0:02 adopting new technology can sometimes be
0:05 a painful and slow process but in this
0:07 learning objective we will show you
0:10 methods to deliver AI in a way that is
0:13 faster less painful and much more
0:16 impactful to your agency's mission first
0:19 let's talk about agile Contracting so
0:21 agile has become a meaningless buzzword
0:24 in so many organizations but we are
0:27 taking it back and you will see the real
0:28 power of agile
0:31 Contracting a ility emphasizes
0:33 collaboration responsiveness and
0:35 continuous Improvement unlike
0:37 traditional procurement which is often
0:40 rigid and linear agile Contracting
0:42 adjusts and refines continuously during
0:45 a project agile fits the fast-paced
0:47 world of technology and AI where
0:50 requirements and solutions evolve
0:52 rapidly what that really means is doing
0:55 a far 15 contract
0:58 negotiation is a little getting ahead of
1:00 the game meaning I would be doing some
1:04 OTAs with maybe 6 months with maybe
1:07 six-month extensions until you start
1:10 seeing the same thing again and again
1:12 meaning oh it's the the software is
1:14 really not getting better you know over
1:17 the last six months or years that's not
1:18 been the case for the last two or three
1:21 years and what I really am concerned
1:24 about is that some prime convinces some
1:27 smart acquisition folks oh name of large
1:29 Prime here we really got thousands of
1:31 people working on this we really got it
1:34 Etc when that's I will guarantee you
1:35 that's not the case that you really
1:39 don't want a far1 15 contract right now
1:42 for most AI stuff if it's if there's
1:45 some very specific narrow things that
1:46 need to be
1:49 deployed maybe but again it' be like
1:51 dialing into the internet and building a
1:53 website in
1:55 1997 and and realizing you just wrote a
1:58 10-year contract it's obsolete by 1998
2:01 The Other Extreme is is well why don't
2:02 we wait it's going to get better in 6
2:05 months or a year or two years yes it's
2:06 like trying to decide when to buy the
2:09 right PC CU next year will be faster or
2:11 the next phone well of course next
2:12 year's will be faster but what do we
2:14 need now and how long is it going to
2:18 last in fact maybe that an analog is is
2:20 probably useful at least in the consumer
2:23 world we don't buy iPhones or Google
2:26 Android phones the last 15 years right
2:28 we buy them for our needs now knowing
2:29 that there's probably something in
2:32 another two or three years that we're
2:34 going to toss this one out or send it
2:37 back for for some refund and and get a
2:39 new one you know the dod traditionally
2:41 has a different mindset is things need
2:45 to have 30-year life cycles and mlpf and
2:48 you know whatever boy if we try to apply
2:50 those same principles right now to an
2:52 industry that's exponentially changing
2:55 we're going to lock ourselves into like
2:58 you know stuff that not only obsolete at
3:01 least for Stuff that Warf Fighters need
3:04 that will put us at a disadvantage we
3:05 need a mindset that says we're going to
3:07 need to refresh this stuff at least
3:10 every 18 months if not sooner boy that's
3:12 what kind of acquisition authorities do
3:14 we have for that that's what I'd be
3:15 thinking about if we're trying to use
3:17 the same ones that we're using for
3:21 buying windows or Macs then we're just
3:24 on the wrong path not until it
3:26 stabilizes and we're certainly not on a
3:30 stable course for AI and if you believe
3:34 that AI should be delivered in some sort
3:37 of dis and or some sort of uh iteration
3:40 of of version 1.0 and the government
3:42 then should drive what are the
3:46 requirements for version 2.0 and then
3:48 contract for that and then what are the
3:50 requirements for version
3:54 3.0 we will lose because our adversaries
3:56 are not going to wait for us to do that
4:00 in fact our adversaries have more access
4:03 to the technologists that are driving
4:06 startups all across America with this
4:09 new technology because our process is so
4:14 slow and painfully um uh linear most of
4:16 the reason actually has to do with the
4:19 resourcing process being so low so slow
4:20 and so
4:22 micromanaged when companies know that
4:25 they can approach uh congress with
4:28 requests for funding in a very narrow
4:29 account that has the name of their
4:31 program program on it the phase of their
4:33 program on it the office that can
4:37 contract for things using that money and
4:40 the compliance folks in fiscal law
4:41 inside our
4:45 agencies uh must adhere to those
4:49 guidelines and so I believe it creates a
4:51 a very very slow process by which we
4:54 create budgets over a couple of years we
4:57 get it to Congress Congress passes A
4:59 continuing resolution and is actually
