The Defense Department (DoD) is considered a model organization for harnessing data to generate insights and empower employees. During COVID-19, DoD’s Joint Artificial Intelligence Center (JAIC) has probed defense and health data to predict where hospitals are strained and where supplies run thin.
The reason DoD is able to thrive on the AI frontier, where so many agencies have barely trodden, is its central platform for data tools and services. ADVANA, which stands for “advanced analytics,” brings in information from 120 systems. Commercial-off-the-shelf products and cloud services on ADVANA mean employees – even nontechnical ones – can find useful analytics and build models.
“The way I view AI and ML right now is that it will always be augmented. And what I mean by augmented is that the human being will always be involved in that process,” said Greg Little, Director of the Chief Financial Officer Data Transformation Office within the Office of the Under Secretary of Defense (Comptroller).
DoD wasn’t always so data-inclusive. In fact, Monica McEwen, Public Sector Vice President of data intelligence company ThoughtSpot, remembers working with DoD 16 years ago, before Little was there. Though DoD was still cutting-edge back then, employees were not as welcomed into the process.
The analytics platform, at the time, had limited bandwidth, meaning some users could execute orders immediately, and others couldn’t at all. Early- and mid-2000s technological constraints and geographical barriers limited just how much data could influence an enterprise.
“Technology has just come so far in terms of helping to eliminate those challenges,” McEwen said.
At GovLoop’s virtual summit on Wednesday, those familiar with DoD’s evolution answered questions on a panel about advanced analytics. Brian Shealey, Director of DoD and Intelligence Community for Trifacta, a data wrangling software company, joined McEwen and Little. Below are the main data challenges DoD has battled – and continues to do so – and solutions the department has found.
Employees generally have two reservations about the use of advanced analytics and AI at their agencies.
First, they might not trust the information, a suspicion that can be validated if data doesn’t have a clear backstory or lineage. Their cynicism can lead them to actively work against the analytics.
Second, employees might fear that increasing automation and analytics will take their jobs. McEwen mentioned how data issues that used to take four days to resolve can now be fixed up in 15 minutes.
Step one to correcting the cultural dilemma is to just start, Little said. Introduce data technologies and processes bit by bit, paired with the commensurate education and exposure for involved employees.
“Visibility into the data is the best disinfectant for it,” McEwen said.
Solutions should be delivered on the same plate as a familiar process too, Little said, so that people are comfortable with the environment and notice tangible improvements.
Even so, not everyone will be receptive. Focus on pockets that are, Little said, because creating a reputation of trust and delivery is crucial to changing the culture.
The volume, velocity and variety of data are all increasing. Agencies need to wrangle it all in, and then distribute it to employees throughout the enterprise.
“The reality is, there’s such a vast amount of data out there, and really to get a point where we have a richer experience in life and a richer experience in business, the data needs to be stitched together,” Shealey said.
Adamantly, experts on the panel endorsed migrating to the cloud, which agencies are at various stages of. Cloud architectures, which are generally maintained off premises and delivered over the internet, are why DoD no longer has to worry about bandwidth on ADVANA, McEwen said. Agencies can scale systems on the cloud up and down to match usage. Cloud service providers can also provide ready-to-go software that removes barriers to entry.
Because of these new capabilities, Little has bold ideas for what’s next. DoD hopes to generate breaking-news-style alerts based on data trends and set up an AI query system, where an answer can be found in one or two clicks – much like with Apple’s Siri. Little said the experience wouldn’t just be easy to use, but “delightful.”
“We want to continue to connect dots across our organization,” Little said.
Don’t miss out on other virtual learning opportunities. Pre-register for GovLoop’s remaining 2020 virtual summits today.
This online training was brought to you by: