WEBVTT

00:03.420 --> 00:05.531
OK , now we can have some fun . I was

00:05.531 --> 00:07.587
trying to be quiet cause I don't , I

00:07.587 --> 00:07.500
didn't know when they were gonna go

00:07.500 --> 00:10.220
live . Um , so this morning we are

00:10.220 --> 00:12.331
going to start off with another large

00:12.331 --> 00:15.140
group session known as a plenary . Um ,

00:15.380 --> 00:17.658
it's one of my new favorite words . Um ,

00:17.819 --> 00:20.219
and this person , you already know

00:20.219 --> 00:22.386
who's on the schedule , but I do wanna

00:22.386 --> 00:24.441
tell you a little story . As we were

00:24.441 --> 00:26.775
seeking folks to come and talk about AI ,

00:26.775 --> 00:29.950
um , I kept hearing this name and

00:31.590 --> 00:33.812
It excited me a little bit because this

00:33.812 --> 00:35.812
person is still active duty , which

00:35.812 --> 00:37.923
means we have some great innovators ,

00:37.923 --> 00:40.090
um , and , and big A I think are still

00:40.090 --> 00:42.312
on active duty . So , that's exciting .

00:42.312 --> 00:45.349
Um , his bio is in the folder , so I'm

00:45.349 --> 00:47.460
not gonna read all of that , but I do

00:47.460 --> 00:49.627
wanna , um , give a special welcome to

00:49.627 --> 00:51.905
today's very first guest , no pressure ,

00:51.905 --> 00:54.071
starting off in the morning , um , but

00:54.071 --> 00:55.793
Master Sergeant Ben Stratton .

01:00.860 --> 01:04.209
Oh you . No pressure . I know the

01:04.209 --> 01:06.265
nerves . I , this is why I have this

01:06.265 --> 01:09.379
like fidget here cause . Can't get over

01:09.379 --> 01:11.435
my nerves . So , uh , good morning ,

01:11.435 --> 01:13.601
everyone . I'm , is this working ? I'm

01:13.601 --> 01:15.712
just too tall , or you're too short .

01:18.000 --> 01:20.278
So , um , I'm Ma Sergeant Ben Stratton .

01:20.278 --> 01:22.278
I'm the emerging Technology Program

01:22.278 --> 01:24.111
Manager for the Air Force Public

01:24.111 --> 01:26.167
Affairs Agency at Randolph Air Force

01:26.167 --> 01:28.111
Base in San Antonio , Texas . Um ,

01:28.111 --> 01:30.519
essentially , uh , my job is , uh ,

01:30.669 --> 01:32.725
artificial intelligence , research ,

01:32.760 --> 01:34.927
development , testing , and evaluation

01:34.927 --> 01:36.982
for the public affairs enterprise in

01:36.982 --> 01:39.989
the Air Force . So , what we're trying

01:39.989 --> 01:42.470
to establish here is , how do we get

01:42.470 --> 01:45.669
time back , um , and spend that time

01:45.669 --> 01:48.150
back where trust is built . And we've

01:48.150 --> 01:50.372
learned a lot about that yesterday . It

01:50.372 --> 01:52.539
seemed like every briefer talked about

01:52.539 --> 01:55.459
how the human relations , those

01:55.459 --> 01:58.430
interactions are the most important

01:58.430 --> 02:00.790
part of how we move forward with AI .

02:02.040 --> 02:04.207
So that's how we ignite storytelling .

02:05.029 --> 02:07.750
It's deciding what those returned hours

02:07.750 --> 02:11.458
are for . AI will give us time

02:11.458 --> 02:14.330
back . Where the time goes is a choice .

02:15.039 --> 02:17.660
And it's ours to make , not the tools .

02:18.529 --> 02:21.779
The false one is more content means

02:21.779 --> 02:24.289
faster . We've been doing this thing

02:24.289 --> 02:26.345
for the last few decades where we're

02:26.345 --> 02:28.678
like , we can do so much more with less .

02:31.179 --> 02:33.235
But that , where did that leave us ?

02:34.279 --> 02:36.559
Training has been lost in the field .

02:37.350 --> 02:39.470
Media operations has been lost in the

02:39.470 --> 02:41.581
field , community engagement has been

02:41.581 --> 02:43.581
lost in the field . Mission partner

02:43.581 --> 02:47.199
support has been lost in the field . So ,

02:47.210 --> 02:49.669
how do we use this moment , our

02:50.240 --> 02:52.440
generation's industrial revolution or

02:52.449 --> 02:55.410
information revolution to give us that

02:55.410 --> 02:58.630
time back ? For training , for media ,

02:58.899 --> 03:01.232
for ComRL , for mission partner support .

03:03.960 --> 03:06.169
But before we get into that , this is

03:06.169 --> 03:09.210
who I am . Uh , you've got the bio

03:09.210 --> 03:11.869
apparently in the , in the folders . Um ,

03:12.169 --> 03:14.391
my first deployment was with the Navy .

03:14.391 --> 03:16.949
I am a shellback for Navy people here

03:16.949 --> 03:19.005
who know that . I see some Navy guys

03:19.005 --> 03:22.910
right there . Um , I've been

03:22.910 --> 03:24.854
deployed all over the place in all

03:24.854 --> 03:28.830
sorts of cocoms , um . And

03:28.830 --> 03:31.369
what I learned is that Those

03:31.369 --> 03:34.100
interactions that we make with our

03:34.100 --> 03:36.210
partners downrange , our partners in

03:36.210 --> 03:38.154
nations that we're there providing

03:38.154 --> 03:41.009
humanitarian support . That's how we

03:41.009 --> 03:43.250
get the stories . But how we , how do

03:43.250 --> 03:45.583
we tell the story faster is with AI now .

03:45.889 --> 03:48.000
This is like the coolest part that we

03:48.000 --> 03:51.309
have . I came back to

03:51.309 --> 03:53.389
Dnfos with a question I could not

03:53.389 --> 03:56.759
ignore . I graduated from Dinfos in

03:56.759 --> 04:00.559
February 2008 from BPASW as a writer .

04:02.160 --> 04:04.399
Our final project was laying out and

04:04.399 --> 04:07.080
designing a newspaper , newspapers .

04:10.360 --> 04:12.360
And then I wrote my first prompt in

04:12.360 --> 04:15.039
January 2025 , my first massive prompt

04:15.039 --> 04:18.079
using JSON , uh , at an Azure demo in

04:18.079 --> 04:20.246
Austin . Microsoft was putting on this

04:20.246 --> 04:22.468
big , uh , thing for the military , and

04:22.468 --> 04:24.579
we learned what we're moving toward .

04:25.609 --> 04:29.019
How we can build prompts that are

04:29.019 --> 04:32.510
smarter . When you go to the barbershop ,

04:33.010 --> 04:35.019
you don't tell somebody , I want a

04:35.019 --> 04:37.859
haircut . I think you're gonna get

04:37.859 --> 04:39.915
whatever that barberra wants to give

04:39.915 --> 04:42.220
you , right ? You tell the barber , for

04:42.220 --> 04:44.549
me , I want a medium high with the one

04:44.549 --> 04:46.827
on the side , a little trim on the top ,

04:46.827 --> 04:48.882
and get my eyebrows because they get

04:48.882 --> 04:52.489
super bushy . That's prompting . We

04:52.489 --> 04:55.690
prompt our own humans all the time . So

04:55.690 --> 04:58.130
do that with AI . Give AI the context

04:58.130 --> 05:00.470
it needs to understand .

05:01.679 --> 05:03.790
How to give you the output you want .

05:05.679 --> 05:07.846
And so we do that with AI research and

05:07.846 --> 05:10.012
development for the PA Enterprise . We

05:10.012 --> 05:12.739
can make AI useful without making trust

05:12.739 --> 05:14.910
negotiable by giving it the right

05:14.910 --> 05:16.743
context , by giving it the right

05:16.743 --> 05:19.470
prompts . So the pressure is real . The

05:19.470 --> 05:21.589
lesson is optional . This is our

05:21.589 --> 05:23.645
industrial revolution . I'm going to

05:23.645 --> 05:25.811
repeat myself a lot because apparently

05:25.811 --> 05:29.309
that's how humans learn . This is our

05:29.309 --> 05:31.309
information revolution in knowledge

05:31.309 --> 05:33.309
work . We can redistribute how that

05:33.309 --> 05:35.476
knowledge work is done , so we can put

05:35.476 --> 05:37.698
the humans out on the flight line where

05:37.698 --> 05:39.698
they belong , getting the content ,

05:39.698 --> 05:41.753
doing the historical documentation .

05:42.130 --> 05:44.352
The AI does not decide what the time is

05:44.352 --> 05:46.989
worth . The AI does not have a one

05:46.989 --> 05:50.510
white common access card . That's us

05:54.540 --> 05:57.600
Production became the shop . Every ,

05:58.140 --> 06:00.369
say , Air Force wing PA shop , the

06:00.369 --> 06:03.179
majority of our manning sits in command

06:03.179 --> 06:06.329
information . Say a shop has 70% of the

06:06.329 --> 06:08.496
Manning sitting in command information

06:08.496 --> 06:10.329
doing production , doing content

06:10.329 --> 06:12.980
generation . The 30% is then for media ,

06:13.209 --> 06:15.809
for CR , for plans and programs , for

06:15.809 --> 06:18.980
administravia work . But what if we

06:18.980 --> 06:21.899
could use AI to reshape our force , to

06:21.899 --> 06:24.119
reshape where we're putting the people

06:24.500 --> 06:27.140
do that human in the loop work that the

06:27.140 --> 06:28.196
AI can't replace .

06:32.720 --> 06:36.029
But trust cannot be automated . AI can

06:36.029 --> 06:38.230
generate , and it can , but it cannot

06:38.230 --> 06:40.119
replace the witness . You are the

06:40.119 --> 06:42.269
witness . You are the historical

06:42.269 --> 06:44.670
documenters . You are the ones writing

06:44.670 --> 06:47.369
history , not the AI . The AI is only

06:47.630 --> 06:51.369
predicting what you want to hear . It's

06:51.369 --> 06:53.369
generating content from what you've

06:53.369 --> 06:55.036
created , because you are the

06:55.036 --> 06:56.425
historical documenters .

06:58.899 --> 07:00.899
So generate is not capture , and we

07:00.899 --> 07:03.010
have to understand how important this

07:03.010 --> 07:04.955
is , especially when we're talking

07:04.955 --> 07:08.790
about . Forensic photography .

07:09.109 --> 07:11.429
How do we draw the line between when

07:11.429 --> 07:12.540
we're talking about

07:15.549 --> 07:17.820
Creating content for social media or

07:17.820 --> 07:19.559
creating content for a safety

07:19.559 --> 07:21.615
investigation board . How do we draw

07:21.615 --> 07:23.615
the line so people understand those

07:23.615 --> 07:26.799
lawyers , those investigators can trust

07:27.459 --> 07:29.681
that the photographs they're looking at

07:29.681 --> 07:31.859
is real , it's from us , so the public

07:31.859 --> 07:33.915
can trust what they're looking at is

07:33.915 --> 07:36.750
from us . That's content provenance

07:36.750 --> 07:38.806
that uh Ken talked about yesterday .

07:40.359 --> 07:42.399
That's making sure that humans are

07:42.399 --> 07:44.510
doing what the humans are really good

07:44.510 --> 07:46.788
at . The Humans are on the flight line .

07:46.788 --> 07:48.677
The humans are next to the tank .

07:51.109 --> 07:53.829
Be present , earn the source , become

07:53.829 --> 07:56.107
the historical record of what occurred .

07:56.230 --> 07:58.429
That is our job , that is your job .

08:00.200 --> 08:01.922
The meme can , the machine can

08:01.922 --> 08:04.089
synthesize an image , it can assist in

08:04.089 --> 08:06.089
editing , captioning , sequencing ,

08:06.089 --> 08:07.978
export , distribution , all those

08:07.978 --> 08:10.359
wonderful fun things that take so much

08:10.359 --> 08:14.350
time . But

08:14.350 --> 08:16.350
we have to keep people at the gates

08:16.350 --> 08:18.559
that carry consequence . AI doesn't

08:18.559 --> 08:21.410
have the release authority . We have

08:21.410 --> 08:23.589
the release authority . So the

08:23.589 --> 08:25.645
correction preserves what happened .

08:26.850 --> 08:29.089
Provenance is the receipt , how we

08:29.089 --> 08:31.311
track it , whether it's the information

08:31.769 --> 08:34.039
or the imagery . Release authority

08:34.039 --> 08:36.039
never moves , it remains with you .

08:38.409 --> 08:40.465
Now map the work before you move the

08:40.465 --> 08:43.609
people . Capacity is a design decision .

08:43.690 --> 08:47.099
It is not something AI does . To you

08:48.489 --> 08:50.750
It's not an infliction on your work .

08:54.580 --> 08:57.039
So In order to understand what the

08:57.039 --> 08:59.960
field needs , AFPA sent out 3

08:59.960 --> 09:02.369
questionnaires to the field . We did a ,

09:02.559 --> 09:04.520
uh , needs assessment , a best

09:04.520 --> 09:07.729
practices questionnaire , and a

09:07.729 --> 09:09.896
contracting questionnaire . We want to

09:09.896 --> 09:11.785
understand what the field needs ,

09:11.785 --> 09:14.007
because we can't build from the top and

09:14.007 --> 09:16.007
expect everyone in the field to get

09:16.007 --> 09:18.062
what they want . We have to get from

09:18.062 --> 09:20.285
the bottom what they actually need , so

09:20.285 --> 09:22.340
that we can take care of them . So ,

09:22.340 --> 09:24.929
what the field needs , Manning . Not

09:24.929 --> 09:28.500
more AI . But they want AI

09:28.500 --> 09:31.289
training . They want policy , and they

09:31.289 --> 09:33.400
want DINOs curriculum that gives them

09:33.400 --> 09:35.400
pipeline airmen who understand AI ,

09:35.450 --> 09:37.394
that gives them intermediate level

09:37.394 --> 09:39.617
training that understands AI and how we

09:39.617 --> 09:41.783
can integrate that everywhere we do at

09:41.783 --> 09:44.840
PA . They asked for this room .

09:46.330 --> 09:48.552
That is why I'm standing in it . That's

09:48.552 --> 09:50.669
why Chief Scaff invited me here from

09:50.770 --> 09:53.960
war media activity . We are here to

09:53.960 --> 09:57.080
talk about this . I can't thank Chief

09:57.080 --> 09:59.479
Staff and the WMA enough for inviting

09:59.479 --> 10:01.840
me here to be a part of this . It's

10:01.840 --> 10:05.260
really , this is why I'm so shaky . To

10:05.260 --> 10:07.482
be here in the building that started my

10:07.482 --> 10:11.210
career is amazing . And then we

10:11.210 --> 10:14.770
mapped all 198 enlisted public affairs

10:14.770 --> 10:18.049
tasks from the Air Force CFETPs against .

10:18.969 --> 10:21.489
AI augmentable , AI replaceable , and

10:21.489 --> 10:23.600
human only . We additionally did that

10:23.600 --> 10:26.010
with all of the officer tasks . So a

10:26.010 --> 10:29.250
total of 428 uh tasks from our

10:29.250 --> 10:32.659
CFUTPs , we assessed . What's AI

10:32.659 --> 10:36.619
touchable ? We found 69% is

10:36.619 --> 10:40.609
AI touchable . It's

10:40.609 --> 10:44.520
a big number . And

10:44.520 --> 10:48.119
then the CSAF Pathfinder program , they

10:48.119 --> 10:51.380
chose the 3NO's , that's the Air Force ,

10:51.570 --> 10:55.179
uh , crew field for enlisted to be

10:55.440 --> 10:58.460
a case study on how we can use AI .

10:59.559 --> 11:02.630
To take care of things that the human

11:02.630 --> 11:05.010
can't do . So we're part of that

11:05.010 --> 11:07.630
Pathfinder program from CISAF . For

11:07.630 --> 11:09.963
those of you who aren't in the military ,

11:09.963 --> 11:12.019
that's the Chief of Staff of the Air

11:12.019 --> 11:14.770
Force . That's our 4 star . And we're

11:14.770 --> 11:17.179
not looking for a people cut target .

11:17.890 --> 11:20.830
We want to reshape how we do business .

11:21.099 --> 11:23.321
We want to reshape where the people are

11:23.321 --> 11:25.488
doing the work . We want to put people

11:25.488 --> 11:27.488
where only people can do the work .

11:30.359 --> 11:32.359
And we know PA professionals are

11:32.359 --> 11:36.090
already using AI . The problem isn't

11:36.090 --> 11:38.859
adoption . It's fragmentation .

11:39.159 --> 11:41.440
Everyone is doing their own thing . And

11:41.440 --> 11:44.539
that's the problem . We have to un-silo .

11:44.919 --> 11:48.359
We have to become one military force ,

11:48.530 --> 11:51.099
one federal government , one .

11:54.260 --> 11:56.299
sorry , words , um , collaborating

11:56.299 --> 11:58.979
force where we work together to

11:58.979 --> 12:02.250
understand how to use AI . The siloing

12:02.250 --> 12:04.472
thing we've been doing for the last few

12:04.472 --> 12:06.840
decades is not working . This is how

12:06.840 --> 12:10.400
the enemy owns us . We have to work

12:10.400 --> 12:14.270
together . So , units are building

12:14.270 --> 12:17.510
similar tools independently . There's

12:17.510 --> 12:19.843
no inheritance , no doctrinal alignment ,

12:19.843 --> 12:22.150
no autonomy path , no memory .

12:23.900 --> 12:26.122
In the architecture that we've built at

12:26.122 --> 12:29.140
AFP is called Quill . Quill solves that

12:29.140 --> 12:31.539
fragmentation by providing one

12:31.539 --> 12:34.219
structural model any PA organization

12:34.219 --> 12:35.219
can adopt .

12:38.450 --> 12:40.561
And then we have to understand how we

12:40.561 --> 12:42.394
work this through the research ,

12:42.394 --> 12:44.672
development , testing , and evaluation ,

12:44.672 --> 12:46.783
you know , and , uh , DNOs , we can't

12:46.783 --> 12:48.894
use acronyms , so I keep spelling out

12:48.894 --> 12:50.894
what RDTNE is . RDTNE pipeline , we

12:50.894 --> 12:54.130
assess it through these steps . If

12:54.130 --> 12:55.979
somebody wants a brand fancy new

12:55.979 --> 12:59.250
contract , why ? If somebody wants a

12:59.250 --> 13:02.200
SAS , a software as a service , why ?

13:02.619 --> 13:05.020
Can we already do it in Office 365 ?

13:06.039 --> 13:08.095
Which is already paid for on Ajela .

13:08.095 --> 13:10.317
Can we already do it with Adobe , which

13:10.317 --> 13:12.372
is already paid for on Jela ? Can we

13:12.372 --> 13:14.317
already do it with P3 ? And if you

13:14.317 --> 13:16.539
don't know what P3 is , this is the Air

13:16.539 --> 13:18.428
Force Public Affairs Productivity

13:18.428 --> 13:20.428
portal that we use to , for project

13:20.428 --> 13:22.539
management and equipment management .

13:22.539 --> 13:25.840
Every PA in the Air Force . Uses this

13:25.840 --> 13:29.270
P3 to manage our projects , to manage

13:29.270 --> 13:32.390
our hours , so we understand where

13:32.390 --> 13:34.359
we're putting people , where we're

13:34.359 --> 13:37.030
spending time , where we have equipment .

13:38.210 --> 13:40.890
And then Quill . Qui is the no-cost ,

13:40.960 --> 13:43.071
low-cost solution . We can build this

13:43.071 --> 13:45.729
in-house . We don't need another

13:45.729 --> 13:47.599
contract . You want the public's

13:47.599 --> 13:49.655
trusted ? Stop wasting their money .

13:54.210 --> 13:56.309
So Quill gives every prompt the same

13:56.309 --> 13:59.320
starting point . A blank chat starts

13:59.320 --> 14:02.989
from 0 . There's no unit context .

14:03.510 --> 14:05.288
There's no review standard , no

14:05.288 --> 14:08.770
continuity . So we have

14:08.770 --> 14:12.150
65+ prompts that we've created

14:12.690 --> 14:15.710
for public affairs . We have a suite of

14:15.710 --> 14:17.988
story coach prompts from print , photo ,

14:18.109 --> 14:20.679
video , and a caption writer . These

14:20.679 --> 14:23.229
help you cover the event , or the story ,

14:23.400 --> 14:25.511
the feature , the information story .

14:25.919 --> 14:27.975
We have prompts that help you handle

14:27.975 --> 14:30.086
the media . We have a media summary ,

14:30.086 --> 14:32.308
tone analyst , press engagement coach .

14:32.820 --> 14:34.931
We have prompts to help you clear the

14:34.931 --> 14:36.876
release , a release of information

14:36.876 --> 14:38.653
advisor , a security and policy

14:38.653 --> 14:40.709
reviewer . We have prompts that help

14:40.709 --> 14:42.376
you brief the boss , with the

14:42.376 --> 14:44.487
commander's communication builder , a

14:44.487 --> 14:46.559
war writer . How many people in this

14:46.559 --> 14:49.109
room have to tell the boss what you did

14:49.109 --> 14:52.979
every week ? Yeah , well , we have a

14:52.979 --> 14:55.219
prompt for that . You can start a chat

14:55.219 --> 14:58.489
with this war writer prompt . At the

14:58.489 --> 15:00.656
beginning of your week , and then tell

15:00.656 --> 15:02.767
it every day what you did , and it'll

15:02.767 --> 15:04.711
give you a repeatable templated uh

15:04.711 --> 15:06.822
output that your boss wants to see at

15:06.822 --> 15:09.080
the end of the week . You don't have to

15:09.080 --> 15:11.247
rethink about all the things you did .

15:11.247 --> 15:13.302
Just have a conversation with the AI

15:13.302 --> 15:13.200
and they'll give you the output at the

15:13.200 --> 15:16.979
end of the week . Oh yeah . If you have

15:16.979 --> 15:19.146
a question , uh , I think the sergeant

15:19.146 --> 15:20.659
Major said , use the mic .

15:28.070 --> 15:30.229
Um , I , I heard what you just said .

15:30.590 --> 15:33.109
Can that be stored and be used at the

15:33.109 --> 15:35.109
end of , be used for our end of the

15:35.109 --> 15:38.390
year evaluation ? So , I , I , I can't

15:38.390 --> 15:40.612
speak for anyone else , but on my job ,

15:40.710 --> 15:43.010
we do a end of the year evaluation

15:43.010 --> 15:45.121
where we have to basically track what

15:45.121 --> 15:47.232
we've done for the year and something

15:47.232 --> 15:49.454
that could , uh , track us for the week

15:49.454 --> 15:51.677
if I could say I did this doing these .

15:51.677 --> 15:53.843
Months in these uh seasons of the year

15:53.843 --> 15:56.010
that would drastically help me . Yes ,

15:56.039 --> 15:58.280
and I have a performance writer for

15:58.280 --> 16:00.739
military and civilian that understands

16:00.739 --> 16:02.469
DP map , that understands your

16:02.469 --> 16:04.760
evaluation process , that understands

16:04.760 --> 16:06.927
how to write for civilian appraisals .

16:07.289 --> 16:09.456
And the performance writer understands

16:09.456 --> 16:11.678
how to write for military evaluations .

16:11.690 --> 16:13.690
For on the Air Force side , we call

16:13.690 --> 16:16.489
that the enlisted , uh , for the EPB or

16:16.489 --> 16:19.809
the OPB for officers . There's a

16:19.809 --> 16:21.809
supervisor . Sorry , go ahead . I'm

16:21.809 --> 16:23.642
sorry . I'm Dave , um , from the

16:23.642 --> 16:25.698
intelligence community . Yeah , uh ,

16:25.698 --> 16:27.642
great programs . Love it . Is this

16:27.642 --> 16:29.476
upscale to the , uh , classified

16:29.476 --> 16:31.365
network , IEJWC ? So , uh , these

16:31.365 --> 16:33.531
prompts are all standalone . I'll have

16:33.531 --> 16:35.698
a QR code eventually in this slideshow

16:35.729 --> 16:37.618
where you can scan it and you can

16:37.618 --> 16:39.729
download . I created a 3 , it's a 300

16:39.729 --> 16:42.270
page uh PDF portfolio with interlinking

16:42.270 --> 16:44.950
links that create this , it's like a

16:44.950 --> 16:47.172
website basically , but in a PDF it's 5

16:47.172 --> 16:49.750
megabytes . Um , it has all of the

16:49.750 --> 16:51.917
prompts . You can see the markdown and

16:51.917 --> 16:54.139
how all of the prompts are written , so

16:54.139 --> 16:56.250
you can customize them and use them .

16:56.250 --> 16:58.250
At your heart's will . This is open

16:58.250 --> 17:00.417
source . This is how we un silo . This

17:00.417 --> 17:02.417
is how we work together . There's a

17:02.417 --> 17:04.583
supervisor coach so you can understand

17:04.583 --> 17:04.359
how if you're a civilian , how do I

17:04.359 --> 17:06.640
supervise the military ? There's the

17:06.640 --> 17:08.696
supervisor coaches who , if you're a

17:08.696 --> 17:10.584
military , supervising a military

17:10.584 --> 17:12.751
person , you're brand new and you need

17:12.751 --> 17:16.530
help , we can use AI . And then we

17:16.530 --> 17:18.752
talked about media operations or crisis

17:18.752 --> 17:20.641
communication and how that's been

17:20.641 --> 17:22.918
neglected . Here's the thing . Give the

17:22.918 --> 17:25.085
junior communicator enough repetitions

17:25.085 --> 17:26.307
to sit at the table .

17:29.030 --> 17:32.410
Thanks to Dinfos and BSW . When I

17:32.410 --> 17:35.079
showed up at Minot Air Force Base . In

17:35.079 --> 17:38.020
February 2008 , 6 months after the

17:38.020 --> 17:40.540
August '07 nuclear incident , if any of

17:40.540 --> 17:43.180
you remember the B-52s that decided to

17:43.180 --> 17:45.180
fly down with nukes on board to the

17:45.180 --> 17:47.069
Barksdale , I showed up there six

17:47.069 --> 17:50.849
months later . And then . 2

17:50.849 --> 17:54.050
months after I arrived , 3-level A1C

17:54.050 --> 17:56.510
Stratton is on PA on call .

17:58.250 --> 18:00.306
And I have to respond to the missile

18:00.306 --> 18:02.361
response cell , which is the missile

18:02.361 --> 18:04.139
wing's version of the emergency

18:04.139 --> 18:06.250
operations center . And I have to sit

18:06.250 --> 18:09.380
at the table as A1C Stratton . But I

18:09.380 --> 18:11.602
knew what to do . I knew how to write a

18:11.602 --> 18:13.658
media advisory . I knew how to write

18:13.658 --> 18:15.824
the press release . I knew how to work

18:15.824 --> 18:17.602
with the EOC director . Because

18:17.602 --> 18:19.824
Dinfel's prepped me for that . I called

18:19.824 --> 18:21.936
my tech sergeant . She said , you got

18:21.936 --> 18:24.102
this . I trust you . I was like , no ,

18:24.102 --> 18:26.780
I don't But I did , because didn'tfos

18:26.780 --> 18:28.947
prepped me for it . Are we still doing

18:28.947 --> 18:31.900
that today ? Could we send an E3 to the

18:31.900 --> 18:34.119
EOC and trust them to do that job ?

18:36.119 --> 18:37.119
I don't know

18:40.160 --> 18:42.680
Senior NCOs , that's us . We should be

18:42.680 --> 18:45.013
orchestrating and developing , training .

18:46.189 --> 18:48.920
And AI will give us that time back . We

18:48.920 --> 18:51.400
can use AI to train and every quill

18:51.400 --> 18:54.489
prompt . is built on a three-part

18:54.489 --> 18:56.520
structure . There's a quill block ,

18:56.770 --> 18:58.992
there's an activity block , and there's

18:58.992 --> 19:01.159
a prompt block . And the quill block ,

19:01.159 --> 19:03.214
every single prompt has this , which

19:03.214 --> 19:05.530
creates this continuity every time you

19:05.530 --> 19:07.880
use a quill prompt . You know what

19:07.880 --> 19:10.102
you're getting , you know it's going to

19:10.102 --> 19:12.269
coach you on that subject matter . You

19:12.269 --> 19:14.491
know it's gonna coach you on how to use

19:14.491 --> 19:17.189
AI better . It also has a confidence

19:17.189 --> 19:20.829
level . So every quill prompt will ask

19:20.829 --> 19:23.030
you follow-up questions until it feels

19:23.030 --> 19:25.910
like it's 95% confident in its output

19:25.910 --> 19:28.243
before it actually gives you the output .

19:28.520 --> 19:30.464
Because that's how we get better .

19:30.464 --> 19:32.687
That's how we make sure that AI doesn't

19:32.687 --> 19:34.853
hallucinate . AI will confidently tell

19:34.853 --> 19:36.742
you that your F-16 crashed . It's

19:36.742 --> 19:39.020
sitting on the flight line , it's fine .

19:39.890 --> 19:41.723
If it doesn't have the context .

19:45.650 --> 19:47.920
So you got it from a human , but it

19:47.920 --> 19:51.380
writes the story . We are embracing AI

19:52.060 --> 19:54.171
to write our stories for us , because

19:54.171 --> 19:56.349
it takes a lot of time . We're

19:56.349 --> 19:58.979
embracing AI to write our scripts for

19:58.979 --> 20:01.201
us , because that takes a lot of time .

20:01.589 --> 20:05.540
An AI can do it in 5 minutes . You use

20:05.540 --> 20:07.651
the story coach print feature story .

20:09.250 --> 20:11.417
Use it to write the feature . You give

20:11.417 --> 20:13.639
it the raw interview because you're the

20:13.639 --> 20:15.639
human who got the interview because

20:15.639 --> 20:17.972
you're really good at talking to people ,

20:17.972 --> 20:20.194
or you're an introvert like me and this

20:20.194 --> 20:22.306
is like nerve-wracking . Um . But one

20:22.306 --> 20:25.089
on one , those interviews , solid .

20:28.420 --> 20:30.469
We can use AI to free us up , to do

20:30.469 --> 20:32.636
more of the human things that AI can't

20:32.636 --> 20:34.858
do . The Story coach prompts were built

20:34.858 --> 20:37.319
on the Morell Media Awards , SOP .

20:38.189 --> 20:40.300
These prompts understand what makes a

20:40.300 --> 20:42.479
really good . Article . These prompts

20:42.479 --> 20:44.535
were built on Pavilion . Thank you ,

20:44.535 --> 20:46.479
Pavilion for providing all of this

20:46.479 --> 20:48.535
amazing data that we can use to make

20:48.535 --> 20:51.000
our jobs easier . Thank you for all of

20:51.000 --> 20:53.239
the humans . I've been a Sme for 5

20:53.239 --> 20:56.160
years for Pavilion . Thank you to all

20:56.160 --> 20:58.327
the humans who have dedicated all this

20:58.327 --> 21:00.719
time and effort to creating a way to

21:00.719 --> 21:03.859
save us time . These prompts understand

21:03.859 --> 21:06.329
AP style because every model

21:06.329 --> 21:09.900
understands AP if you tell it . They

21:09.900 --> 21:12.229
understand tongue and quill . If you

21:12.229 --> 21:14.619
tell it , you want tongue quill . It's

21:14.619 --> 21:16.841
not guessing at a story . It is writing

21:16.841 --> 21:19.063
to the standard your supervisor already

21:19.063 --> 21:22.810
grades against . So

21:22.810 --> 21:25.032
feature story from a flightline visit .

21:25.032 --> 21:27.143
My interview notes and shot list from

21:27.143 --> 21:29.143
this morning's flightline visit are

21:29.143 --> 21:31.310
below . Write the feature story . This

21:31.310 --> 21:33.532
you give it after you give it the , the

21:33.532 --> 21:36.329
full prompt . Hold to Daffy 35101 ,

21:36.410 --> 21:38.630
which is the Air Force uh instruction

21:38.770 --> 21:41.760
for public affairs . In the morale

21:41.760 --> 21:44.219
category fit . AP style tongue quill ,

21:45.119 --> 21:47.286
flag every assumption and every fact ,

21:47.479 --> 21:49.760
I still have to confirm because we

21:49.760 --> 21:51.704
don't want it to give us an output

21:51.704 --> 21:53.871
that's terrible . We want it to always

21:53.871 --> 21:56.093
ask us more questions so it understands

21:56.093 --> 21:59.380
the context . And if you don't .

22:00.400 --> 22:03.010
Understand AI you can ask AI to help

22:03.010 --> 22:05.177
you prompt better . Ask it to give you

22:05.177 --> 22:07.288
feedback on how your prompting was in

22:07.288 --> 22:09.177
that session , and it'll give you

22:09.177 --> 22:11.288
feedback . It'll help you get smarter

22:11.288 --> 22:13.621
on how to use it . And that's the point .

22:14.869 --> 22:17.036
Every quill workflow ends with a human

22:17.036 --> 22:20.020
decision . Input , mission and facts

22:20.020 --> 22:22.131
you gathered because you're the human

22:22.131 --> 22:24.739
in the loop . Quill understands the

22:24.739 --> 22:26.699
context , the coaching , review

22:26.699 --> 22:29.310
behavior that you expect . The draft .

22:29.390 --> 22:31.557
This is the output to inspect , not to

22:31.557 --> 22:33.668
trust , not to just blindly release ,

22:33.829 --> 22:35.885
because again , AI does not have the

22:35.885 --> 22:38.051
white common access card . AI does not

22:38.051 --> 22:40.162
have the release authority , you do .

22:41.300 --> 22:43.467
You take the draft , you edit it , and

22:43.467 --> 22:46.099
you release it . Because you're the

22:46.099 --> 22:48.900
human , you verify and decide , output

22:48.900 --> 22:51.520
does not equal authority . The tool can

22:51.520 --> 22:54.680
create the next move . But you own

22:54.680 --> 22:55.736
whether it moves .

22:58.689 --> 23:00.911
But I don't have time to learn AI . But

23:00.911 --> 23:03.078
that's the thing , the more you use AI

23:03.078 --> 23:07.060
to get the job done . The

23:07.060 --> 23:09.400
more you learn AI . So if you're using

23:09.400 --> 23:12.260
the thing that's helping you save time .

23:13.750 --> 23:15.750
You're going to learn how to use it

23:15.750 --> 23:17.583
while you're using it to get the

23:17.583 --> 23:19.694
mission done . You already have to do

23:19.694 --> 23:21.861
this work . Why not use it to save you

23:21.861 --> 23:24.083
time to get back into training , to get

23:24.083 --> 23:26.083
back into comR , to get back to the

23:26.083 --> 23:28.250
media , to get back into crisis coms .

23:30.239 --> 23:32.160
So these are the 4 programs we're

23:32.160 --> 23:34.800
building on how we can address AI .

23:35.349 --> 23:37.293
Quill prompts , we've talked a lot

23:37.293 --> 23:39.310
about that . Quill apps , we're

23:39.310 --> 23:43.229
building 3 apps . One Quil OS is

23:43.229 --> 23:46.979
a Mac native Swift app . That will that

23:46.979 --> 23:48.812
intuitively understands based on

23:48.812 --> 23:51.709
context . What you need . It'll

23:51.709 --> 23:54.540
automatically pull the prompt that most

23:55.160 --> 23:57.216
works for the context of what you're

23:57.216 --> 23:59.438
working at . Hey , I gotta go to Dinfos

23:59.438 --> 24:01.493
and brief a bunch of people about AI

24:01.550 --> 24:04.160
and I need to tell my boss how much

24:04.160 --> 24:06.160
it's gonna cost . Well , there is a

24:06.160 --> 24:08.359
travel estimator prompt that pulls in

24:08.359 --> 24:10.415
the per diem for the local area that

24:10.415 --> 24:12.959
pulls in , uh , flight stats and data

24:12.959 --> 24:15.126
from the website , from the internet ,

24:15.126 --> 24:17.292
so you can get the drafts you and it ,

24:17.292 --> 24:19.699
sorry . Draft , I'm talking more with

24:19.699 --> 24:22.349
my hands now . So it drafts you that

24:22.349 --> 24:24.293
travel estimate , so your boss can

24:24.293 --> 24:26.516
approve it and then you can go into DTS

24:26.516 --> 24:28.571
and , and , uh , submit it , right ?

24:29.760 --> 24:31.482
But it's gonna intuitively and

24:31.482 --> 24:33.859
dynamically understand . What you need .

24:34.199 --> 24:36.366
There's this field kit . The field kit

24:36.366 --> 24:38.439
is the step before this quill app

24:38.439 --> 24:40.719
becomes available . Both are in alpha .

24:41.540 --> 24:43.762
The field kit , it will be a zip file .

24:43.900 --> 24:45.844
It'll include all of the prompts .

24:45.844 --> 24:47.956
It'll include all of the awareness of

24:47.956 --> 24:50.067
what these programs are and how we're

24:50.067 --> 24:53.060
addressing AI . And you give it Chat

24:53.060 --> 24:55.750
GPT codex . You give it Olama .

24:56.839 --> 24:59.959
Olama is a service that allows you to

24:59.959 --> 25:03.719
use on-device models like GEMA 412B .

25:04.069 --> 25:07.189
GEMA 412B is a very smart model that

25:07.189 --> 25:09.739
uses very little hardware requirements .

25:10.719 --> 25:14.180
So you can keep PA data on a PA device

25:14.180 --> 25:16.291
and never touch the cloud to get most

25:16.291 --> 25:18.513
of your information work done . You can

25:18.513 --> 25:20.736
be deployed downrange , not have access

25:20.736 --> 25:22.791
to the internet . And still get work

25:22.791 --> 25:24.902
done . GAI.mil is wonderful , but you

25:24.902 --> 25:26.709
have to have a Nipper computer

25:26.709 --> 25:28.900
connected to NipperNet to use it . But

25:28.900 --> 25:31.109
what if we could do it on our Macs and

25:31.109 --> 25:33.276
don't need the internet to get our job

25:33.276 --> 25:36.410
done ? What if you could do it on a

25:36.410 --> 25:38.729
Windows stand-alone computer ? With the

25:38.729 --> 25:41.939
chat GPT codex converted for the local

25:41.939 --> 25:45.739
models . Also building a war gaming

25:45.739 --> 25:48.829
app . This is a web app that's built on ,

25:49.260 --> 25:51.093
anyone heard of Odin Dateworld ?

25:52.550 --> 25:56.280
01 hand , 2 hands . Odin Date World is

25:56.280 --> 25:59.670
built by the US Army . Wonderful in uh

25:59.670 --> 26:02.469
platform that has all sorts of fake

26:02.469 --> 26:04.691
countries with fake country leaders and

26:04.691 --> 26:07.030
fake culture and language and all of

26:07.030 --> 26:09.989
this amazing stuff . So the war gaming

26:09.989 --> 26:12.100
app will pull all is pulling all that

26:12.100 --> 26:14.699
data in and helping us run white cell

26:14.750 --> 26:16.800
and blue cell and red cell in an

26:16.800 --> 26:18.900
interactive information environment

26:19.239 --> 26:21.295
that helps us get back to training ,

26:21.295 --> 26:23.680
that helps us prepare for the enemy

26:23.680 --> 26:26.949
that's already ahead of us . We can

26:26.949 --> 26:29.116
play like the enemy , because the only

26:29.116 --> 26:31.229
way for us to get ahead of them is to

26:31.229 --> 26:33.589
get in their brains , to understand who

26:33.589 --> 26:36.349
they are as the human in the loop on

26:36.349 --> 26:40.199
their side . And then cool Alphas

26:42.079 --> 26:45.239
This is a network of really smart PAs .

26:46.359 --> 26:48.303
Who may not be smart at AI but are

26:48.303 --> 26:50.581
really smart at the job and understand ,

26:50.581 --> 26:53.150
hey , that prompt actually sucks , and

26:53.160 --> 26:56.410
we need to make it better . So , We can

26:56.410 --> 26:58.632
send it to the Alpha Network , and they

26:58.632 --> 27:00.577
can help us get better . These are

27:00.577 --> 27:02.688
practitioners who test it against the

27:02.688 --> 27:03.688
real work .

27:06.609 --> 27:08.776
So 3 rules , they hold no matter which

27:08.776 --> 27:11.050
rung you pick . Show the sources ,

27:11.319 --> 27:13.263
every claim traceable to something

27:13.263 --> 27:15.239
dated and real . Keep the human at

27:15.239 --> 27:17.461
every gate that carries consequence and

27:17.461 --> 27:20.369
reinvest the time where humans , where

27:20.369 --> 27:22.036
their presence builds trust .

27:26.150 --> 27:28.372
So we started this presentation talking

27:28.372 --> 27:30.483
about how we're gonna get time back ,

27:30.483 --> 27:32.483
how we're going to spend it , where

27:32.483 --> 27:34.829
trust is built . So pick the wrong one

27:34.829 --> 27:37.630
rung , one rung this week . Should have

27:37.630 --> 27:39.880
chose better words there . Bring one

27:39.880 --> 27:42.119
workflow , keep the human at the gate .

27:43.420 --> 27:45.476
So this is the QR code . If you want

27:45.476 --> 27:48.209
access to all the prompts and uh

27:48.209 --> 27:50.265
explainer uh for all of the programs

27:50.265 --> 27:52.431
we're building , scan this . This will

27:52.431 --> 27:54.431
take you to a Google Drive that has

27:54.431 --> 27:57.640
these slides and the PDF portfolio .

27:57.900 --> 28:00.560
The point is that AI isn't replacing PA ,

28:00.939 --> 28:03.569
it's refocusing PA , and I don't know ,

28:03.729 --> 28:05.785
this is , oh , the mouse does work .

28:06.270 --> 28:09.859
This Is the quill mark .

28:10.160 --> 28:14.010
It's a reinvention . Of our badge

28:15.829 --> 28:17.949
Maximum disclosure , minimum delay .

28:18.900 --> 28:22.030
The quill and the lightning bolt . This

28:22.310 --> 28:24.366
is how we move forward , because you

28:24.366 --> 28:26.366
are still the storyteller , but the

28:26.366 --> 28:28.477
tools , they just got better . And if

28:28.477 --> 28:30.910
you want to join us tomorrow , War

28:30.910 --> 28:34.229
Media Activity has invited anyone who

28:34.229 --> 28:36.285
in this room who wants to join WMA .

28:37.069 --> 28:40.660
I'm hosting an AI Quill collab . And

28:40.660 --> 28:42.716
this is an overview of what we'll be

28:42.716 --> 28:44.882
talking about . We'll be diving deeper

28:44.882 --> 28:46.882
into , uh , Quill , we'll be diving

28:46.882 --> 28:49.104
deeper into the AI and its background .

28:49.104 --> 28:51.438
We'll be diving deeper into the prompts ,

28:51.569 --> 28:53.736
how to use them , and then we're going

28:53.736 --> 28:55.680
to do a stress test and exercise ,

28:55.680 --> 28:57.791
where you'll have 5 deliverables that

28:57.791 --> 28:59.847
usually take hours to get done , and

28:59.847 --> 29:02.013
you have 45 minutes to get them done .

29:05.599 --> 29:07.766
So that is my briefing . Do I have any

29:07.766 --> 29:10.189
questions ? Sure .

29:13.890 --> 29:16.168
Hello , um , I've got a quick question .

29:16.168 --> 29:18.410
So I'm finding that our models that

29:18.410 --> 29:20.077
we're building have a general

29:20.077 --> 29:22.132
understanding of the AP style book ,

29:22.132 --> 29:24.132
but not a specific one because it's

29:24.132 --> 29:26.243
copyrighted information . How are you

29:26.243 --> 29:25.689
guys navigating that ? Do you have a

29:25.689 --> 29:28.229
license agreement ? Is this more like

29:28.449 --> 29:30.849
the first run at AP and then it needs

29:30.849 --> 29:33.016
to be followed up by a human who has a

29:33.016 --> 29:34.960
better understanding in a physical

29:34.960 --> 29:37.530
style book ? That is the right answer .

29:37.930 --> 29:39.986
That's the answer that we've been to

29:39.986 --> 29:41.874
hear it . Sorry , I know , it's a

29:41.874 --> 29:43.986
bridge . We're PAs . We're trained on

29:43.986 --> 29:47.369
bridging , aren't we ? Um , You always

29:47.369 --> 29:49.313
have to be the last reviewer . You

29:49.313 --> 29:51.258
always , you still have to know AP

29:51.258 --> 29:54.390
cause you still , just like . Treat

29:54.390 --> 29:57.630
AI's drafts as your three-level ,

29:58.010 --> 30:00.709
um , as your apprentice's draft that

30:00.709 --> 30:02.820
you have to edit as the journeyman or

30:02.820 --> 30:06.589
craftsman in the room . That still

30:06.589 --> 30:08.811
saves you time cause you didn't have to

30:08.811 --> 30:11.319
do the drafting . Don't rely on the AI

30:12.030 --> 30:14.469
that you're still a human in the room .

30:16.119 --> 30:19.989
The An important part two is that ,

30:20.390 --> 30:22.723
and I was talking about this with uh Mr .

30:22.723 --> 30:24.612
Coslett last night , is that it's

30:24.612 --> 30:26.670
understanding that every model has a

30:26.670 --> 30:29.819
training date cut off . I don't know if

30:29.819 --> 30:31.819
you understand that , but if you're

30:31.819 --> 30:33.986
talking with , say you're in Gea.mil ,

30:33.986 --> 30:36.097
and you're having a conversation with

30:36.097 --> 30:38.380
the latest Gemma 3.5 model . Its

30:38.380 --> 30:40.324
training date cutoff was roughly 6

30:40.324 --> 30:43.979
months ago . From when it released .

30:44.430 --> 30:47.550
So a 3.5 released two months ago , then

30:47.550 --> 30:49.717
six months prior to that , at least at

30:49.717 --> 30:52.050
minimum , was its training date cut off .

30:52.050 --> 30:53.717
So if you're trying to have a

30:53.717 --> 30:55.883
conversation about the Iran conflict ,

30:55.883 --> 30:58.750
has no idea anything's wrong in Iran

30:58.750 --> 31:01.500
right now . Unless you enable web

31:01.500 --> 31:03.333
search , you have to enable that

31:03.333 --> 31:05.333
plug-in for it to have awareness of

31:05.333 --> 31:07.556
current events , current data , current

31:07.556 --> 31:08.239
AP style .

31:11.349 --> 31:12.905
Yeah . First of all , great

31:12.905 --> 31:15.071
presentation . I'm excited to read the

31:15.071 --> 31:17.238
prompts , but to poke the bear because

31:17.238 --> 31:16.619
that's what I like to do with some of

31:16.619 --> 31:18.563
these things . So you talked about

31:18.563 --> 31:20.619
using it to write articles . At what

31:20.619 --> 31:22.563
point then are you not writing the

31:22.563 --> 31:24.619
article and AI is ? And then are you

31:24.619 --> 31:26.619
crediting them in the article ? And

31:26.619 --> 31:28.730
then does that make um the trust that

31:28.730 --> 31:30.897
we talked about yesterday go down ? So

31:30.897 --> 31:33.989
Doty 5400.19 AI in public affairs makes

31:33.989 --> 31:37.180
it very clear if you use AI , you admit

31:37.180 --> 31:39.410
that you use AI . When I went through

31:39.410 --> 31:42.250
Dinfos almost 20 years ago , we were

31:42.250 --> 31:43.972
told , put an editor's note in

31:43.972 --> 31:46.028
parentheses if you had help with the

31:46.028 --> 31:47.972
article or if you got it from some

31:47.972 --> 31:50.139
somewhere else . Why not do that now ?

31:50.139 --> 31:52.139
Why not put an editor's note at the

31:52.139 --> 31:54.250
bottom ? I use at the bottom , not at

31:54.250 --> 31:56.329
the byline . In my mind , me writing

31:56.329 --> 31:58.662
something , sitting down and writing it .

31:58.662 --> 32:00.662
Yeah , excuse me , if you're taking

32:00.662 --> 32:00.530
notes or whatever it is , you're

32:00.530 --> 32:02.641
getting facts . You got the quotes in

32:02.641 --> 32:04.863
there , but you're not sitting down and

32:04.863 --> 32:07.086
writing it . AI is , so why wouldn't AI

32:07.086 --> 32:09.252
be the byline ? You directed it . It's

32:09.252 --> 32:11.419
your creativity . The AI isn't as good

32:11.419 --> 32:13.530
be the source at the bottom . I would

32:13.530 --> 32:15.474
put it in an editor's note because

32:15.474 --> 32:17.530
that's still a transparency that the

32:17.530 --> 32:19.586
public requires , right ? We want to

32:19.586 --> 32:21.641
always be transparent if we use AI .

32:21.641 --> 32:23.752
That's what the dode tells we have to

32:23.752 --> 32:25.919
do . So we have to do that . We'll put

32:25.919 --> 32:28.141
an editor's note that says , or this is

32:28.141 --> 32:27.890
still like a conversation , right ?

32:27.930 --> 32:30.097
We're still in the , we're in , in the

32:30.097 --> 32:31.930
middle of figuring out what this

32:31.930 --> 32:33.930
revolution means . We're , I'm just

32:33.930 --> 32:33.619
like poking the bear , so we just ,

32:33.660 --> 32:35.827
just kind of getting that conversation

32:35.827 --> 32:37.827
started and everybody else thinking

32:37.827 --> 32:39.938
because on the side I'm an author . I

32:39.938 --> 32:39.900
don't use AI to use my writing because

32:39.900 --> 32:42.122
it's my creative process , and I do the

32:42.122 --> 32:44.233
same thing at work . It's my creative

32:44.233 --> 32:46.400
process to write , and AI is great for

32:46.400 --> 32:48.622
things like DSIs and that kind of stuff

32:48.622 --> 32:50.622
and doing performance , but doesn't

32:50.622 --> 32:52.456
that take away from what we were

32:52.456 --> 32:54.567
trained to do to write , and that's ,

32:54.567 --> 32:57.859
I'll stop there . It does But where do

32:57.859 --> 33:01.650
we need the people ? Sorry ? Sure .

33:03.739 --> 33:05.239
OK , so I do public health

33:05.239 --> 33:06.795
communication . A lot of my

33:06.795 --> 33:09.017
communication is designed to persuade ,

33:09.017 --> 33:11.750
change behavior , not just inform or

33:11.750 --> 33:13.861
increase knowledge . And so I've been

33:13.861 --> 33:17.069
using personas to help . I'm wondering

33:17.069 --> 33:19.402
if you , if you have any prompts or you ,

33:19.402 --> 33:21.625
if you've explored that at all with any

33:21.625 --> 33:23.625
of your communication . I , mine is

33:23.625 --> 33:23.510
called Sam , and I have one called

33:23.510 --> 33:25.810
Micah . They're very helpful , yeah .

33:27.800 --> 33:31.390
Um , It's interesting ,

33:31.689 --> 33:34.609
the , so Janet Upmill is apparently

33:34.609 --> 33:37.890
opening up soon to work with Office 365 .

33:38.810 --> 33:41.032
So then you could have a , I don't know

33:41.032 --> 33:42.866
if this is really answering your

33:42.866 --> 33:45.032
question or if you just wanted to like

33:45.032 --> 33:47.310
geek out , and we can totally geek out ,

33:47.310 --> 33:49.829
um , but . Once you're able to use AI

33:49.829 --> 33:52.939
to read your emails . To read your

33:52.939 --> 33:55.420
calendar , to read your SharePoints .

33:56.780 --> 33:58.891
That is where we have to move to . We

33:58.891 --> 34:01.002
have to move to a time when we're not

34:01.002 --> 34:03.239
just talking with a chatbot . We're

34:03.239 --> 34:05.517
talking with an AI that understands us ,

34:05.599 --> 34:07.599
that understands our mission , that

34:07.599 --> 34:09.821
understands what we need to get the job

34:09.821 --> 34:11.821
done , what we need to maintain the

34:11.821 --> 34:14.449
public's trust . If we keep that as our ,

34:14.590 --> 34:17.129
our lighthouse , our guiding light ,

34:17.590 --> 34:19.312
the public's trust is the most

34:19.312 --> 34:21.479
important thing that we need , that we

34:21.479 --> 34:23.590
have or don't have , according to the

34:23.590 --> 34:25.757
slide yesterday at 46% . How do we get

34:25.757 --> 34:28.969
that back ? Did I answer your question

34:28.969 --> 34:31.429
or I just kind of like . Talk , I guess

34:31.429 --> 34:34.270
what I was getting at was , regardless

34:34.270 --> 34:36.159
of whether the AI is writing it ,

34:36.159 --> 34:38.214
you're writing it , you , the AI can

34:38.214 --> 34:41.310
assess the product against certain , uh ,

34:41.570 --> 34:43.590
intent intention like , so , so the

34:43.590 --> 34:45.646
persona , I can have a persona for a

34:45.646 --> 34:47.701
junior enlisted E3 that's married or

34:47.701 --> 34:49.812
junior enlisted E3 that's single or ,

34:49.812 --> 34:51.923
or NCO with , uh , same , same , same

34:51.923 --> 34:54.090
all the way up to senior leaders . And

34:54.090 --> 34:56.146
I can tailor the communication or at

34:56.146 --> 34:58.469
least uh uh have uh evaluated against

34:58.469 --> 35:01.050
that persona to see if I'm , and , and

35:01.050 --> 35:03.250
see how that's impacting reach , how

35:03.250 --> 35:06.820
that's impacting um Uh ,

35:07.139 --> 35:10.620
uh , uh , better engagement , uh , with ,

35:10.820 --> 35:13.500
with the article to , to lead to that

35:13.500 --> 35:15.556
hopefully intended outcome , and you

35:15.556 --> 35:17.778
can add metric audience to assessment ,

35:17.778 --> 35:19.611
correct , yes , right , audience

35:19.611 --> 35:21.667
assessment , um , or you're going to

35:21.667 --> 35:23.833
red team the information you put out .