5:01 happy about it as an achievement uh
5:03 while our adversaries are moving money
5:06 within 20 minutes inside of our own
5:09 markets uh talking to founders of
5:11 companies uh through intermediaries in
5:14 some cases so they can remain hidden uh
5:17 and then buying that technology buying
5:19 that company or stealing it and giving
5:21 it to Chinese companies and then locking
5:23 the United States out of the very
5:25 Innovation that our own Marketplace is
5:29 is is generating we have to find ways to
5:31 connect with industry on a much more
5:34 meaningful level and much faster than we
5:36 are the next important thing to
5:39 understand is how to make AI
5:42 stick people and organizations are often
5:45 resistant to new technology in this
5:47 graphic we see that there is a wide
5:50 Chasm where many projects go to die
5:52 between the early Market innovators and
5:54 the rest of the mainstream market the
5:58 bridge across that Chasm is concentrated
6:00 effort or organizations are juggling
6:03 numerous priorities and challenges
6:06 simultaneously adopting new technology
6:08 involves significant investment training
6:11 and risk management which can be
6:14 daunting amid other pressing concerns so
6:16 there are three main barriers to AI
6:19 adoption that we must overcome
6:23 fear inertia and
6:25 bureaucracy let's look at each of these
6:26 a little
6:30 closer fear apprehension about I can
6:31 stem from concerns about Job
6:33 displacement privacy breaches and ethical
6:34 ethical
6:37 implications there may be a fear of the
6:39 unknown or misconceptions about ai's
6:40 capabilities and
6:42 limitations addressing these fears
6:46 requires education transparency and
6:48 clear communication about the potential
6:51 benefits and risks of AI adoption as
6:53 well as strategies for mitigating any negative
6:54 negative
6:58 impacts inertia government agencies
7:00 struggle with entrenched processes
7:03 systems and cultures that resist change
7:06 adopting AI requires embracing new
7:08 technology workflows and ways of
7:12 thinking which upsets the status quo and
7:14 sometimes stir resistance overcoming
7:16 inertia requires strong leadership
7:19 organizational buyin and willingness to
7:21 challenge existing norms and
7:23 practices demonstrating the tangible
7:26 benefits of AI adoption such as
7:28 increased efficiency improve decision-
7:31 making and enhan Service delivery will
7:35 help and finally we have bureaucracy the
7:38 complex regulatory Frameworks and
7:40 bureaucratic structures of government
7:42 slow the adoption of new
7:44 technology procurement processes
7:46 compliance requirements and legal
7:50 constraints pose barriers to AI
7:53 Solutions cutting red tape requires
7:55 streamlining processes reducing
7:58 administrative burdens and fostering
8:00 collaboration Flex ible policies and
8:02 Regulatory Frameworks can accommodate
8:05 unique AI challenges and
8:08 opportunities one theme in overcoming
8:11 all of these barriers is building trust
8:13 end users must trust new AI technology
8:16 on two levels they must trust that the
8:18 new AI technology will function as
8:22 intended and as needed and two they must
8:24 trust that the AI will be used
8:27 responsibly we will cover this in much
8:30 detail in module 9 your expertise in
8:32 acquisition leadership and policy will
8:35 clear the way to successfully adopt and
8:36 integrate AI
8:39 Technologies so let's hear from a few
8:41 experts who have gone before you and how
8:45 they overcame resistance to AI adoption
8:46 uh the role in of procurement and
8:48 facilitating adoption um in particular
8:50 of AI and and kind of new technologies
8:53 is to create that safe space to
8:55 incentivize people to want to try it and
8:57 that means that there will be mistakes
8:59 things may not turn out the way that um
9:00 they would have expected they'll have a
9:03 hypothesis and it doesn't work then we
9:05 will not chastise people we not penalize
9:07 people we'll actually we're looking to
9:08 be able to reward them for looking at
9:10 new ways of doing business new ways of
9:13 using processes and I do have at Nasa a
9:16 team looking at all of those use cases
9:18 for how to more efficiently um make use
9:21 of the um the AI technology um I have
9:23 two or three use cases going on right
9:24 now and we've awarded a small contract
9:27 to a small business to be able to allow
9:29 us to demonstrate uh we believe these
9:31 pilot to these demonstrations through
9:33 what we call our NASA acquisition
9:35 Innovation Launchpad that's our uh
9:37 Innovation laboratory by demonstrating
9:39 those we're recording what we call
9:41 microlearning um you know what was the
9:44 actual process we were trying to Pilot
9:45 um what was our expected or intended
9:47 outcome it's kind of like scientific
9:49 method but for procurement processes and
