State governments have unique challenges when it comes to budgets, workforce readiness, and technology adoption. Increasingly state and local governments have also been innovative, pushing forward AI technology to help with many operational and constituent facing operations. Carlos Rivero, Chief Data Officer for the Commonwealth of Virginia, was recently a guest on the AI Today podcast as well as a panelist during the recent Data for AI Week Virtual Conference to discuss how states, and in particular the Commonwealth of Virginia, is leveraging data, automation, and cognitive technologies.
In an interview for this article, Carlos shares how automation, advanced data analytics, and AI play an increasing role in state government, some unique challenges around data at the state level, some interesting or surprising insights you can share about how Virginia is using ML and AI and more.
What are some of the unique challenges around data at the state level?
Carlos Rivero: An important component to successfully use data to enhance agency operations is getting people involved in the data governance process. Part of the mission of the Office of the Chief Data Officer is to figure out how to leverage data as a strategic asset. As we’re working toward that goal it has become apparent that the participation of individuals both horizontally (across different lines of business) and vertically (chain of command) within the organization is critical in being able to share and integrate data assets in a meaningful way. Having data custodians, data stewards, data owners, and executive sponsors participate in the governance discussion is important because everyone has a role to play in our ability to leverage data for analytics, machine learning, and intelligence applications.
A benefit that can also be a challenge at the state level is having the ability and responsibility to work with all the lines of business, in this case state agencies, at the same time. It’s critical to include the unique perspectives from each agency and figure out how to engage all stakeholders in a way where they see the value in data sharing and want to integrate their data assets to make data-driven decisions.
What are some of the technical and practical challenges states such as Virginia face around data collection and data usage?
Carlos Rivero: A very practical, yet non-technical challenge, is establishing and maintaining trust. People will not enter or maintain a relationship with you if trust is absent or compromised. In order to overcome this challenge we created the Commonwealth Data Trust, a safe, secure, and legally compliant information sharing environment that establishes consistent requirements for trust members through a standardized data sharing agreement process. The data trust provides a scalable alternative to multiple “point-to-point” sharing, promoting trust among its members through common rules for data security, privacy, and confidentiality, and reduces technical costs by onboarding to a single environment leveraging standard NIEM protocols.
The Commonwealth Data Trust also reduces the possibility of engaging in what I call the “data dance.” This is when organizations go out on their first “data date” ask each other:
“What data do you want?”
“Well, what data do you have?”
“Tell me what you want and I’ll tell you if I have it…” and so on.
In Virginia, the Chief Data Officer is a catalyst or broker for data sharing transactions (legally the trustee in the data trust), which facilitates data sharing through the implementation of enterprise solutions maintaining security and privacy. The trustee is responsible and accountable for ensuring that only authorized organizations and individuals access restricted-use data for the specific projects defined in the Data Trust User Agreements.
How can automation be applied at the state level?
Carlos Rivero: There are over 2,000 operational data systems in production within the Commonwealth of Virginia so I see a lot of opportunity for automation. Many of these systems are decades old. In some cases, business processes need to be re-engineered to take advantage of advanced computing architectures, automation, and machine learning. How do you eat an elephant? One bite at a time.
Within the Framework for Addiction Analysis and Community Transformation (FAACT), which started as a pilot project in Winchester, VA and has now expanded to Roanoke and Southwest Virginia, we’re leveraging machine learning models as we’re bringing in large amounts of data from behavioral health and community health organizations to identify some of the risk factors and indicators associated with substance use disorder and how we can mitigate them with services we offer. The purpose is to look at the human side of the services we provide and derive better intelligence to determine how we can evolve those services to promote positive outcomes and have a greater impact on the lives of our residents. The key is to augment human intelligence with AI by using machine learning models to sift through vast amounts of disparate data identifying patterns and trends, quantifying relationships, and recommending the best course of action for a given situation. However, it is still a human being who takes the action.
What are some interesting or surprising insights you can share about how Virginia wants to apply automation, ML and AI?
Carlos Rivero: Virginia is currently laying the foundation for AI by creating an enterprise data intelligence infrastructure as a value-added component to participating in the Commonwealth Data Trust providing Commonwealth of Virginia agencies the infrastructure, system development lifecycle administration services, and human resources needed to manage the deployment and consumption of intelligence products. Essentially, agencies do not have to build and manage their own intelligence infrastructure to take advantage of advanced computing capabilities. Organizations can allocate resources to and focus on solution development and let the Office of the CDO handle deployment and production management. This frees up valuable resources both human and financial to implement innovative solutions.
The Office of the Chief Data Officer is identifying common needs across agencies within the Commonwealth of Virginia and implementing enterprise solutions to address those needs. AI and ML are certainly on the horizon as the solutions mature.
Where do you see AI having the biggest impact on user experience?
Carlos Rivero: I see AI looking at all of the services we provide to our residents and determining which combination of services will provide the best outcomes for our constituents, but doing so in a transparent manner.
What can states and local governments do to attract the skilled workforce it needs to keep up with technological innovations?
Carlos Rivero: Virginia colleges and universities are producing an amazing amount of data analytics talent so we’ve leveraged our relationships with academic institutions and their engagement with the Commonwealth to develop two programs that increase the data acumen of these students while also providing value to state agencies. We are facilitating student access to restricted-use data to better understand and quantify the relationships between various measures. The metrics generated as a product of that research are used to operationalize data-driven informed decision-making. It’s about bridging the gap between investigative research and business operations. There are two commonwealth programs and an annual event that help attract this talent.
During the 2019 legislative session, the General Assembly created the Rural IT Apprenticeship Program. This program allows student apprentices to work on data technologies under the mentorship of an IT professional while engaging with data stewards and subject matter experts. It’s a two-fold education that allows students to gain experience in the field of data while learning the business of government. The program awards grants to small, rural information technology businesses in specific rural localities in the Southwest and Southside regions of Virginia.
The Commonwealth Data Internship Program was initiated to enhance the state’s efficiency while providing real-world experience to students. It began in 2014 with 20 participating state agencies and 45 graduate interns from the Virginia Commonwealth University School of Business. The program has expanded to include several participating colleges and universities across the Commonwealth. Essentially it’s a matchmaking service that connects commonwealth organizations that have data and problems to solve with academic institutions that have students seeking professional experiences developing real-world solutions.
Last, but not least, we just wrapped up our 2020 Virginia Datathon which is an event that brings together state agencies, students, localities, non-profits, and the private sector to leverage technology and data to develop innovative solutions addressing complex problems. Our theme this year was Communities and COVID-19 and served to explore the multi-faceted effect of COVID-19 and contrasting community experiences developing solutions that can mitigate the impact of another crisis and ultimately aid Virginia’s recovery.
What can states do to reskill and upskill their current workforce around data and AI skills?
Carlos Rivero: Earlier this year Virginia built a web-based data governance training course specifically tailored for state employees with the goal of increasing data literacy across the Commonwealth. This course is available to all employees through the Commonwealth of Virginia Learning Management System and is designed for both technical and non-technical audiences and has already increased the awareness and understanding of the importance of data governance within state agencies.
Our goal is to develop a data acumen assessment tool, provide a roadmap for individuals to follow, and develop training tools through web-based training, coaching, mentoring, and project-work that will help commonwealth employees increase their data acumen. I think getting people excited about using data to solve critical issues and giving them the tools they need to develop solutions is key to skills development.
What do you see as critical data-related challenges in your role as Chief Data Officer?
Carlos Rivero: The most critical challenge I face is the interpretation of federal privacy laws and regulations by agency data owners and stewards. Many agencies I work with have a very narrow, risk-averse interpretation of federal privacy laws and regulations which limits their ability to share data with other government agencies. Unfortunately, this is a cultural issue that will require a dramatic shift in behavior. Fortunately, guidance from our federal partners on allowable exemptions will help us move in the right direction. We are currently working with other state CDOs and non-profit organizations that have established relationships with multiple federal agencies to standardize guidance that all states can follow.
What AI technologies are you most looking forward to in the coming years?
Carlos Rivero: I am very excited to see how Natural Language Generation evolves as a technology to translate intelligence reports, visualizations, and dashboards into human readable text. This is one of the areas where we are failing our customers by not meeting the needs of our visually challenged residents. We live in a very visual world and I happen to be a visual learner. I take for granted the way information is presented to me in a way that I can easily and quickly understand. This is not the case for everyone. Utilizing technologies like NLG will help create a more equitable access to intelligence.