35:23.833 --> 35:26.780
You want to understand if , um , late

35:26.780 --> 35:30.050
adopters , uh , or if women in their

35:30.050 --> 35:33.639
50s , or if men in their 60s , or if

35:34.219 --> 35:37.379
Um , people of a certain cultural

35:37.379 --> 35:39.889
background , how the , this , uh ,

35:39.899 --> 35:42.066
content will resonate to them . That ,

35:42.066 --> 35:44.177
and so you built all these personas ,

35:44.177 --> 35:47.250
uh , so that way you can foreshadow ,

35:47.290 --> 35:49.580
you can build , you can understand how

35:49.580 --> 35:51.580
this communication plan is going to

35:51.580 --> 35:53.939
work , how , and this is where we can

35:53.939 --> 35:56.919
use AI because we can use AI to create

35:56.919 --> 36:00.020
these personas , to create this

36:00.020 --> 36:02.187
environment where we can use AI to red

36:02.187 --> 36:05.760
team . Trying to use words that are

36:05.760 --> 36:08.040
appropriate for this setting . Uh , we

36:08.040 --> 36:11.840
can red team the poop emoji out of , uh ,

36:12.080 --> 36:14.247
our content , our communication plan .

36:14.247 --> 36:17.360
We want to understand , uh , we don't

36:17.360 --> 36:19.582
have enough people to do that anymore ,

36:19.582 --> 36:21.638
so we can create these personas like

36:21.638 --> 36:24.800
you said . That replicate that

36:24.879 --> 36:27.760
cultural background . So that way , if

36:27.760 --> 36:30.879
we're in Japan , Misawa Airbase , we

36:30.879 --> 36:32.823
want to make sure that the content

36:32.823 --> 36:34.979
we're putting out isn't going to harm

36:35.120 --> 36:37.360
our relations with Japanese . Is that

36:37.360 --> 36:40.370
kind of , yeah . So build those

36:40.370 --> 36:43.590
personas so we can understand . Mr .

36:43.629 --> 36:45.669
Coslett , hello , Ben . Um , so a

36:45.669 --> 36:47.780
couple of questions to , to play with

36:47.780 --> 36:49.949
the bear . Uh , we've had , I , I

36:49.949 --> 36:51.949
believe , uh , you educated me last

36:51.949 --> 36:53.893
night , we've had AI for decades .

36:54.620 --> 36:58.469
Autocorrect , autoc that

36:58.469 --> 37:01.629
it's still AI . So it , but it's still

37:01.629 --> 37:04.510
AI . So , uh , basically anything

37:04.510 --> 37:08.010
that's auto in your Adobe software is a

37:08.510 --> 37:10.510
generative , it's an algorithm with

37:10.510 --> 37:12.843
generative AI is just a catchphrase now .

37:12.989 --> 37:16.250
So what would you recommend on when .

37:16.860 --> 37:19.082
You know , you're generating it . You ,

37:19.082 --> 37:21.082
you've , that's your photo , you're

37:21.082 --> 37:23.249
editing it . Something else is helping

37:23.249 --> 37:23.219
you to move levels versus something

37:23.219 --> 37:26.939
helping you to review words . Versus .

37:27.889 --> 37:30.449
What we have to do to counter the 7th

37:30.449 --> 37:32.505
warfighting function , because right

37:32.505 --> 37:35.209
now our enemies . Exponentially

37:35.209 --> 37:37.209
outperforming us , but yet we still

37:37.209 --> 37:39.431
have to protect the truth . Second part

37:39.431 --> 37:41.376
of the question , so where in your

37:41.376 --> 37:43.431
opinion , if you were looking at the

37:43.431 --> 37:45.542
policy and you could write the policy

37:45.542 --> 37:47.209
for the day , where should it

37:47.209 --> 37:50.010
identifying AI go ? And then part two ,

37:50.250 --> 37:53.090
especially for the schoolhouse . How do

37:53.090 --> 37:55.709
we get after the human element ? So ,

37:55.850 --> 37:58.072
the younger generation , some folks are

37:58.072 --> 38:00.294
graduating with a degree and don't know

38:00.294 --> 38:02.406
nothing because they use so much AI .

38:02.406 --> 38:04.128
How do we protect that in this

38:04.128 --> 38:05.906
schoolhouse , in your , in your

38:05.906 --> 38:05.310
expertise ?

38:10.729 --> 38:12.840
He's right . AI has been around since

38:12.840 --> 38:16.469
the 1950s . It's been

38:16.469 --> 38:18.979
progressing and generative AI hit the

38:18.979 --> 38:22.050
scene in 2023 when Chat GPT uh became

38:22.050 --> 38:24.383
super popular , right ? And that became ,

38:24.383 --> 38:26.606
everyone's like generative AI , right ?

38:28.560 --> 38:31.750
But we've had an AI for But as long as

38:31.750 --> 38:35.300
the B-52 has been flying . Technically ,

38:35.679 --> 38:37.790
although back then it was like a full

38:37.790 --> 38:41.000
room of uh massive machines doing it ,

38:41.280 --> 38:45.280
um . And he's right too , that anytime

38:45.280 --> 38:47.800
you ever use the auto enhancer auto

38:47.800 --> 38:51.389
levels or auto white balance . It's

38:51.389 --> 38:55.350
AI So where do we

38:55.350 --> 38:58.780
draw the line that 5040.02 talks about ,

38:58.830 --> 39:00.941
the manipulation versus enhancement .

39:02.000 --> 39:04.167
What I was always taught of was if you

39:04.167 --> 39:07.459
can do it in a dark room , then that's

39:07.459 --> 39:10.169
enhancement , but if you can't , then

39:10.169 --> 39:14.149
it's manipulation . So When

39:14.149 --> 39:16.371
we're talking about the Aquafina bottle

39:16.371 --> 39:17.982
yesterday . That would be an

39:17.982 --> 39:21.260
illustration , not a photo or a

39:21.260 --> 39:25.159
straight video . If we're talking about

39:25.580 --> 39:27.949
generating Articles .

39:29.439 --> 39:32.199
The point that we on the Air Force side

39:32.199 --> 39:34.479
are embracing AI writing the articles

39:34.479 --> 39:38.030
for us is because we understand there's

39:38.030 --> 39:40.199
other work that we have to do as PA

39:40.199 --> 39:42.959
that's more important than sticking

39:42.959 --> 39:45.181
with writing the article . What we have

39:45.181 --> 39:47.719
to do is use this moment in time to

39:47.719 --> 39:50.889
understand that . We don't have to own

39:50.889 --> 39:52.833
the byline . Maybe the byline just

39:52.833 --> 39:55.090
becomes the shop . This , this article

39:55.090 --> 39:57.257
is written by the thirty-fifth Fighter

39:57.257 --> 39:59.257
Wing . Editor's note , we use AI to

39:59.257 --> 40:01.312
write this article . Then the byline

40:01.312 --> 40:03.479
isn't the human , it's the shop coming

40:03.479 --> 40:06.120
together as a team , uh , to produce

40:06.120 --> 40:09.350
content , but also Now , we have a

40:09.350 --> 40:11.294
communication plan for this single

40:11.294 --> 40:13.406
article , and this communication plan

40:13.406 --> 40:15.461
says , we got to write the article ,

40:15.461 --> 40:17.628
but also , we have to see if the media

40:17.628 --> 40:19.794
would be interested in this story . We

40:19.794 --> 40:21.961
also have to see if this could improve

40:21.961 --> 40:23.739
our relations with the honorary

40:23.739 --> 40:25.850
commander's program . We also have to

40:25.850 --> 40:27.628
see if this story would lead to

40:27.628 --> 40:30.810
training better for a crisis event . If

40:30.810 --> 40:33.629
the comp plan is thinking about our job

40:33.629 --> 40:37.050
as a whole picture and not just one

40:37.050 --> 40:39.790
article or one social media post ,

40:40.409 --> 40:44.070
think in campaigns . Not in the

40:44.070 --> 40:47.469
singular product . We have to think big

40:47.469 --> 40:50.030
picture with the public affairs moving

40:50.030 --> 40:52.389
forward because AI can do , it can

40:52.389 --> 40:54.669
write the comp plan , but then we

40:54.669 --> 40:56.725
review and we make it better because

40:56.725 --> 40:58.669
that's what we do . We're humans .

40:58.669 --> 41:00.447
Humans are very good at solving

41:00.447 --> 41:02.550
problems . AI is very good at

41:02.550 --> 41:05.189
replicating us , solving problems , but

41:05.189 --> 41:07.100
it's still us .

41:09.550 --> 41:11.661
Does that kind of help a little bit .

41:14.209 --> 41:17.689
Yeah , OK , no , it's great , we have

41:17.689 --> 41:19.800
to poke the bear . We have to do this

41:19.800 --> 41:23.300
better . This is the . Mhm I haven't

41:23.300 --> 41:25.356
had enough coffee , but I mean , the

41:25.356 --> 41:27.522
nerves kind of make up for the lack of

41:27.522 --> 41:29.578
coffee , so yeah , we're good . Um .

41:29.578 --> 41:30.600
Oops .

41:38.330 --> 41:40.689
So , uh , I've noticed it , and I'm

41:40.689 --> 41:43.129
sure everyone here maybe is falling ,

41:43.250 --> 41:47.209
you know , on a side , but there is a

41:47.209 --> 41:49.889
pushback , you know , in the general

41:49.889 --> 41:53.459
community about using AI . Um , you

41:53.459 --> 41:55.719
know , and it's whether it's through

41:55.719 --> 41:57.886
actors guilds or , you know , wherever

41:57.886 --> 41:59.830
it's coming from popular culture ,

41:59.830 --> 42:03.800
there , there is a , a pushback of

42:04.270 --> 42:07.439
using AI . From your position , you

42:07.439 --> 42:10.560
know , um , we're being told , you know ,

42:10.679 --> 42:12.790
to use it in the military , that it's

42:12.790 --> 42:15.239
OK , but there's still this cultural

42:15.239 --> 42:17.679
pushback . What's , uh , a

42:17.679 --> 42:19.623
recommendation ? How do you get at

42:19.623 --> 42:21.790
maybe overcoming that ? I know there's

42:21.790 --> 42:23.901
people in my shop that just refuse to

42:23.901 --> 42:26.068
use it , period . They're not going to

42:26.068 --> 42:26.000
say that out loud because there's a

42:26.000 --> 42:28.320
kind of a directive to use it . My wife

42:28.320 --> 42:30.487
also hates AI , but yeah , where , you

42:30.487 --> 42:32.542
know , where , what's the best , and

42:32.542 --> 42:34.709
you're kind of doing it through your ,

42:34.709 --> 42:36.280
um . You know , bringing up

42:36.280 --> 42:38.391
recommendations and stuff like that ,

42:38.391 --> 42:40.558
but is there a better way or how , how

42:40.558 --> 42:42.724
would we make , you know , maybe build

42:42.724 --> 42:44.891
the trust on our team , because that's

42:44.891 --> 42:46.836
maybe a trust issue . Maybe people

42:46.836 --> 42:49.058
don't want to use it because they don't

42:49.058 --> 42:51.113
trust it or understand it or there's

42:51.113 --> 42:53.336
some fear of a skill set that they have

42:53.336 --> 42:55.447
that might be replaced or that they ,

42:55.447 --> 42:55.340
you know , that they're replaceable by

42:55.340 --> 42:57.889
this , by AI . How do we get at , like ,

42:57.939 --> 43:00.161
you know , making everybody comfortable

43:00.161 --> 43:02.610
with that ? What we've seen in

43:02.610 --> 43:04.666
researching the industry is that the

43:04.666 --> 43:06.929
industry is looking for people who are

43:06.929 --> 43:09.639
really smart on a specific subject ,

43:10.010 --> 43:11.570
but also are AI fluent .

43:13.879 --> 43:16.212
So how do we do that for public affairs ?

