent  van Gogh Sir Isaac Newton even people you know  

47:11

a little more mundane like Dan Aykroyd the the  uh the uh the actor from Saturday Night Live  

47:17

and Ghostbusters uh who have that Tourette  Syndrome right so uh the ability of people  

47:23

that have certain neurodivergence to come to  the table and contribute in a spectacular way  

47:29

is something that we absolutely cannot ignore  and that we absolutely need to foster right   there's some savant uh you know cases where  because of certain of the conditions of the  

47:39

individual there are traits that are I want  I don't want to say exaggerated but augmented  

47:45

right they provide them with amazing geospatial uh  you know consciousness and capability that lead to  

47:52

innovation in a variety of areas so leveraging  all of these people and the beautiful spectrum  

47:58

of of of abilities that exist in in humankind it's  got to be one of our one of our uh main objectives  

48:06

because we are all different and some of the  people that are sometimes maybe considered to   be on the fringes of those spectrums have amazing  capabilities to bring to the table Innovation and

48:17

progress um there's a question in the chat  how does AI impact diversity equity inclusion  

48:28

and accessibility and are there current  efforts within DOC to leverage AI in these

48:34

areas um I'll I'll let my my panelists address  that but I think Elham had something that uh  

48:44

that she wanted to say that maybe was a follow on  yeah but it it was uh I don't have anything to say  

48:50

to this particular question because I don't know  what's happening across the Commerce but they just  

48:57

want to say this that um what we learned through  development of the AI RMF is when you increase the  

49:04

uh diversity of thoughts um uh the not only the  demographic background but the thought process the  

49:12

the life experiences uh in any process uh magic  happens and for the AI particularly we're talking  

49:19

about making sure that the technology is usable  for everybody is uh fair and responsible towards  

49:28

every individuals and groups and communities it  becomes extremely important to make sure that all  

49:34

these groups are involved or we're going to get  technologies that are going to miss a portion of  

49:41

the society and we all know that that's that that  how maybe even disastrous that can be yep yeah I  

49:48

also want to add on to that too I think this is a  really great opportunity to talk about talent and   Workforce in Commerce and the ways in which we  do our hiring here and just across the federal  

49:57

government in general I think forcing people to go  through and I'm speaking from experience USA Jobs  

50:04

uh where you're kind of pushed out and you don't  really know why and it takes a very long time to   hear back um and there's no opportunity to kind  of flag whether you are or aren't neuro-  

50:13

divergent I think the big tech companies have kind  of settled for this they have their own kind of portal   for neurodivergent folks oftentimes when they're  applying and they have other opportunities to kind  

50:23

of show their skills and their talents besides  their resume most neurodivergent people again   speaking from experience struggled to kind of  figure out the ways in which they didn't fit  

50:32

into typical society slash oftentimes maybe didn't  have the best grades during a specific period of  

50:38

time and then they kind of dealt with their you  know whatever dyslexia etc um in the you know  

50:43

in their own time frame and they've learned how  to kind of fit into society but our our typical  

50:48

application process doesn't really work for that  type of archetype of person so there should be  

50:54

other ways that that people can kind of come in  and show their skills uh without having to to go  

51:00

through these very typical processes and I might  myself have pretty much never gotten a job through  

51:05

any of these portals just be because of the fact  that oftentimes I'm booted out of the system even  

51:10

here at Commerce I'm here on IPA and that was  through um really great networking and etc uh but  

51:16

it's it's very unfortunate that that we don't have  really a process for that yet and I think we could   definitely do better to work on that so uh one one  of the projects that we're working on uh you know  

51:27

trying to leverage AI it's one of our test cases  is in the creation of position descriptions of  

51:33

course and when it comes to creating position  descriptions uh you know being being you know  

51:38

very cognizant of of diversity inclusion and  and equity is very very important right um  

51:45

because there were certain uh you know traits  and characteristics that were required before   and even uh you know qualifications that were  required before and that are no no longer part of  

51:53

the equation because people realize that uh that  there are there were certain concepts about for  

51:59

example a college degree right having a college  degree being so important for the deployment of  

52:05

of of just about any job and that's really not a  reality I'll tell you on a personal note right I  

52:11

I got my undergraduate degree when I was 48  years old okay I had lectured at MIT before I  

52:18

got my my undergraduate degree and I got my  masters when I was 50 right everything that  

52:24

I've done before that wasn't worthwhile because  I didn't have a college degree that's ridiculous   right so we need to make sure that we have those  things at the table and understand that right now  

52:33

on the other side of the equation you're going  to see AI being applied without a shadow of a   doubt right it probably is already is a lot of  machine learning being done for for looking at  

52:43

resumés right and looking at CVs and and doing  calculations that if that engine is biased are  

52:50

going to absolutely be terrible for anybody who  doesn't fit a particular profile of somebody uh  

52:57

that the the machine thinks it's it's optimal  right so we need to be we we need to be willing  

53:03

to use them in order to improve the process so it  doesn't take forever to hire but we also need to   be extremely cognizant that we do so we need to do  so in a very deliberate methodical and systematic  

53:14

manner so as to avoid any kind of bias that could  potentially take off the table amazing talent that  

53:21

does not fit a specific profile that uh that that  historically has been potentially considered the  

53:28

most desirable but it's necessarily not  necessarily the one that will bring the best  

53:36

a lot more questions for you and the audience have  a lot more questions but uh to respect your time   you know please uh log off when you can and uh if  anyone of the panelists can stay a little longer  

53:46

or moderators or anybody can stay longer uh to for  the conversation um I I want to welcome you but uh  

53:53

you know at the 11 o'clock sharp the program  will end but that doesn't mean that it's over  

53:58

so uh anyone who can want to stay let's continue  the conversation thank you so much everyone and   uh Sharon my apologies for upsetting the Apple  cart and changing the order of everything but  

54:10

no problem run out it worked it worked so for  whatever it's worth uh you know send me send me  

54:15

a message send me a an email and I'll be more more  than glad to give you my opinion that doesn't mean  

54:20

it's it's it's Department of Commerce policy but  I'll be more to glad to give you my opinion I'm  

54:26

sure the other panelists will think the same right  and uh and you know and there's always been with  

54:32

the introduction of Technology a fear that people  will be displaced right that has always been the  

54:38

case so what we need to do is make sure that uh  that just because that there's a fear that there's  

54:44

going to be a realization of that fear we need to  continue to strive to make sure that you humankind   is fully engaged in its life right regardless of  our machines come to the table and historically  

54:56

we've been able to succeed in that by and large  okay not not optimally but nothing is optimal so  

55:02

we just need to keep continue to be engaged and uh  what a pleasure and and and I wanted to thank you  

55:09

and Jerry I know you're still here if you wanted  to come on and just thank the panelist that'd   be great I wanted to say personally wonderful  wonderful discussion um obviously the chat was  

55:18

blowing up through through out the whole entire  event we had so many questions and discussions   we want to continue this uh I did mention in the  chat that that we do have plans to continue this  

55:28

discussion with future events look out stay on  stay stay tuned for more details uh as a as a  

55:34

the the ERG the differing abilities ERG and the  AAPI said we want to come together and do it so  

55:40

Jerry did you want to say anything before Andre  has to jump off real quick I really just want   to thank all of you for this really engaging  discussion clearly our audience is interested  

55:50

and we are just beginning this discussion so look  forward to continuing other dialogue about this   topic in the in the near future so thank you  everybody thank you for joining our program  

55:59

and we have uh a Native American Heritage Month  starting on November 1 so keep an eye out for um  

56:07

our uh a broadcast announcing the theme for the  month and then the event for that month as well