Governments are awash in data. To make sense of this data, gain insights, and to better serve citizens, agencies are turning to technologies such as automation, RPA, and ML and AI to better manage data, extract additional value, and improve processes and workflows. The US Department of Labor (DOL) is one such agency that is crafting a unique approach for using emerging technologies across its data rich environment.
The DOL’s Chief Technology Officer Sanjay Koyani and his team are working to integrate innovative technologies like responsible AI, RPA, and chatbots as well as a planned initiative to create an enterprise-wide data platform at DOL. In an upcoming AI in Government event on September 15, 2022, Sanjay will discuss the agency’s AI, automation, and data journey, what needs to be done to explore culture change considerations, and how best to identify problems and customer needs and then craft solutions that actually address and solve these problems.
In this preview interview for Forbes, Sanjay shares how DOL is applying AI and ML in a data rich environment, some of the challenges associated in adopting transformative technology in the public sector, as well as how DOL is looking at trustworthy and responsible AI.
What are some innovative ways you’re leveraging data and AI to benefit the Department of Labor (DOL)?
Sanjay Koyani: Every IT modernization initiative works towards our goal of best in federal IT solutions, which supports our DOL mission to enhance services for the American public and offer better customer service in support of a more digital workplace.
A little over a year ago we created a new branch within our Technology, Innovation and Engineering (TIE) Division that specializes in Emerging Technologies to create a human-centered design approach for future technologies at the Department. The first emerging tech capability we have launched and are working to scale enterprise-wide surrounds the use of automation – Robotics Process Automation (RPA). In the past year, we have launched five RPA bots – software applications to automate repetitive, rule-based tasks considered as administrative in nature – and are piloting an additional six. Currently, we are exploring additional opportunities across the Department’s agencies with several in development for future use. The overall goal being to allow staff to focus their ability on mission-critical work instead of administrative based tasks and lay the groundwork for other advanced technologies such as machine learning and Artificial Intelligence.
In TIE, we are also exploring how to use AI more responsibly as a service to improve performance and increase value. We have several AI pilots underway where we are innovating in the cloud by using native AI-enabled capabilities to evaluate program needs like speech-to-text, text-to-speech, translation services, and form recognition services to extract text and structure documents for faster decision making. In parallel, we have also started to explore the practice of designing and evaluating AI in an ethical and responsible manner so that we can scale it with greater confidence.
To fuel our AI and automation efforts, our team is also building out our analytics capabilities through the creation of an Enterprise Data Platform to bolster data-based decision making in innovative ways. Data is the foundation for AI and machine learning, so we are investing in data management and analytics tools. By using Technology Modernization Funding awarded for this initiative, the Department can accelerate data management and advanced analytics capabilities to strengthen cross-agency data sharing and sharing and make better and faster decisions. We can also advance elements of the Executive Order on Worker Empowerment by equipping investigator and policy teams with improved intelligence, and high quality and prompt worker protection data that makes jobs safer.
How do you identify which problem area(s) to start with for your data and cognitive technology projects?
Sanjay Koyani: We have started to identify projects through our Innovation Incubator, which helps evaluate proofs of concepts – show the risks and evaluate it against existing tools. This has allowed us to expand on our current pilot programs to see if they might solve other problems and explore innovative solutions as well.
Another tactic we used recently is an enterprise-wide Bot-a-Thon to help inform employees on using bots and ideate on how it could help the workforce with administrative based tasks like reporting, populating forms, or research. The result involved nine different bot processes being started in development in FY21, with thousands of work hours saved across five operationalized bots.
What are some of the unique opportunities the public sector has when it comes to data and AI?
Sanjay Koyani: We are experiencing a greater visibility and focus on the importance of modernizing IT in government, and how IT impacts multiple government services. This Presidential administration has made IT modernization, including data and AI, a priority. Congress continues to focus on IT efforts through the Federal IT Acquisition Reform Act (FITARA) that puts agency CIOs in control of IT investments and grades agencies on seven key IT areas. Cybersecurity breaches have also brought a renewed focus to how AI can help the public sector in mitigating threats and responding faster to potential risks.
What are some use cases you can share where you have successfully applied AI?
Sanjay Koyani: We developed a new user inspired website for DOL’s Employment and Training Administration (ETA) based on a customer-centric design and enhanced the customer experience through the incorporation of AI. As a result, AI helped to improve candidate sourcing/matching for opportunities on Apprenticeship.gov.
Another example is our use of AI-enabled form recognition services to speed beneficiary determinations. Our team assessed how AI-enabled cloud technologies could aid claim examiners in assessing benefits forms for accuracy and fraud for faster determinations. Using existing cloud technology, we trained AI models to extract and organize data from several claims forms so that examiners received the consolidated information faster. Prior to this, examiners spent considerable manual hours sorting and comparing forms instead of fully focusing on beneficiary customer support, faster decisions.
Can you share some of the challenges when it comes to AI and ML in the public sector?
Sanjay Koyani: There are a few challenges I will touch on. One is data management, which is a big focus for the Department. While having an abundance of data is good, you need to know what information is available and understand how it is being used. To use AI and ML correctly, you need to know what data exists, have it cataloged, and agency stakeholders aligned with how DOL can use data for faster and better decision making. This requires continued education and investments into our data strategy.
Human-centered design is also key for AI/ML. So, you must make sure you are communicating with all the stakeholders involved to understand the process and how they would use the technology. This is when it is important to decide if AI/ML would even fix the problem. Not all problems can be fixed through technology.
Another key challenge is cultural acceptance. Cultural change can be difficult, so make sure to show the workplace benefits, how new technologies will be used responsibly, and how it will be accessed across the agency.
At the end of the day for the Department of Labor, enterprise-wide scalability is the long-term goal. So, we are considering both cultural and technical considerations, evaluating effectiveness and then building upon our successes.
How are you navigating privacy, trust, and security concerns around the use of AI?
Sanjay Koyani: We are incorporating the use of a Responsible AI Framework, to ensure AI is used in a trustworthy way. The Department is doing this by collaborating with both non-profit practitioners and government-based subject matter experts to end bias in development of AI algorithms and to help us navigate the complex landscape of creating safe AI.
Additionally, we have several policies and procedures currently in place to help navigate security concerns. Those include a sound governance policy and a holistic strategy to incorporate security from the start.
In the Executive Order on Responsible AI, OSTP outlines 10 principles for responsible implementation of AI systems. Furthermore, privacy is a huge consideration when considering the use of an AI system. Not only do we want to ensure that we are not introducing bias, but we also want to ensure that the privacy of those whose information is contained in the data is protected. We do this through compliance with federal regulations as well as a dedicated privacy assessment.
What are you doing to develop an AI ready workforce?
Sanjay Koyani: We are building out an Enterprise Architecture and IT governance process to support the use of all emerging technology solutions. This will help ensure the alignment of tools in support of agencies’ business needs and standardize processes. Another way we are developing an AI ready workforce is through education, training and hiring of subject matter experts. For instance, we recently had a Presidential Innovation Fellow (PIF) evaluate our AI pilot use cases for trustworthy AI in support of the Administration’s Executive Order on promoting the use of trustworthy AI in the federal government. Our PIF enabled us to work with agency experts to design and test new models for assessing how we design, develop, and deploy AI in a more responsible way that helps create more transparency, trust in scaling AI with confidence.
What AI technologies are you most looking forward to in the coming years?
Sanjay Koyani: I am looking forward to seeing more responsible AI testing programs that will help fulfill the gaps of our modernization efforts for legacy IT systems and using more automation to empower transformation. Each of which will allow us to mature our enterprise architecture and use of emerging technologies.
Another area I am excited to see AI assist with is in cybersecurity. I think there will be even more solutions available to help automate responses to cyber threats and reduce risk to the organization, given the ever changing environment and continual stress on resources to protect systems and network solutions.
In his upcoming presentation in September 2022, Sanjay will dig deeper into some of the topics discussed above as well as share highlights from his team’s work on integrating innovative technologies like responsible AI, RPA, and chatbots.