43:16.212 --> 43:19.250
Well , That's why DNOs exists . Dinfos

43:19.250 --> 43:21.770
exists to make us really smart about PA ,

43:22.129 --> 43:24.296
to make us really smart about building

43:24.296 --> 43:26.570
a command , a campaign that honors a

43:26.570 --> 43:28.792
communication plan . We're really smart

43:29.290 --> 43:32.929
or we need to be . About how we

43:33.110 --> 43:35.949
build public trust , and we can use AI

43:36.310 --> 43:38.510
to shift that , but you're right . How

43:38.510 --> 43:40.177
do we get through this change

43:40.177 --> 43:42.620
management , uh , moment in time where

43:43.350 --> 43:45.629
Like my wife , who will spend as a

43:45.629 --> 43:47.909
speech therapist , she spends hours in

43:47.909 --> 43:50.699
the evening writing her notes . And I

43:50.699 --> 43:52.755
keep telling her I could write you a

43:52.755 --> 43:56.169
prompt for that . She's like , I hate

43:56.169 --> 44:00.080
AI . The data centers , they're

44:00.080 --> 44:02.540
killing all of the fishes and things .

44:03.040 --> 44:06.340
But how much of that is I owe from the

44:06.340 --> 44:09.889
enemy ? We have to be

44:09.889 --> 44:13.530
smarter , and we have to be open ,

44:13.949 --> 44:16.010
which is easier for me to say cause

44:16.629 --> 44:18.796
clearly I'm very into AI and I love AI

44:18.796 --> 44:21.310
and More AI , um .

44:23.100 --> 44:25.156
But it's a change management issue ,

44:25.156 --> 44:28.530
and the way that it seems like , uh ,

44:28.689 --> 44:30.800
hasn't worked with my wife yet , um ,

44:30.800 --> 44:34.639
but we have to Show

44:34.639 --> 44:38.620
people . How it's useful . We have to

44:38.620 --> 44:40.564
show people , hey , you just spent

44:40.564 --> 44:42.787
hours writing that civilian appraisal ,

44:42.860 --> 44:45.550
uh , for your , so you're the 14 , you

44:45.550 --> 44:49.219
just wrote this year 12 , um , their

44:49.219 --> 44:51.570
appraisal . But what if it only took

44:51.570 --> 44:54.260
you 30 minutes because you use AI and

44:54.260 --> 44:56.371
now you have that time back to , um ,

44:57.550 --> 45:00.189
In one of my boss's examples to watch

45:00.189 --> 45:02.300
the brewers do amazing or something ,

45:02.300 --> 45:06.120
right ? So , If we

45:06.620 --> 45:08.731
look at it from a change management ,

45:08.731 --> 45:10.953
and I'm sure a Sergeant Major is really

45:10.953 --> 45:13.064
smart about change management , and I

45:13.064 --> 45:15.176
need to learn from you , uh , we have

45:15.176 --> 45:17.342
to approach people at every level with

45:17.342 --> 45:19.676
like the personas we brought up earlier .

45:19.676 --> 45:21.731
We have people at every level of the

45:21.731 --> 45:24.739
adopter , uh , bell curve , and

45:24.739 --> 45:27.409
understanding . If we show people how

45:27.409 --> 45:29.610
it's useful , um , if we show people

45:29.610 --> 45:31.721
how we're saving time using it , then

45:31.721 --> 45:34.030
maybe , um , we have a chance . But

45:34.739 --> 45:38.010
until the majority of people are moving

45:38.010 --> 45:40.177
forward with it , we're still going to

45:40.177 --> 45:42.288
be stuck in this minority that we are

45:42.288 --> 45:44.649
right now of , well , if two people in

45:44.649 --> 45:46.871
my shop are using it , then it must not

45:46.871 --> 45:49.038
be good , it must be terrible . But if

45:49.038 --> 45:52.669
most , sorry . But if most of my shop

45:52.669 --> 45:55.389
is using it , then maybe we have this

45:55.389 --> 45:58.620
FOMO . We can bring the late adopters

45:58.620 --> 46:00.787
on board . Now , does that answer your

46:00.787 --> 46:02.320
question ? Yeah .

46:05.649 --> 46:07.816
One of the aspects of it that I didn't

46:07.816 --> 46:11.030
bring up is that . Produce a product

46:11.030 --> 46:13.689
that says AI generated .

46:15.830 --> 46:19.209
That Criticized

46:20.860 --> 46:24.100
Uh So then be smart about that when

46:24.100 --> 46:25.989
you're writing your comp plan and

46:25.989 --> 46:28.211
you're building the campaign , what has

46:28.211 --> 46:31.659
to be human only ? Based on the

46:31.659 --> 46:33.881
audience . What's the audience appetite

46:33.881 --> 46:35.881
for AI versus what's the audience's

46:35.881 --> 46:37.937
appetite for human only ? We have to

46:37.937 --> 46:40.048
understand where to put it , where we

46:40.048 --> 46:43.870
can save time , what is , uh , And

46:44.020 --> 46:46.131
it's interesting too , you can see in

46:46.131 --> 46:47.964
the industry that it seems to be

46:47.964 --> 46:50.020
there's , there's gonna be this push

46:50.020 --> 46:51.853
where , uh , like in Hollywood ,

46:51.853 --> 46:54.540
human-only uh content will be more

46:54.540 --> 46:57.580
expensive than the AI assisted content .

46:57.939 --> 47:00.050
So then , like in the food industry ,

47:00.050 --> 47:02.106
you've got organic stuff and the not

47:02.106 --> 47:04.217
organic stuff . Organic stuff is more

47:04.217 --> 47:06.439
expensive . So then human-only stuff is

47:06.439 --> 47:10.020
gonna be more expensive . So I don't

47:10.020 --> 47:12.076
know if that answers your question .

47:12.076 --> 47:14.298
That's another rabbit hole I was almost

47:14.298 --> 47:16.409
to fall down . Maybe not . I see your

47:16.409 --> 47:19.429
face , it's not working . Sounds like

47:19.429 --> 47:21.651
we need to have a , a working group and

47:21.651 --> 47:24.610
talk about this more in depth . Ma'am ,

47:25.419 --> 47:27.475
you kind of stepped on my question .

47:27.475 --> 47:29.808
Good morning , Master Sergeant Stratton .

47:29.808 --> 47:31.586
Thank you so much . Um , and my

47:31.586 --> 47:33.641
question really is probably more for

47:33.641 --> 47:35.808
the forum . Um , which is , I see that

47:35.808 --> 47:37.919
the Air Force is very pushing forward

47:37.919 --> 47:40.141
on the tip of the spear of this , but I

47:40.141 --> 47:42.363
was curious if there was a community of

47:42.363 --> 47:44.475
practice or something with all of the

47:44.475 --> 47:46.363
services and perhaps even federal

47:46.363 --> 47:48.320
agencies that was included in this

47:48.320 --> 47:50.542
research or is being considered for the

47:50.542 --> 47:52.653
future . What I would like to know is

47:52.653 --> 47:55.580
if there are other M's in the services ,

47:55.800 --> 47:57.967
so I can collaborate with them . Right

47:57.967 --> 48:00.300
now , I don't know of anybody else . Uh ,

48:00.300 --> 48:02.989
yeah , there is a federal communicators

48:02.989 --> 48:06.010
practice in AI . I they , they meet

48:06.010 --> 48:08.177
weekly . Oh , yeah , I'll give you the

48:08.177 --> 48:10.343
information . Excellent . I would love

48:10.343 --> 48:13.719
to join that . Uh , so I would love to

48:13.719 --> 48:15.886
know , I would love to meet , um , the

48:15.886 --> 48:19.350
other , if there's PA RDT&E for the

48:19.350 --> 48:21.461
other service branches , I would love

48:21.461 --> 48:23.128
to meet them . Uh , so we can

48:23.128 --> 48:25.239
collaborate and we can build together

48:25.239 --> 48:28.350
and again , on silo . Sir ,

48:29.600 --> 48:31.822
Thank you so much for your presentation

48:31.822 --> 48:33.933
and sharing these resources . Uh , my

48:33.933 --> 48:36.909
question is , how are you ? What

48:36.909 --> 48:39.739
qualitative metrics or ways of

48:39.739 --> 48:41.906
assessing the outputs from , from this

48:41.906 --> 48:44.669
writing process are you using ? Um , I

48:44.669 --> 48:46.558
run a team right now . I have two

48:46.558 --> 48:48.939
junior writers . They're using way too

48:48.939 --> 48:52.439
much AI and I , I can tell that it's ,

48:52.510 --> 48:54.590
it , the , the , the articles we put

48:54.590 --> 48:56.812
out , the information we put out , it's

48:56.812 --> 48:59.629
not tracking as well . So what , what

48:59.629 --> 49:01.851
processes are you using to , you know ,

49:02.149 --> 49:05.169
assess these , these time savings and

49:05.389 --> 49:07.750
audience engagement ? Are you seeing

49:07.750 --> 49:09.979
anything worthwhile on that end , or is

49:09.979 --> 49:12.790
this more of the input and , and

49:12.790 --> 49:13.901
productivity side ?

49:16.919 --> 49:20.360
Uh , right now , uh , AFP

49:20.360 --> 49:23.649
OSZ emerging technology is me . OK .

49:24.370 --> 49:26.219
Um , I'm supposed to get a tech

49:26.219 --> 49:28.163
sergeant in in the fall , and then

49:28.163 --> 49:30.275
it'll be the two of us doing this for

49:30.275 --> 49:32.580
the DAPA Enterprise , um .

49:34.409 --> 49:37.090
So we don't have a good way to track

49:37.090 --> 49:40.600
metrics , um . We don't have ,

49:40.810 --> 49:42.977
because the , we're still at the early

49:42.977 --> 49:45.143
stages here where we don't have a good

49:45.143 --> 49:47.254
way to connect with the field , which

49:47.254 --> 49:49.850
is why Quil Alphas uh is still building

49:49.850 --> 49:53.449
up an existing , um , and I know my

49:53.449 --> 49:55.782
inbox is flooded with , uh , people who ,

49:55.850 --> 49:58.072
uh , so the Matchcoms have been sending

49:58.072 --> 50:00.239
us people , uh , while I've been TDI I

50:00.239 --> 50:02.183
was in Lansing last week for a TDI

50:02.183 --> 50:04.370
supporting , uh , NorthCom , and here

50:04.370 --> 50:06.850
this week , I just We're , we're ,

50:06.989 --> 50:08.878
we're still building up the alpha

50:08.878 --> 50:11.156
network . We're still building up this ,

50:11.156 --> 50:13.211
uh , council of PA practitioners who

50:13.211 --> 50:15.322
want to be smart about AI and help us

50:15.322 --> 50:17.545
move forward , but we , we're not there

50:17.545 --> 50:19.656
yet . We're still early stages here ,

50:19.656 --> 50:21.822
and that's not the answer you wanted .

50:21.822 --> 50:23.878
I'm sorry . No , that's fine . I was

50:23.878 --> 50:23.790
just wondering . If , if you've done we

50:23.790 --> 50:26.590
need to get testing on , on audience

50:26.590 --> 50:28.701
engagement with content , yeah , uh ,

50:28.701 --> 50:30.812
we need to get there because , but we

50:30.812 --> 50:33.146
need the field to give us that feedback ,

50:33.146 --> 50:35.090
and we just , we like , we haven't

50:35.090 --> 50:37.090
built that , um , we're still early

50:37.090 --> 50:38.979
stages , and if we could get more

50:38.979 --> 50:41.034
people , human on the loop , uh , to

50:41.034 --> 50:43.239
work with , uh , to get more data from

50:43.239 --> 50:45.350
the field , then , um , that would be

50:45.350 --> 50:47.572
amazing . So if you have contacts and ,

50:47.572 --> 50:49.517
uh , we can , uh , work that out ,

50:49.517 --> 50:51.461
let's do that . Thank you , yeah .

50:55.909 --> 50:57.965
When's the training tomorrow ? Where

50:57.965 --> 51:01.770
where is it ? The training with the war

51:01.770 --> 51:05.090
media activity is across the street at

51:05.090 --> 51:07.312
the war media activity and it starts at

51:07.312 --> 51:09.479
8:15 . Now , the room , that's a great

51:09.479 --> 51:11.701
question , but Chief Scaff told me this

51:11.701 --> 51:13.923
is like a TED Talk thing and that makes

51:13.923 --> 51:16.034
me even more anxious than this room .

51:16.034 --> 51:17.979
Uh , so , uh , I probably have all

51:17.979 --> 51:19.979
sorts of sweat on my , uh , magnets

51:19.979 --> 51:23.449
here . Um . So , yeah , 8:15 across the

51:23.449 --> 51:26.659
street at the WMA . I have another

51:26.659 --> 51:29.280
question to build on that , as we see

51:29.300 --> 51:31.467
AI being used more and more to write ,

51:31.580 --> 51:33.479
how are you picturing , how do we

51:33.479 --> 51:35.659
continue to grow and develop and make

51:35.659 --> 51:37.603
sure that our junior employees are

51:37.603 --> 51:39.715
maintaining their skills ? If we have

51:39.715 --> 51:41.881
staff that's coming in who has used AI

51:41.881 --> 51:43.881
since they were children , which is

51:43.881 --> 51:45.659
what's happening as the younger

51:45.659 --> 51:47.715
generation comes in , and if we have

51:47.715 --> 51:47.419
staff that's relying on AI to write

51:47.419 --> 51:49.419
things for them , I find that I'm ,

51:49.419 --> 51:52.449
we're losing some . Foundational skills ,

51:52.610 --> 51:54.721
uh , both for our writers and for our

51:54.721 --> 51:56.943
editorial staff . Our editors coming in

51:56.943 --> 51:59.850
are just not . Quite there . So , how

51:59.850 --> 52:01.729
do we protect the foundational

52:01.729 --> 52:04.050
knowledge that we need to verify the AI

52:04.050 --> 52:06.429
is correct and the AI is doing its job ,

52:06.810 --> 52:08.532
also leveraging these skills ,

52:08.620 --> 52:10.787
particular or leveraging these tools ,

52:10.787 --> 52:13.360
particularly in writing . We were

52:13.360 --> 52:16.850
already losing trade value , expertise

52:17.129 --> 52:19.489
before AI hit the scene . You look at

52:19.489 --> 52:21.879
Air Force and we've been , now we're

52:23.260 --> 52:25.371
When I came in , we could go out as a

52:25.371 --> 52:27.649
PB&J team , photographer , broadcaster ,

52:27.649 --> 52:29.939
and journalist . And I was really good

52:29.939 --> 52:32.217
at writing , I was , probably still am .