9:51 then we share those Lessons Learned we
9:53 are sharing our failures meaning the
9:54 things that didn't work as well as the
9:57 things that did work so people can learn
9:58 from the things that didn't work as well
10:00 as those things that did work and then
10:01 in that way you're going to speed up
10:03 learning and when you speed up learning
10:05 you speed up the um will to adopt new
10:08 technologies so it's an overarching
10:09 community and a different way of
10:11 thinking um and that's what we're
10:13 pushing at Nasa so there's what I would
10:15 call these two steps first step is just
10:18 doing AI understanding what a single AI
10:19 project is for whatever purpose you're
10:22 trying to do medical Finance business
10:24 War fighting operations whatever just
10:27 building a technology an ai ai enabled
10:29 system an AI enabled capab that's sort
10:32 of Step One the most important step over
10:36 time is how the enduser actually uses
10:38 that capability and that may be a very
10:41 different operating concept which is put
10:43 it in the user's hands as fast as
10:45 possible in the form of a minimal viable
10:47 product because we in the Pentagon could
10:50 never pretend that we had the answers
10:52 how the users would end up using those
10:54 capabilities so giving it over to them
10:57 early and often and allowing them to
10:58 play around with it so this idea of
11:00 diffusion I think is is more critical
11:02 than most people give it credit for a
11:04 few people like Jeff ding write about
11:06 this that a technology that you just
11:09 when you build it that's step one and
11:11 the fact that the United States is doing
11:14 it and China's doing it we Mo we both
11:16 may get these Technologies but I think
11:18 what I'm most interested in is how do we
11:20 use those Technologies in new creative
11:22 and innovative ways that give us a
11:24 competitive advantage against our
11:26 adversaries and potential adversaries
11:28 that's the diffusion part of it and and
11:30 you mentioned electricity printing press
11:32 all of those when they first came out I
11:35 don't think anybody truly could say this
11:37 is going to transform Society they
11:38 didn't know right away but of course the
11:41 printing press completely changed
11:43 globally how people communicated with
11:45 each other electricity completely
11:47 changed the entire world so this point
11:50 about AI being a dual use enabling
11:52 technology I would also say a general
11:54 purpose technology it comes largely from
11:56 commercial technology companies that
11:59 were then adapting into the department
12:01 of defense or int into the intelligence
12:03 Community but that's still step one the
12:05 rest of this is all about how the user
12:07 takes these Technologies brings them
12:10 into their daily lives and you ask I
12:11 think what you're getting at is what's
12:13 the endgame we don't call it AI anymore
12:15 it's just another tool that everybody's
12:17 using and it's great it makes their life
12:19 better they're more effective they're
12:21 more efficient things are better but
12:23 they don't have to stay AI anymore it's
12:25 magic until all of a sudden it's just
12:26 another piece of software you know there
12:29 are lots of barriers about um AI
12:33 adoption for acquirers and and one is um
12:35 and I'm going to use this word not as a
12:37 perjorative but just as a fact of
12:39 ignorance of both leadership and the
12:41 acquisition professional what the heck
12:43 this stuff is you know do I buy it per
12:46 pound or do I buy it for feature or do I
12:48 buy it to for outcomes which maybe I
12:51 should actually be thinking about um
12:53 because it's new to everybody a you
12:55 shouldn't feel dumb but you don't have a
12:58 clue about like what is this and what's
13:01 the best Etc you know the best advice is
13:04 um is working with the Departments or
13:07 the mission centers that actually need
13:08 some outcomes
13:11 delivered and then it becomes a little
13:13 easier to start looking at what do I
13:16 want to acquire that matches the outcome
13:19 versus single out of vendors with shiny
13:21 objects and well gee my large language
13:23 model is bigger than this or I could do
13:26 X or Y or whatever you really want that
13:28 match between you know Warf fighter or
13:31 user needs and what is the value
13:33 proposition or what is the product or
13:35 service capable of doing and more
13:38 importantly is how rapidly is that
13:40 solution space changing and what type of
13:43 contract do I want to get the government
13:46 into um before I commit to a multi-year
13:48 purchase of something that as I said
13:50 might deliver features not wanted or
13:52 needed or Worse become obsolete or
13:55 superseded by something else much better
Chrome Extension