Agencies across government can learn from innovative leaders in developing data strategies that leverage AI and other emerging technologies. Specifically, AI and intelligent automation (IA) technologies can improve grants management by expediting access to information and standardizing grant processes.
As government agencies strive to improve service by implementing advanced technologies, how that data informs strategy and business decisions represents a critical element of success. The IBM Center for The Business of Government and the Partnership for Public Service recently hosted five virtual panels on this subject, focused on readying agency data by using artificial intelligence and other technology solutions. These sessions focused on detecting and addressing agency data quality issues that might stem from bias or inaccuracy, and on addressing barriers to data sharing within and across agencies to promote better access through technology. The insights and lessons learned from these discussions will help government to advance data strategies within and between agencies.
The fourth in our series of reports from the five panel discussions summarizes insights and lessons learned from the Department of Health and Human Services’ (HHS) “ReImagine HHS Grants” initiative, led by Michael Peckham. Peckham addressed how his team uses intelligent automation, especially artificial intelligence and blockchain, to make the grants management process more efficient. The HHS initiative, which concluded in September and whose lessons learned are now being implemented throughout HHS, provides significant insight for applying emerging technologies for modernizing grants management (for more information about the Reimagine initiative, see this summary post from HHS).
The full webinar discussion is available here. Highlights follow.
HHS engaged grant recipients, and other stakeholders to ask what type of tool would be most useful to the grant process. A common request was something that added functionality, usability and flexibility. On average, grants management professionals used to spend four hours to complete the required pre-reward risk assessment for grants, but under the Reimagine initiative HHS introduced the Grant-Recipient Digital Dossier to reduce that time to only fifteen minutes after adapting AI through user-centered design. The change from four hours to fifteen minutes in work required to complete a risk assessment will be a return to mission of approximately $142 million a year. This gave HHS insights into how to similarly implement intelligent automation technologies such as robotic process automation to ingest information, natural language processing to comprehend information, and blockchain to store and share information. The HHS lesson also demonstrates that while AI can predict future outcomes and assess potential risk, agencies interested in using the technology should not simply apply AI to any problem without a clear program goal – in this case the grants process — in mind.
What has HHS Learned from Implementing Intelligent Automation?
IA includes multiple technologies. For example, robotic process automation (RPA) is one of the processes HHS currently manages. RPA enables rapid intake of information, which can be reviewed and “cleansed” to apply AI and deepen understanding of a process like grants management. Peckham noted that maintaining a specific purpose when applying AI to all of the information is essential: “Once you’ve ingested information and you have a specific purpose around that information, and you put the AI on top of it, you will start to see things in the data that you’ve never seen before. The computer does a great job of being able to extract information and present it to you in multiple different layers and different manners, depending on who you’re working with as a contractor.”
AI can then support making predictions for what will happen in the future, and potential risks later on. AI allows humans to see scenarios much more quickly, such as issues that may arise in setting up a grants system. At the same time, simply throwing AI at a problem is ill-advised – agencies should identify a process – in an area like grants management — where applying AI has a clear outcome. Peckham found, “I do caution that running out and just throwing AI at a problem, just because AI is a cool tool, is probably not the best approach. Think about it very logically, think about a process or something that you know intimately, and then apply the AI on top of some other type of automation. You’re going to start to see some real results.”
Peckham also noted that blockchain can be used for oversight and monitoring. Blockchain technologies bring partners together in a shared business network to improve the flow of information between partners and with the people tasked with oversight. Blockchain is based on trust among the partners in sharing information. Peckham noted, “if we have a blockchain, and all the information residing, then we can allow the people tasked with overseeing what’s going on—allow them real-time access to it so that they can tell us as we move through the process what the perceived issues are, and we can of course correct.”
Finally, Peckham advised other agencies interested in applying these technologies to avoid fear of failure – rather, to learn from failures – in applying new innovation. Lessons learned can pinpoint where to apply intelligent automation effectively.
How does HHS Apply IA?
User involvement and feedback in the engineering process is critical. HHS engaged with grants recipients, users of the technology, and other stakeholders from the early stages of the engineering process, concluding that both Federal employees and grant recipients shared similar challenges. Peckham noted, “They wanted to have something introduced to them that added functionality, usability and flexibility that they didn’t have previously.”
They worked with users to move through the grants management process, asking users questions like:
- Why do you do that in that system?
- Why are you copy and pasting that?
- Why are you clicking here?
Once the process was mapped the team could then ask:
- What emerging technology would solve that particular issue?
- Is this an RPA issue or is this an AI issue?
- Should we apply natural language processing here?
- Is this right for blockchain?
HHS could then modify intelligent automation technologies such as natural language processing by engaging recipients in conferences, meetings, and breakout sessions:
“We engaged recipients very, very closely. We went to conferences, we sat down with them, we had breakout sessions, sometimes there were six people, sometimes there were twenty-five people in a room. But we walked through the process, and sometimes we would say ‘If someone were to apply natural language processing to this particular aspect of your grant world, and it was going to tell you that at the end of the day, that you have more information here that can be utilized in a different manner, would that be helpful?”
The idea to build a risk assessment tool was inspired by users’ frustration with the lack of access to information required to complete a pre-reward risk assessment. This arose as a repeated challenge that required lots of redundant analysis. Applying AI to the information in the pre-award process, HHS developed the Grant-Recipient Digital Dossier or GDD to reduce the amount of time recipients would need to complete this required assessment — producing a pre-reward risk assessment that takes a person about fifteen minutes to go through – down from four hours.
This effort was then followed by development of an external Grant-Recipient Digital Dossier, to aid recipients required to do pre-reward risk assessments for sub-recipients. This tool facilitates gaining information from the sub-recipients’ dossiers. As Peckham reported: “Imagine the idea and the power of, once I am a prime recipient and I look at three sub-recipients and I say to them, ‘Hey, I’d like to do a risk assessment on you all, I want to use the external Grant-Recipient Digital Dossier to do that, would you give me access to your profile?’”
The change from four hours to fifteen minutes in work required to complete a risk assessment will be a return to the mission of approximately $142 million a year. And the tool is completely intuitive. After a five-minute instruction video, people understand how to move forward quickly – a key enabling lesson for any agency applying intelligent automation.