52:32.217 --> 52:34.050
I don't know . I haven't written

52:34.050 --> 52:36.328
articles in a while , um , or I use AI .

52:36.328 --> 52:39.310
So I was the journalist on the team

52:39.310 --> 52:40.977
with the photographer and the

52:40.977 --> 52:43.088
broadcaster , so then we weren't jack

52:43.088 --> 52:45.254
of all trades , master of none , but .

52:45.254 --> 52:47.421
Is that not where we're at today ? Was

52:47.421 --> 52:49.600
that not where we're at in 2022 before

52:49.600 --> 52:51.433
AI really hit the scene with the

52:51.433 --> 52:53.750
generative AI ? How do we get , how do

52:53.750 --> 52:55.989
we make sure that that training is

52:55.989 --> 52:59.790
still out there ? Well , If we

52:59.790 --> 53:02.012
don't have the expertise in the field ,

53:02.012 --> 53:04.510
which there isn't a whole lot , um ,

53:04.790 --> 53:07.600
and all of us old writers are , um ,

53:07.909 --> 53:09.820
getting ready to retire , um ,

53:12.669 --> 53:15.189
We can use AI , which is why the

53:15.189 --> 53:18.370
prompts have the coach built into them .

53:19.379 --> 53:21.729
And the , the purpose of having coach

53:21.729 --> 53:25.340
built into it is to augment

53:25.620 --> 53:28.300
the expertise in the field , uh , to

53:28.300 --> 53:30.411
help train us on how to write . Now ,

53:31.030 --> 53:33.252
Does Hollyhey know how to write as well

53:33.252 --> 53:35.419
as Dinfos trained us back in the day ?

53:35.560 --> 53:38.969
Probably not . But how do we get that

53:38.969 --> 53:40.889
training in the field ? That's a

53:40.889 --> 53:43.111
question , uh , for a working group , I

53:43.111 --> 53:45.389
think , cause I don't have the answers .

53:45.389 --> 53:48.949
Uh , I would love to see . More

53:48.949 --> 53:51.350
writing skills , um , brought back . I

53:51.350 --> 53:55.340
would love to see more . Less

53:55.340 --> 53:57.173
of the jack of all trades , uh ,

53:57.173 --> 53:59.284
approach , but probably not gonna get

53:59.284 --> 54:01.562
that in the current budget environment ,

54:01.562 --> 54:04.560
um . That doesn't really answer your

54:04.560 --> 54:06.671
question or give you warm and fuzzy ,

54:06.671 --> 54:08.893
but I don't know that there is any warm

54:08.893 --> 54:11.360
and fuzzy to give , uh , unless we can

54:11.360 --> 54:15.179
bring back , uh , BPW . Um ,

54:15.840 --> 54:17.618
that , that doesn't answer your

54:17.618 --> 54:17.600
question . That was , that was a

54:17.600 --> 54:20.000
terrible response . I'm sorry . So the ,

54:20.159 --> 54:22.381
so the Air Force does a pretty good job

54:22.381 --> 54:24.548
with this , with other career fields .

54:24.548 --> 54:26.659
It's an exercise . It would be like a

54:26.659 --> 54:28.881
communication out , meaning you have no

54:28.881 --> 54:30.937
computers , you're going back to pad

54:30.937 --> 54:33.320
and pen , papers , and your leaderships

54:33.320 --> 54:35.153
run that . Now , AI can help you

54:35.153 --> 54:37.439
generate this . So at a wing , I've sat

54:37.439 --> 54:40.479
down and done an ORI level exercise ,

54:40.560 --> 54:42.727
which normally takes about 120 days to

54:42.727 --> 54:45.810
plan . 15 to 40 people , experts to put

54:45.810 --> 54:48.032
it in a room . I can put them in a room

54:48.032 --> 54:50.989
and in 4 hours give you a 120 day plan

54:51.330 --> 54:53.729
of metal for that exercise . We can do

54:53.729 --> 54:57.379
the same in PA . The medical . Practice

54:57.379 --> 54:59.620
the field . They are required to have

54:59.620 --> 55:01.398
combat skills that they have to

55:01.398 --> 55:03.398
maintain , but when they're here in

55:03.398 --> 55:05.731
Konis , they don't get to practice them .

55:05.731 --> 55:07.898
What do they do ? They exercise . They

55:07.898 --> 55:10.120
put themselves in austere conditions to

55:10.120 --> 55:12.120
practice and sharpen their skills .

55:12.520 --> 55:14.742
Number 2 , maintenance . I used to be a

55:14.742 --> 55:16.639
maintainer and we used to work on

55:16.639 --> 55:18.695
airplanes . In order to do that , we

55:18.695 --> 55:20.919
would have quality assurance . In each

55:20.919 --> 55:24.000
quarter you can cycle and you can have

55:24.000 --> 55:26.167
a plan . I can help you make that plan

55:26.560 --> 55:28.760
to test . OK , this quarter you're

55:28.760 --> 55:30.719
going to be tested 4 times on a

55:30.719 --> 55:32.552
surprise and a moment's notice ,

55:32.552 --> 55:34.663
meaning it's a no notice inspection ,

55:34.663 --> 55:36.775
and I'm going to see if you can write

55:36.775 --> 55:38.997
this article , and then you give a pass

55:38.997 --> 55:41.163
or fail grade . You can take a photo ,

55:41.163 --> 55:43.386
pass , or feel great . So the answer is

55:43.386 --> 55:43.320
already out there in other career

55:43.320 --> 55:45.376
fields , and we've been doing it for

55:45.376 --> 55:47.542
decades . So I think it's no different

55:47.542 --> 55:49.764
for PA . So if we want , if you want to

55:49.764 --> 55:51.653
keep your skills sharp . It's the

55:51.653 --> 55:53.820
leader , it's the human in that office

55:53.820 --> 55:55.987
that sets that condition . If you're a

55:55.987 --> 55:58.320
boss that just says , give me , give me ,

55:58.320 --> 55:57.689
give me , give me , give me , give me ,

55:57.760 --> 56:00.189
you're just gonna get AI . If you do

56:00.189 --> 56:02.570
not create an atmosphere like I like 75 ,

56:02.580 --> 56:05.909
25 , I like 75% of the schedule is

56:05.909 --> 56:08.020
going to be dictated to us . We don't

56:08.020 --> 56:10.242
have a control . But as a boss , if you

56:10.242 --> 56:12.353
can control that 25% for creativity ,

56:12.353 --> 56:15.229
creativity , creativity . So for me , I

56:15.229 --> 56:17.949
told him my philosophy is automate the

56:17.949 --> 56:20.227
mundane to make time for the wonderful ,

56:20.350 --> 56:22.128
and it's up to the leader , the

56:22.128 --> 56:24.239
communicator in the room to make that

56:24.239 --> 56:26.183
happen . The answer is already out

56:26.183 --> 56:28.350
there . So to me , we can do a working

56:28.350 --> 56:28.090
group . But we could probably sit down

56:28.090 --> 56:30.257
in an afternoon and create a program ,

56:30.739 --> 56:32.739
cross all DOD and be done with it .

56:36.949 --> 56:39.227
Assume there wasn't a question in that .

56:39.929 --> 56:42.151
It was a really great statement . Hey ,

56:42.151 --> 56:43.810
uh , good good morning . I'm

56:43.810 --> 56:45.810
Carlsberger with the Army's digital

56:45.810 --> 56:48.110
media division . Uh , two things , so

56:48.489 --> 56:51.189
the Army recently , uh , last year , we

56:51.570 --> 56:53.792
tried to build a knowledge management ,

56:53.792 --> 56:55.903
uh , section within our digital media

56:55.903 --> 56:57.959
division that was doing a lot of the

56:57.959 --> 56:59.959
same things you were doing . So the

56:59.959 --> 56:59.689
first part of my statement is , I would

56:59.689 --> 57:01.967
love to actually get with you and , uh ,

57:02.060 --> 57:04.060
make that connection . We , we were

57:04.060 --> 57:06.570
trying to do what you do . Um , but my

57:06.570 --> 57:09.780
question is , Uh , with the time saving

57:10.000 --> 57:12.222
in your research and your development ,

57:12.830 --> 57:15.052
do you have a specific number , because

57:15.052 --> 57:17.108
I'm a very results driven . How much

57:17.108 --> 57:19.274
time have you found that you're saving

57:19.274 --> 57:22.159
overall ? Because it sounds excellent

57:22.159 --> 57:24.560
and it's good in theory , and I , I

57:24.560 --> 57:26.782
personally am a fan of AI and I , and I

57:26.782 --> 57:29.004
buy into everything you're saying , but

57:29.004 --> 57:31.338
for our commanders , for our leadership ,

57:31.338 --> 57:33.449
the numbers matter more . So how much

57:33.449 --> 57:35.338
time are we , have you found that

57:35.338 --> 57:37.560
you're saving doing these things versus

57:37.560 --> 57:37.340
to do them the organic way ?

57:41.699 --> 57:44.139
The problem is , like I , uh , told the

57:44.139 --> 57:46.195
gentleman earlier , I don't have the

57:46.195 --> 57:48.250
data from the field , and I need the

57:48.250 --> 57:50.417
data from the field to give you a real

57:50.417 --> 57:52.639
answer , but in the collabs that I have

57:52.639 --> 57:54.861
done , uh , and I forced them to do the

57:54.861 --> 57:57.379
5 deliverables in 45 minutes , um ,

57:57.620 --> 58:00.250
they had to , so I did , um , at the

58:00.250 --> 58:02.417
Air Force PA superintendent workshop ,

58:02.417 --> 58:05.750
um , a few weeks ago , um . I did 15 ,

58:05.800 --> 58:07.967
I had an , I had an hour with them . I

58:07.967 --> 58:09.967
gave them a 15-minute overview on ,

58:09.967 --> 58:12.189
here's Quill , and here's everything we

58:12.189 --> 58:14.356
have on the AFNET PA SharePoint , um .

58:15.199 --> 58:17.143
Now , here are the five things you

58:17.143 --> 58:19.310
gotta do in the next 45 minutes . Uh ,

58:19.310 --> 58:22.840
so we used the Kirtland F-35 crash from

58:22.840 --> 58:25.520
2024 as our example because all the

58:25.520 --> 58:27.464
data and all the media reports are

58:27.464 --> 58:29.520
already out there . We could use the

58:29.520 --> 58:31.853
media summary and media analyst prompts .

58:31.853 --> 58:35.379
Um , so their , their job then was to

58:35.550 --> 58:37.860
draft the media advisory . Their job

58:37.860 --> 58:41.379
was also to write um an achievement

58:41.379 --> 58:43.739
medal , uh , for their airman who was

58:43.739 --> 58:46.320
the first on scene . Their job was to

58:46.560 --> 58:49.100
um create uh talking points for their

58:49.100 --> 58:51.156
boss to talk in front of the media .

58:51.156 --> 58:53.540
Their job was to um create a media

58:53.540 --> 58:55.729
summary on all of the events and

58:55.729 --> 58:57.840
articles that have been written up to

58:57.840 --> 59:00.669
that point or that point , right ? And

59:00.669 --> 59:03.000
their job was to um

59:03.949 --> 59:07.030
Prepare a crisis communication plan for

59:07.030 --> 59:09.252
future events like that . Uh , and they

59:09.252 --> 59:11.363
had to get all 5 of those things done

59:11.750 --> 59:14.820
in 45 minutes . They got all of them

59:14.820 --> 59:18.500
done in 35 minutes . So if that is a

59:18.500 --> 59:22.030
metric for you , the point is these

59:22.030 --> 59:24.030
things normally would take hours or

59:24.030 --> 59:26.086
days to do , but now we can do it in

59:26.086 --> 59:29.969
under 1 hour . Were they really

59:29.969 --> 59:32.191
good products ? I didn't get to look at

59:32.191 --> 59:34.399
them . Cause they hit their 45 minute

59:34.399 --> 59:37.540
mark and there was an ex-briefer , but

59:37.760 --> 59:41.639
the point is . Try to save time by

59:41.639 --> 59:43.760
putting AI where it can help you get

59:43.760 --> 59:45.959
the work done faster that the AI is

59:45.959 --> 59:48.219
good at . It's good at writing a comp

59:48.219 --> 59:51.649
plan , it's good at writing . A staff

59:51.649 --> 59:54.300
report article . Um ,

59:56.709 --> 01:00:00.129
Um But it's not good at the

01:00:00.129 --> 01:00:02.296
interpersonal communication . It's not

01:00:02.296 --> 01:00:04.889
good at , uh , understanding that the

01:00:04.889 --> 01:00:07.030
EOD airmen or the firehouse airmen ,

01:00:07.810 --> 01:00:09.699
well , they'll let you take their

01:00:09.699 --> 01:00:11.977
photos if you give them a case of beer ,

01:00:11.977 --> 01:00:14.088
uh , and you show up and you hang out

01:00:14.088 --> 01:00:16.254
with them and understand them . That's

01:00:16.254 --> 01:00:19.219
stuff that AI can't replace . Uh , to

01:00:19.219 --> 01:00:21.330
your metrics question , how much time

01:00:21.330 --> 01:00:23.580
are we saving ? It's , it's more so

01:00:23.580 --> 01:00:25.820
that we're saving time to redistribute

01:00:25.820 --> 01:00:27.876
and reshape where we're spending our

01:00:27.876 --> 01:00:30.489
time right now . So one final question

01:00:30.489 --> 01:00:32.711
that . So with that , the time saving ,

01:00:32.711 --> 01:00:34.878
do you find or would you , would it be

01:00:34.878 --> 01:00:36.989
a safe assumption that more time will

01:00:36.989 --> 01:00:38.989
be spent in the editorial process ?

01:00:38.989 --> 01:00:41.156
Because looking at , just speaking for

01:00:41.156 --> 01:00:42.933
myself , if I have , you know ,

01:00:42.933 --> 01:00:45.649
soldiers that are using AI , I'm going

01:00:45.649 --> 01:00:48.780
to make certain that my editorial

01:00:48.780 --> 01:00:51.649
process is more robust so that it's

01:00:51.649 --> 01:00:54.770
less of that AI feel or I understand my

01:00:54.770 --> 01:00:56.949
audience a little bit better . So ,

01:00:57.010 --> 01:00:59.066
would you say that shift in terms of

01:00:59.066 --> 01:01:00.788
creation . Go more towards the

01:01:00.788 --> 01:01:02.788
editorial to make that process that

01:01:02.788 --> 01:01:04.774
product correct , uh , versus the

01:01:04.774 --> 01:01:07.215
actual creation of the process . So ,

01:01:07.254 --> 01:01:10.084
the point behind developing an , uh ,

01:01:11.614 --> 01:01:13.725
quill prompts that are really good at

01:01:13.725 --> 01:01:16.455
writing articles is that you , if you

01:01:16.455 --> 01:01:18.534
prompt an AI effectively , it won't

01:01:18.534 --> 01:01:20.635
sound like AI . And if it does sound

01:01:20.635 --> 01:01:22.810
like AI , Uh , there is a prompt for

01:01:22.810 --> 01:01:24.810
that , just like there used to be ,

01:01:24.810 --> 01:01:27.032
there's an app for that . Now there's a

01:01:27.032 --> 01:01:29.032
prompt for that . There's a quality

01:01:29.032 --> 01:01:30.977
review prompt that is available in

01:01:30.977 --> 01:01:33.032
those prompts , and it walks through

01:01:33.032 --> 01:01:34.921
your article before you can hit ,

01:01:34.921 --> 01:01:36.921
before you hit the editorial , uh ,

01:01:36.921 --> 01:01:38.840
step . Um , it walks it through a

01:01:38.840 --> 01:01:41.590
five-step process , uh , it assesses it

01:01:41.850 --> 01:01:45.179
for , um , Copyright infringement , it

01:01:45.179 --> 01:01:47.540
assesses it for AIisms , it , it

01:01:47.540 --> 01:01:49.651
assesses it for flow , it assesses it

01:01:49.651 --> 01:01:51.860
for uh all sorts of great things . It

01:01:51.860 --> 01:01:53.971
also does a thing called pre-mortem .

01:01:54.260 --> 01:01:56.316
Um , and pre-mortem is a really cool

01:01:56.316 --> 01:01:58.500
thing that if you tell an AI , uh , do

01:01:58.500 --> 01:02:00.679
a pre-mortem assessment on this , uh ,

01:02:00.939 --> 01:02:02.661
thing that I want to put out ,

01:02:02.661 --> 01:02:05.679
pre-mortem , um , tells the AI assume

01:02:06.169 --> 01:02:08.280
everything that could go wrong did go

01:02:08.280 --> 01:02:10.300
wrong , uh , and then help me build

01:02:10.300 --> 01:02:13.080
this better so that Things don't go

01:02:13.080 --> 01:02:15.302
wrong , and then you can keep iterating

01:02:15.302 --> 01:02:17.524
on that pre-mortem concept . Everything

01:02:17.524 --> 01:02:19.719
went wrong . My comp plan was ended up

01:02:19.719 --> 01:02:21.879
being terrible , and now the media is

01:02:21.879 --> 01:02:23.899
at the door and they want answers .

01:02:24.120 --> 01:02:26.176
Well , with the pre-mortem , you can

01:02:26.176 --> 01:02:28.600
prepare better . Uh , so the quality

01:02:28.600 --> 01:02:30.879
review prompt does that for you . So ,

01:02:30.949 --> 01:02:33.280
the point is , how do we save time with

01:02:33.280 --> 01:02:35.113
AI ? Well , you can save time by

01:02:35.113 --> 01:02:37.280
writing the article , by drafting it ,

01:02:37.280 --> 01:02:39.336
then you can save time , uh , with a

01:02:39.336 --> 01:02:40.891
better review . So , Really

01:02:40.891 --> 01:02:43.113
well-structured prompts will get a much

01:02:43.113 --> 01:02:45.336
better result out of the AI . Is that ,

01:02:45.336 --> 01:02:47.870
OK . I think Mel's gonna yell at me in

01:02:47.870 --> 01:02:50.037
a second , but do we have time for one

01:02:50.037 --> 01:02:53.139
more ? No breakdown . It's your break

01:02:53.139 --> 01:02:55.620
time , OK . Eric Toomer , Strategic

01:02:55.620 --> 01:02:57.953
Communications and Public Affairs , TSA .

01:02:58.989 --> 01:03:01.219
Uh , my question is , what's being done

01:03:01.600 --> 01:03:05.399
to , uh , educate , uh , writers

01:03:05.399 --> 01:03:07.510
before they come in ? Because if , if

01:03:07.510 --> 01:03:09.566
all you've done is AI and all you've

01:03:09.566 --> 01:03:11.600
been trained to do is AI and you

01:03:11.600 --> 01:03:13.767
haven't been taught structure , when I

01:03:13.767 --> 01:03:15.822
was in school , we got taught how to

01:03:15.822 --> 01:03:18.044
write a story , we got taught how to do

01:03:18.044 --> 01:03:20.239
things long before AI existed and if

01:03:20.239 --> 01:03:22.679
you get a person into your , your job

01:03:22.679 --> 01:03:25.600
and they've only done AI . It's

01:03:25.600 --> 01:03:27.544
difficult at that point in time to

01:03:27.544 --> 01:03:29.489
teach them how to actually write a

01:03:29.489 --> 01:03:31.739
story themselves . I think AI is just

01:03:31.750 --> 01:03:34.159
an extra step in that learning journey .

01:03:34.330 --> 01:03:36.560
So like when I came through BSW , our

01:03:36.560 --> 01:03:39.199
first two weeks was gram cram , uh ,

01:03:39.879 --> 01:03:41.935
understand grammar , uh , understand

01:03:41.935 --> 01:03:44.157
the AP style book , how to memorize all

01:03:44.157 --> 01:03:46.212
this grammar that maybe weren't , we

01:03:46.212 --> 01:03:48.435
weren't taught in school . And that was

01:03:48.435 --> 01:03:50.657
step one . And then step two was honing

01:03:50.657 --> 01:03:52.935
the craft , understanding how to write .

01:03:52.935 --> 01:03:55.101
And then maybe the next step is , OK ,

01:03:55.101 --> 01:03:57.268
now that you know how to write and you

01:03:57.268 --> 01:03:59.490
know what good looks like . Let me show

01:03:59.490 --> 01:04:01.601
you what this AI tool can do and save

01:04:01.601 --> 01:04:03.823
you time . That makes sense , right ? I

01:04:03.823 --> 01:04:06.101
mean , like if we build the foundation ,

01:04:06.101 --> 01:04:08.212
then we can use AI to save the time .

01:04:08.212 --> 01:04:10.323
Does that answer your question , no ,

01:04:10.323 --> 01:04:10.080
that makes a lot of sense . OK , thank

01:04:10.080 --> 01:04:14.030
you , yeah . Guess what ?

01:04:15.050 --> 01:04:16.209
You get a tumblr .

01:04:20.439 --> 01:04:24.129
I like . Let it go . I do have some

01:04:24.129 --> 01:04:26.296
comments based on some of the stuff he

01:04:26.296 --> 01:04:28.407
said , but I didn't want to interrupt

01:04:28.407 --> 01:04:31.060
anybody , um . Those of you who are in ,

01:04:31.479 --> 01:04:34.820
I hope we're not , um , Still alive .

01:04:35.040 --> 01:04:37.120
Um , for those of you who are in the

01:04:37.120 --> 01:04:39.287
Department of War , and you're looking

01:04:39.287 --> 01:04:41.231
for some writing training for your

01:04:41.231 --> 01:04:43.679
people , give me a call , send me an

01:04:43.679 --> 01:04:45.790
email . My job is the mobile training

01:04:45.790 --> 01:04:48.929
team program manager for Denfos , and I

01:04:48.929 --> 01:04:51.479
support the Department of War across

01:04:51.479 --> 01:04:54.239
all activities , not just the services ,

01:04:54.290 --> 01:04:56.600
but across all activities . I have

01:04:56.600 --> 01:04:58.378
writing workshops that I've put

01:04:58.378 --> 01:05:00.378
together . I'm working on a grammar

01:05:00.378 --> 01:05:03.800
slam . Uh , for , uh , one of the units .

01:05:04.100 --> 01:05:07.459
Um , so I'm here to help you . I also

01:05:07.459 --> 01:05:10.360
want to let you know that we I know

01:05:10.360 --> 01:05:12.304
that uh Rhonda talked a little bit

01:05:12.304 --> 01:05:14.193
about the intermediate journalism

01:05:14.193 --> 01:05:16.360
course that they're putting together ,

01:05:16.360 --> 01:05:15.649
still talking Department of War a

01:05:15.649 --> 01:05:18.080
little bit here . Um , we're hoping

01:05:18.080 --> 01:05:20.136
that gets a little bigger and we can

01:05:20.136 --> 01:05:22.479
get more people in here for that . And

01:05:22.479 --> 01:05:24.600
then going back to our entry level ,

01:05:24.679 --> 01:05:26.830
we're not teaching them AI , we are

01:05:26.830 --> 01:05:28.830
still teaching them foundation , so

01:05:28.830 --> 01:05:31.179
they know what right looks like . The

01:05:31.179 --> 01:05:33.860
last thing I will say is there are so

01:05:33.860 --> 01:05:36.939
many great things on Pavilion that talk

01:05:36.939 --> 01:05:39.699
about writing , talk about editing ,

01:05:39.899 --> 01:05:42.199
copy editing , proofreading , giving

01:05:42.199 --> 01:05:45.580
you , uh , step by step on how to do

01:05:45.580 --> 01:05:49.020
that . Um , now , if you are outside

01:05:49.020 --> 01:05:51.620
the Department of War and you need some

01:05:51.620 --> 01:05:53.787
training , I can't give you training ,

01:05:53.860 --> 01:05:56.979
but what I can do is help you put

01:05:56.979 --> 01:05:59.201
together training for your people . I'm

01:05:59.201 --> 01:06:01.689
here to help . Um , not just here at

01:06:01.689 --> 01:06:04.129
training days , but every day . So give

01:06:04.129 --> 01:06:06.240
me a call if you have some stuff that

01:06:06.240 --> 01:06:08.407
you want me to work on . Like I said ,

01:06:08.407 --> 01:06:10.351
I can't do training outside of the

01:06:10.351 --> 01:06:12.518
Department of War , but I can show you

01:06:12.518 --> 01:06:15.790
how , and you can request materials

01:06:16.090 --> 01:06:18.370
from DNOs . There's a way to do that .

01:06:18.449 --> 01:06:20.671
You can , I can help you put together a

01:06:20.671 --> 01:06:22.838
workshop and you can request materials

01:06:22.838 --> 01:06:25.060
through DNOs . So that's kind of how we

01:06:25.060 --> 01:06:27.439
collaborate in that . So , don't give

01:06:27.439 --> 01:06:30.000
up hope . Um , many of us still believe

01:06:30.000 --> 01:06:32.600
in those foundational structures on

01:06:32.600 --> 01:06:36.399
writing . I too am a writer , um , Navy

01:06:36.399 --> 01:06:38.750
writer is how I started out , uh , Navy

01:06:38.750 --> 01:06:41.520
journalist . So , um , I'm here . And

01:06:41.520 --> 01:06:43.409
then also you've just built a big

01:06:43.409 --> 01:06:45.520
network , right ? Not yet . You gotta

01:06:45.520 --> 01:06:48.800
wait till after lunch . Um , but , um ,

01:06:49.080 --> 01:06:51.191
yeah , don't give up hope . Give me a

01:06:51.191 --> 01:06:53.413
holler . How's that sound ? Good ? OK ,

01:06:53.413 --> 01:06:55.100
you are now on break till 10:30 .